Risk factors and predictive model construction for lower extremity arterial disease in diabetic patients

Yingjie Kuang; Zhixin Cheng; Jun Zhang; Chunxu Yang; Yue Zhang

doi:10.1371/journal.pone.0314862

. 2024 Dec 30;19(12):e0314862. doi: 10.1371/journal.pone.0314862

Risk factors and predictive model construction for lower extremity arterial disease in diabetic patients

Yingjie Kuang ¹, Zhixin Cheng ², Jun Zhang ¹, Chunxu Yang ¹, Yue Zhang ^2,^*

Editor: Jincheng Wang³

PMCID: PMC11684652 PMID: 39775606

Abstract

Objective

To understand the prevalence and associated risk factors of lower extremity arterial disease (LEAD) in Chinese diabetic patients and to construct a risk prediction model.

Methods

Data from the Diabetes Complications Warning Dataset of the China National Population Health Science Data Center were used. Logistic regression analysis was employed to identify related factors, and machine learning algorithms were used to construct the risk prediction model.

Results

The study population consisted of 3,000 patients, with 476 (15.9%) having LEAD. Multivariate regression analysis indicated that male gender, atherosclerosis, carotid artery stenosis, fatty liver, hematologic diseases, endocrine disorders, and elevated glycosylated serum proteins were independent risk factors for LEAD. The risk prediction models constructed using Logistic regression and MLP algorithms achieved moderate discrimination performance, with AUCs of 0.73 and 0.72, respectively.

Conclusion

Our study identified the risk factors for LEAD in Chinese diabetic patients, and the constructed risk prediction model can aid in the diagnosis of LEAD.

Introduction

Lower extremity arterial disease (LEAD) is significantly associated with diabetes, with the risk increasing as the severity of diabetes worsens [1]. By 2035, the number of diabetes patients is expected to increase rapidly to approximately 600 million people [2]. Diabetic patients are at a higher risk of developing LEAD compared to non-diabetic patients [3, 4]. Consequently, the prevalence of lower extremity arterial disease in diabetic patients (LEADDP) is expected to increase significantly in the near future. However, statistics indicate that over 60% of LEAD patients may be asymptomatic, leading to early stages of the disease often being overlooked [5]. Compared to non-diabetic LEAD patients, diabetic peripheral neuropathy may mask LEAD symptoms, making the diagnosis of LEADDP more challenging. Current guidelines recommend initial screening for LEAD based on patient interviews and clinical examinations, using the Fontaine or Rutherford scales for assessment [6, 7]. This diagnosis heavily relies on the expertise of specialists and examination conditions, and the diagnostic performance is not ideal [8].

LEAD increases the risk of foot ulcers and lower limb amputations, significantly impacting the quality of life and treatment for diabetic patients, while also imposing a substantial economic burden on society [9]. Therefore, identifying risk factors and accurately predicting LEADDP risk is crucial for the early initiation of prevention and treatment in high-risk patients. In recent years, constructing risk prediction models using machine learning algorithms has become increasingly popular in medical research. These algorithms can automatically select informational variables and capture nonlinear relationships between variables, enhancing predictive capabilities.

The primary aim of this study is to describe the prevalence of LEAD in diabetic patients and to identify potential risk factors. Subsequently, we trained machine learning models to predict LEADDP risk by integrating all available clinical objective information.

Data and methods

In this study, we used the Diabetes Complications Warning Dataset obtained from the China National Population Health Science Data Center to analyze the significant risk factors for LEADDP and to construct a risk prediction model. This dataset includes general information, vital signs, laboratory measurements, and comorbidities for 3,000 diabetic patients, totaling 81 variables (S1 Table). It encompasses traditional risk factors [1, 8]: dyslipidemia, hypertension, cardiovascular diseases, kidney disease, inflammation, obesity, male gender, aging, and infections. Due to the complex pathogenesis of LEADDP and the unclear associated risk factors, we also included more information on comorbidities and additional laboratory indicators in order to identify new risk factors and avoid omissions. We aimed to construct a risk prediction model that can be used by non-specialists; hence, treatment factors were not included.

In this study, we utilized traditional statistical methods to describe the cohort’s demographic and health characteristics. For normally distributed continuous variables, independent sample t-tests or one-way ANOVA were employed to compare the means ± standard deviations between groups. Categorical variables were presented as frequencies (percentages), and the chi-square test was used to assess differences between groups. LEADDP was taken as the dependent variable, while general information, vital signs, laboratory measurements, and comorbidities were considered independent variables to explore the influence of various risk factors. Based on the results of the univariate analysis, factors with a p-value less than 0.1 were included in the multivariate analysis. We examined the correlations between all variables and visualized data with strong associations (Fig 1). Before training the model, we excluded incomplete cases from the dataset, resulting in the removal of 15 cases. After this preprocessing step, a total of 2,985 cases were included in the model. The collinearity of the included factors was examined. The dataset was randomly split into training and validation sets in a 7:3 ratio, with LEADDP as the dependent variable. The training set included 2,089 cases, with 296 cases of LEADDP and 1,793 cases of non-LEADDP. The validation set included 896 cases, with 127 cases of LEADDP and 769 cases of non-LEADDP (Table 1). We used the training set data to construct the LEADDP risk prediction model and validated it using the validation set data.

Fig 1 — Dark blue areas indicate strong negative correlations, while dark red areas indicate strong positive correlations. Light colors show weaker correlations. Abbreviations: AS, Atherosclerosis; FLD, Fatty Liver Disease; HBA1C, Hemoglobin A1c; TP, Total Protein.

Table 1. Characteristics of samples in the training and validation sets.

	NO LEADDP	LEADDP	Total
Training set	1793	296	2089
Validation set	769	127	896
Total	2562	423	2985

Open in a new tab

Results

Among the 3,000 diabetic patients included in this study, 476 (15.9%) had LEADDP. The average age was 57.8 years, with 37.5% being female. Significant differences were observed between patients who had LEADDP and those who did not in terms of general information, vital signs, laboratory tests, and comorbidities (Table 2).

Table 2. Baseline characteristics according to the presence of LEADDP.

Category/variable		NO LEADDP	LEADDP	P value
Sex, n (%)	male	1560 (61.8)	314 (66)	0.089
Sex, n (%)	female	964 (38.2)	162 (34)	0.089
Age, years		57.51±11.28	59.3±10.37	0.001
Height, meter		166.51±7.03	166.34±8.54	0.665
Weight, kg		73±13.00	73.56±12.43	0.392
SBP, mmHg		138.22±20.84	141.18±21.72	0.005
DBP, mmHg		80.40±12.01	80.82±11.57	0.48
Heart Rate, n/min		79.88±26.81	69.64±35.03	0.107
BMI, kg/m²		26.25±3.82	26.58±3.61	0.088
Hypertention, n (%)	no	835(33.1)	119(25)	0.001
Hypertention, n (%)	yes	1689(66.9)	357(75)	0.001
Hyperlipidemia, n (%)	no	1973(78.3)	371(77.9)	0.952
Hyperlipidemia, n (%)	yes	551(21.8)	105(22.1)	0.952
AS, n (%)	no	1420(56.3)	37(7.8)	<0.0001
AS, n (%)	yes	1104(43.7)	439(92.2)	<0.0001
Cerebral Apoplexty, n (%)	no	2351(93.1)	425(89.3)	0.004
Cerebral Apoplexty, n (%)	yes	173(6.9)	51(10.7)	0.004
Carotid Artery Stenosis, n (%)	no	2440(96.7)	431(90.5)	<0.0001
Carotid Artery Stenosis, n (%)	yes	84(3.3)	45(9.5)	<0.0001
FLD, n (%)	no	1793(71)	270(56.7)	<0.0001
FLD, n (%)	yes	731(29)	206(43.3)	<0.0001
Cirrhosis, n (%)	no	2483(98.4)	470(98.7)	0.689
Cirrhosis, n (%)	yes	41(1.6)	6(1.3)	0.689
CLD, n (%)	no	2191(968)	400(84)	0.109
CLD, n (%)	yes	333(13.2)	76(16)	0.109
Pancreatic Disease, n (%)	no	2482(98.3)	470(98.7)	0.563
Pancreatic Disease, n (%)	yes	42(1.7)	6(1.3)	0.563
Biliary Tract Disease, n (%)	no	2190(86.8)	383(80.5)	<0.0001
Biliary Tract Disease, n (%)	yes	334(13.2)	93(19.5)	<0.0001
Nephropathy, n (%)	no	1520(60.2)	203(42.6)	<0.0001
Nephropathy, n (%)	yes	1004(39.8)	273(57.4)	<0.0001
Renal Faliure, n (%)	no	2375(94.1)	442(92.9)	0.347
Renal Faliure, n (%)	yes	149(5.9)	34(7.1)	0.347
Nervous System Disease, n (%)	no	2382(94.4)	44(92.9)	0.202
Nervous System Disease, n (%)	yes	142(5.6)	34(7.1)	0.202
CHD, n (%)	no	1693(67.1)	322(67.6)	0.832
CHD, n (%)	yes	831(32.9)	154(32.4)	0.832
MI, n (%)	no	2363(93.6)	447(93.9)	0.839
MI, n (%)	yes	161(6.4)	29(6.1)	0.839
CHF, n (%)	no	2351(93.1)	437(91.8)	0.329
CHF, n (%)	yes	173(6.9)	39(8.2)	0.329
Arrhythmias, n (%)	no	2392(94.8)	434(91.2)	0.003
Arrhythmias, n (%)	yes	132(5.2)	42(8.8)	0.003
Respiratory System Disease, n (%)	no	2121(84)	407(85.5)	0.451
Respiratory System Disease, n (%)	yes	403(16)	69(14.5)	0.451
Hematonosis, n (%)	no	2175(86.2)	381(80)	0.001
Hematonosis, n (%)	yes	349(13.8)	95(20)	0.001
Rheumatic Immunity, n (%)	no	2432(96.4)	465(97.7)	0.169
Rheumatic Immunity, n (%)	yes	92(3.6)	11(2.3)	0.169
Pregnant, n (%)	no	2515(99.6)	475(99.8)	0.715
Pregnant, n (%)	yes	9(0.4)	1(0.2)	0.715
Endocrine Disease, n (%)	no	1773(70.2)	225(47.3)	<0.0001
Endocrine Disease, n (%)	yes	751(29.8)	251(52.7)	<0.0001
MEN, n (%)	no	2430(96.3)	462(97.1)	0.426
MEN, n (%)	yes	94(3.7)	14(2.9)	0.426
PCOS, n (%)	no	2521(99.9)	476(100)	1
PCOS, n (%)	yes	3(0.1)	0(0)	1
Digestive Carcinoma, n (%)	no	2385(94.5)	462(97.1)	0.022
Digestive Carcinoma, n (%)	yes	139(5.5)	14(2.9)	0.022
Urologic Neoplasms, n (%)	no	2496(98.9)	473(99.4)	0.462
Urologic Neoplasms, n (%)	yes	28(1.1)	3(0.6)	0.462
Gynecolgical Tumor, n (%)	no	2434(96.4)	467(98.1)	0.068
Gynecolgical Tumor, n (%)	yes	90(3.6)	9(1.9)	0.068
Breast Tumor, n (%)	no	2515(99.6)	475(99.8)	0.715
Breast Tumor, n (%)	yes	9(0.4)	1(0.2)	0.715
Lung Tumor, n (%)	no	2478(98.2)	467(98.1)	1
Lung Tumor, n (%)	yes	46(1.8)	9(1.9)	1
Intracranial Tumor, n (%)	no	2512(99.5)	472(99.2)	0.493
Intracranial Tumor, n (%)	yes	12(0.5)	4(0.8)	0.493
Other Tumor, n (%)	no	2312(91.6)	442(92.9)	0.366
Other Tumor, n (%)	yes	212(8.4)	34(7.1)	0.366
GLU, mmol/L		8.40±3.93	8.69±3.72	0.139
GLU_2H, mmol/L		14.52±4.58	15.81±4.46	0.001
HBA1C, %		7.70±1.71	8.29±1.79	<0.0001
GSP, μmol/L		224.11±79.15	237.69±79.13	0.005
TG, mmol/L		2.03±1.65	1.99±1.50	0.704
TC, mmol/L		4.60±1.38	4.69±1.61	0.233
HDL_C, mmol/L		1.07±0.32	1.06±0.30	0.762
LDL_C, mmol/L		2.84±1.09	2.95±1.38	0.048
FBG, g/L		7.73±37.90	8.69±42.44	0.621
UPR_24, g/24h		1.44±1.50	1.08±1.41	<0.0001
BU, mmol/L		7.07±4.97	7.37±4.91	0.222
SCR, μmol/L		106.24±119.97	108.29±114.39	0.731
UCR, mmol/24h		5.49±2.66	6.61±12.16	0.021
SUA, μmol/L		328.86±103.68	327.81±93.56	0.838
HB, g/L		131.77±23.35	131.4±21.94	0.75
CP, nmol/L		2.21±1.53	2.17±2.1	0.777
INS, μU/ml		14.71±39.04	11.61±13.54	0.213
PCV		0.39±0.06	0.38±0.06	0.493
PLT, 10⁹/L		219.18±73.16	213.34±63.99	0.105
ESR, mm/h		27.08±29.49	23.3±25.81	0.056
TBILI, μmol/L		11.02±12.35	10.79±7.14	0.698
DBILI, μmol/L		3.64±9.67	2.99±2.16	0.147
TP, g/L		65.7±7.43	64.36±7.51	<0.0001
ALB, g/L		39.53±5.832	38.67±6.1	0.004
LDH_L, U/L		174.29±64.78	171.48±54.01	0.377
ALT, U/L		24.47±29.252	21.83±15.86	0.055
AST, U/L		20.74±24.03	18.26±10.06	0.028
GGT, U/L		43.3±73.3	38.844±53.28	0.209
ALP, U/L		75.72±47.7	72.31±30.28	0.138
LP_A, mg/dl		30.81±43.58	36.84±53.03	0.661
PL, mmol/L		2.46±0.60	2.46±0.69	0.972
PT, s		13.15±1.28	13.1±1.19	0.413
PTA		99.19±18.88	98.37±19.07	0.387
APTT, s		36.74±7.33	36.42±5.61	0.361
FIBRIN, g/L		13.96±25.29	16.12±30.71	0.833
ALB_CR, mg/g		149.66±230.06	154.12±260.96	0.748
LPS, U/L		164.58±425.35	157.88±300.38	0.808
CA199, U/L		30.7±226.85	18.21±16.1	0.247
CRP, mg/L		1.26±2.88	0.85±2.15	0.008
TH2, ng/L		0.45±0.1	0.35±0.11	0.155
IBILI, μmol/L		7.37±4.35	7.78±5.59	0.072
GLO, g/L		26.18±4.95	25.68±4.56	0.044

Open in a new tab

Note: Data were expressed as mean ± SD or n (%).

Abbreviations: SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure; BMI, Body Mass Index; CHD, Coronary Heart Disease; MI, Myocardial Infarction; CHF, Congestive Heart Failure; AS, Atherosclerosis; FLD, Fatty Liver Disease; CLD, Chronic Liver Disease; PCOS, Polycystic Ovary Syndrome; MEN, Multiple Endocrine Neoplasia; GLU, Glucose; GLU_2H, 2-hour Postprandial Glucose; HBA1C, Hemoglobin A1c; GSP, Glycated Serum Protein; TG, Triglycerides; TC, Total Cholesterol; HDL_C, High-Density Lipoprotein Cholesterol; LDL_C, Low-Density Lipoprotein Cholesterol; FBG, Fibrinogen; UPR_24, 24-hour Urinary Protein; BUN, Blood Urea Nitrogen; BU, Blood Urea; SCR, Serum Creatinine; UCR, Urine Creatinine; SUA, Serum Uric Acid; HB, Hemoglobin; CP, C-Peptide; INS, Insulin; PCV, Packed Cell Volume; PLT, Platelets; ESR, Erythrocyte Sedimentation Rate; TBILI, Total Bilirubin; DBILI, Direct Bilirubin; TP, Total Protein; ALB, Albumin; LDH_L, Lactate Dehydrogenase; ALT, Alanine Aminotransferase; AST, Aspartate Aminotransferase; GGT, Gamma-Glutamyl Transferase; ALP, Alkaline Phosphatase; LP_A, Lipoprotein(a); PL, Phospholipids; PT, Prothrombin Time; PTA, Prothrombin Activity; APTT, Activated Partial Thromboplastin Time; FIBRIN, Fibrinogen; ALB_CR, Albumin/Creatinine Ratio; LPS, Lipase; CA199, Cancer Antigen 19–9; CRP, C-Reactive Protein; TH2, T Helper Cell 2; IBILI, Indirect Bilirubin; GLO, Globulin.

LEADDP patients were predominantly male, older in age, and had higher proportions of hypertension, stroke, carotid artery stenosis, fatty liver, biliary diseases, kidney disease, arrhythmias, hematologic diseases, and other endocrine disorders. They also exhibited poorer glycemic control, with the majority having atherosclerosis in other areas. Additionally, LEADDP patients had higher levels of low-density lipoprotein and urinary creatinine, and lower levels of urinary microalbumin, total protein, serum albumin, aspartate aminotransferase, C-reactive protein, and globulin. Other laboratory indicators showed no statistically significant differences. In the multivariable analysis, all variables that showed a strong association with LEADDP in the univariate analysis (P < 0.1) were included. The results indicated that gender, atherosclerosis, carotid artery stenosis, fatty liver, hematologic diseases, endocrine disorders, and glycosylated serum protein were significantly associated with LEADDP (Table 3).

Table 3. Results of the multivariate analysis.

	P value	OR (95% CI)
Sex	0.018	1.422 (1.062, 1.904)
AS	<0.001	19.911 (12.349, 32.103)
Carotid Artery Stenosis	0.002	2.198 (1.345, 3.591)
FLD	0.004	1.518 (1.140, 2.023)
Hematonosis	<0.001	2.094 (1.467, 2.989)
Endocrine Disease	<0.001	2.155 (1.623, 2.862)
GSP	0.014	1.002 (1.000, 1.004)

Open in a new tab

Abbreviations: AS, Atherosclerosis; FLD, Fatty Liver Disease; GSP, Glycated Serum Protein.

We constructed five LEADDP risk prediction models using Python, specifically the Logistic regression, decision tree, random forest, k-nearest neighbors, and neural network algorithms. We plotted the ROC curves for each algorithm (Fig 2). Among them, the risk prediction models based on the Logistic regression and MLP algorithms achieved the best performance, with AUCs of 0.73 and 0.72, respectively.

Discussion

In our study, approximately 15.9% of Chinese diabetic patients had LEAD. Male gender, atherosclerosis, carotid artery stenosis, fatty liver, hematologic diseases, other endocrine disorders, and glycated serum protein were found to be independently associated with the prevalence of LEADDP. The risk prediction model constructed using Logistic regression and MLP algorithms achieved the best performance, with moderate discriminative ability.

In our study, approximately 15.9% of Chinese diabetic patients had LEADDP, whereas another study from the southern coastal region of China reported a prevalence of about 4.9% [10]. The reason for this difference is not yet clear but may be related to regional variations. Our study data are from a medical center in northern China, where the climate is colder compared to the south. The diagnosis of LEAD primarily relies on ABI (Ankle-Brachial Index). At low temperatures, the pressure in the distal arteries of the lower limbs significantly decreases, which reduces the ABI value [11]. Therefore, we believe this difference may be due to the influence of different climates in the regions where the studies were conducted. Additionally, there are significant differences in dietary habits, lifestyles, and economic levels among people in different provinces of China, which may, to some extent, affect the development of diabetes and its complications. Existing research data indicate that the prevalence of diabetes varies across different provinces in China [12]. However, there have been no epidemiological reports specifically on LEADDP in China.

The risk of LEAD increases with age, a finding also supported by our study [8]. However, existing research indicates that the number of peripheral arterial disease (PAD) cases is higher in females than in males across all age groups [13]. Since females have lower ABI values than males, the actual prevalence of PAD in females might be higher than current estimates [14, 15]. However, in our study, we found that there were more men than women among diabetic patients with concurrent LEAD. A study on LEAD in the diabetic population also reached similar conclusions [16]. Males have a higher risk of LEADDP compared to females, with male diabetic patients being about 1.4 times more likely to develop LEADDP than female diabetic patients. This may be related to the unique pathophysiological processes in diabetic patients.

Diabetic vascular complications share common features, with their main pathological manifestations being endothelial dysfunction and atherosclerosis [17, 18]. The relationship between diabetic macrovascular complications and LEADDP is evident and confirmed in our study. Additionally, we found that carotid artery stenosis can be considered a risk factor for LEADDP. The risk of LEADDP in patients with carotid artery stenosis is 2.198 times higher than in those without. Because symptoms are more evident, diabetic patients with carotid artery stenosis are more likely to seek medical help. Therefore, emphasizing the predictive value of carotid artery stenosis will aid in the diagnosis and early treatment of LEADDP. Due to the widespread vascular complications of diabetes, cardiovascular disease cannot be ignored [17]. However, in our study, there were no significant differences in the prevalence of complications such as coronary heart disease, myocardial infarction, and chronic heart failure. Our study shows that hypertension is associated with LEADDP, but not all types of hypertension. Consistent with some studies, increased systolic pressure is associated with LEADDP, while differences in diastolic pressure are not statistically significant [19, 20]. This suggests that we should pay attention to the LEAD risk in diabetic patients with increased pulse pressure.

Glycemic control is considered an important measure to prevent the risk of LEADDP. Hyperglycemia promotes oxidative stress, glycoxidation, and systemic inflammation, damaging the endothelial cells of the arterial wall, leading to lipid deposition and the development of atherosclerosis [21]. Some studies have shown a link between glycemic control levels and the incidence of LEAD, indicating that poor glycemic control may lead to the development of LEADDP [10, 22, 23]. Other studies, however, have found little association between blood glucose levels and LEADDP [24, 25]. The reasons for these differences are unclear, possibly involving differences in study populations or data analysis methods. In our study, LEAD patients had poorer blood glucose control compared to those without LEAD. Although fasting blood glucose levels did not show significant differences, the postprandial 2-hour blood glucose levels were significantly elevated. Some studies have suggested that insulin resistance is an important cause of LEADDP and can be harmful even with normal blood glucose levels [26, 27]. Studies have also shown that arterial vascular damage can begin before blood glucose levels increase, with elevated insulin levels potentially being a contributing factor [21]. This may indicate that poor glycemic control and LEADDP are not the cause but the result. This may suggest a deeper connection between glycemic control and the occurrence of LEADDP, warranting further research.

Lipid abnormalities play a crucial role in the development of atherosclerosis. Unlike in the past, where only low-density lipoprotein was emphasized, both low-density lipoprotein and triglyceride abnormalities should now be considered [28]. For LEADDP, a recent study has shown that plasma concentrations of HDL-C, TC/HDL-C, and apolipoproteins are associated with the incidence of LEADDP [29]. Other studies have not found a link between lipid parameters and the incidence of LEADDP [23, 30]. These inconsistencies may be due to lipid-lowering treatments in the study populations [31]. In our study, low-density lipoprotein was significantly elevated in the LEADDP group, but other lipid indicators, particularly triglycerides, showed no statistically significant differences. Elevated low-density lipoprotein levels indicate dyslipidemia. In our study, patients with both LEAD and fatty liver had an increased risk of LEAD, and the difference in BMI was not significant. In some studies, obesity is considered a risk factor for the development of LEAD [32, 33]. However, obesity does not cause LEADDP in all populations [34]. This suggests that obesity may not be a direct risk factor for LEADDP, but rather an indicator of dyslipidemia.

Impaired renal function is not a traditional risk factor for LEAD, but studies have shown that it is associated with an increased risk of atherosclerosis, cardiovascular events, and LEAD [10, 35, 36]. In our study, individuals with LEADDP exhibited a higher prevalence of kidney disease and significantly reduced kidney function compared to those without LEADDP. Univariate analysis showed that concurrent kidney disease, UPR_24, and UCR were associated with LEADDP events, but this association was excluded after adjusting for potential confounding factors.

Certain endocrine diseases other than diabetes can also increase the risk of LEADDP. Hypothyroidism is associated with an increased risk of atherosclerosis and may cause peripheral vascular constriction [37, 38]. Patients with Cushing’s syndrome have elevated cortisol levels, leading to hypertension, hyperglycemia, and lipid abnormalities, increasing the risk of atherosclerosis [39]. In our study, we found that having other endocrine diseases increases the risk of LEADDP.

In this study, hematologic diseases were considered one of the risk factors for LEADDP. Possible mechanisms include coagulation disorders, abnormal platelet counts or function, and increased blood viscosity. Abnormalities in fibrinogen, thrombin formation, and fibrin degradation occur not only in acute thrombotic complications but may also occur in stable forms of LEAD, as research suggests [40]. Platelet abnormalities and increased blood viscosity may lead to the development and exacerbation of LEAD [41]. Currently, no studies have analyzed the correlation between hematologic diseases and LEADDP. This finding opens new avenues for research into LEADDP.

Machine learning can assist clinicians in making diagnoses, thereby reducing their workload, increasing the sensitivity of disease diagnosis, avoiding missed diagnoses, and lowering healthcare costs [42]. The pathogenesis of LEADDP is complex, and current research does not clearly define its risk factors. Machine learning algorithms can learn from existing data and identify relationships between independent and dependent variables. This capability makes them well-suited for constructing risk prediction models for LEADDP [43]. Previous studies have employed machine learning algorithms to develop risk prediction models for LEAD, achieving relatively good predictive performance [44, 45]. However, these studies did not specifically focus on diabetic patients, and thus could not capture the unique characteristics of LEAD occurrence in this population. The application of machine learning algorithms in LEAD risk prediction models specifically targeting diabetic patients remains relatively limited. In our study, we aimed to develop a preliminary machine learning model to predict LEAD in diabetic patients. Our results indicated that Logistic Regression and MLP were the most effective algorithms, exhibiting the highest AUC and superior statistical performance. Generally, the AUC ranges from 0.5 to 1.0, with values between 0.5 and 0.7 indicating low discrimination ability, values between 0.7 and 0.9 indicating moderate discrimination ability, and values above 0.9 indicating high discrimination ability [46]. In our study, the accuracy of the model’s predictive ability was evaluated by calculating the area under the ROC curve. The AUC values of the risk prediction models based on Logistic regression and MLP were 0.73 and 0.72, respectively, indicating that these two models achieved moderate discrimination ability. This model is built using easily collectable objective clinical data, and it can be directly integrated with data from medical systems without the need for secondary collection. This allows for improved LEAD risk prediction without increasing the workload of medical personnel.

This study identified risk factors for LEAD in the Chinese diabetic population, further explored the mechanisms related to LEADDP, and developed a relatively user-friendly risk prediction model using machine learning algorithms. The model can assist specialists in identifying high-risk LEADDP patients. The limitations of this study include the following. First, the data used in this study were obtained from the China National Population Health Science Data Center, and while the data were collected by reliable medical centers, the comorbidities in the dataset were not clearly defined, which may pose limitations for broader application. Second, for the sake of model usability, this study did not include treatment factors, so the potential effects of medications on the outcomes have not been explored. Third, as this is a cross-sectional study, the research design does not allow us to establish causal relationships between the identified risk factors. Finally, the data primarily come from northern China, which may limit its generalizability to other regions.

Conclusion

In conclusion, our study identified several important risk factors for LEAD in the Chinese diabetic population. These findings contribute to a deeper understanding of the pathogenesis of LEADDP. The accumulation of risk factors further increases the risk of disease, highlighting the urgent need for early diagnosis and intervention. The LEADDP risk prediction models constructed using machine learning can assist in early screening, reducing adverse outcomes and long-term risks for patients.

Supporting information

S1 Table. Patient characteristics in the dataset.

(PDF)

pone.0314862.s001.pdf^{(125KB, pdf)}

Acknowledgments

We thank the China National Population Health Science Data Center for providing data support. We also appreciate the constructive comments from the editors and each reviewer during the revision of our manuscript.

Data Availability

The data that support the findings of this study are openly available in the Diabetes Complications Warning Dataset at https://www.ncmi.cn/phda/dataDetails.do?id=CSTR:A0006.11.A0005.201905.000282-V1.0.

Funding Statement

This work was supported by the Shandong Provincial Natural Science Foundation (ZR2022MH268). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.Fowkes FGR, Aboyans V, Fowkes FJI, McDermott MM, Sampson UKA, Criqui MH. Peripheral artery disease: epidemiology and global perspectives. Nat Rev Cardiol. 2017;14(3):156–170. doi: 10.1038/nrcardio.2016.179 [DOI] [PubMed] [Google Scholar]
2.Guariguata L, Whiting DR, Hambleton I, Beagley J, Linnenkamp U, Shaw JE. Global estimates of diabetes prevalence for 2013 and projections for 2035. Diabetes Res Clin Pract. 2014;103(2):137–149. doi: 10.1016/j.diabres.2013.11.002 [DOI] [PubMed] [Google Scholar]
3.Jarrett RJ. Type 2 (non-insulin-dependent) diabetes mellitus and coronary heart disease-chicken, egg or neither? Diabetologia. 1984;26(2):99–102. doi: 10.1007/BF00281114 [DOI] [PubMed] [Google Scholar]
4.Stern MP. Diabetes and cardiovascular disease. The “common soil” hypothesis. Diabetes. 1995;44(4):369–374. doi: 10.2337/diab.44.4.369 [DOI] [PubMed] [Google Scholar]
5.Li J, Li Y, Ma W, Liu Y, Yin X, Xie C, et al. Association of Time in Range levels with Lower Extremity Arterial Disease in patients with type 2 diabetes. Diabetes Metab Syndr. 2020;14(6):2081–2085. doi: 10.1016/j.dsx.2020.09.028 [DOI] [PubMed] [Google Scholar]
6.American Diabetes Association. 11. Microvascular Complications and Foot Care: Standards of Medical Care in Diabetes-2021. Diabetes Care. 2021;44(Suppl 1):S151–S167. doi: 10.2337/dc21-S011 [DOI] [PubMed] [Google Scholar]
7.Rutherford RB, Baker JD, Ernst C, Johnston KW, Porter JM, Ahn S, et al. Recommended standards for reports dealing with lower extremity ischemia: revised version. J Vasc Surg. 1997;26(3):517–538. doi: 10.1016/s0741-5214(97)70045-4 [DOI] [PubMed] [Google Scholar]
8.Nativel M, Potier L, Alexandre L, Baillet-Blanco L, Ducasse E, Velho G, et al. Lower extremity arterial disease in patients with diabetes: a contemporary narrative review. Cardiovasc Diabetol. 2018;17(1):138. doi: 10.1186/s12933-018-0781-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Buso G, Aboyans V, Mazzolai L. Lower extremity artery disease in patients with type 2 diabetes. Eur J Prev Cardiol. 2019;26(2_suppl):114–124. doi: 10.1177/2047487319880044 [DOI] [PubMed] [Google Scholar]
10.Chen T, Xiao S, Chen Z, Yang Y, Yang B, Liu N. Risk factors for peripheral artery disease and diabetic peripheral neuropathy among patients with type 2 diabetes. Diabetes Res Clin Pract. 2024;207:111079. doi: 10.1016/j.diabres.2023.111079 [DOI] [PubMed] [Google Scholar]
11.Carter SA. The effect of cooling on toe systolic pressures in subjects with and without Raynaud’s syndrome in the lower extremities. Clin Physiol. 1991;11(3):253–261. doi: 10.1111/j.1475-097x.1991.tb00456.x [DOI] [PubMed] [Google Scholar]
12.Zuo H, Shi Z, Hussain A. Prevalence, trends and risk factors for the diabetes epidemic in China: a systematic review and meta-analysis. Diabetes Res Clin Pract. 2014. Apr;104(1):63–72. doi: 10.1016/j.diabres.2014.01.002 [DOI] [PubMed] [Google Scholar]
13.Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al. Global Burden of Cardiovascular Diseases and Risk Factors, 1990–2019: Update From the GBD 2019 Study. J Am Coll Cardiol. 2020;76(25):2982–3021. doi: 10.1016/j.jacc.2020.11.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Aboyans V, Criqui MH, McClelland RL, Allison MA, McDermott MM, Goff DC Jr, et al. Intrinsic contribution of gender and ethnicity to normal ankle-brachial index values: the Multi-Ethnic Study of Atherosclerosis (MESA). J Vasc Surg. 2007;45(2):319–327. doi: 10.1016/j.jvs.2006.10.032 [DOI] [PubMed] [Google Scholar]
15.Collins TC, Suarez-Almazor M, Bush RL, Petersen NJ. Gender and peripheral arterial disease. J Am Board Fam Med JABFM. 2006;19(2):132–140. doi: 10.3122/jabfm.19.2.132 [DOI] [PubMed] [Google Scholar]
16.Mohammedi K, Woodward M, Hirakawa Y, Zoungas S, Williams B, Lisheng L, et al. Microvascular and Macrovascular Disease and Risk for Major Peripheral Arterial Disease in Patients With Type 2 Diabetes. Diabetes Care. 2016. Oct;39(10):1796–803. doi: 10.2337/dc16-0588 Epub 2016 Jul 25. [DOI] [PubMed] [Google Scholar]
17.Li Y, Liu Y, Liu S, Gao M, Wang W, Chen K, et al. Diabetic vascular diseases: molecular mechanisms and therapeutic strategies. Signal Transduct Target Ther. 2023;8(1):152. doi: 10.1038/s41392-023-01400-z [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Zhou X, Yu L, Zhao Y, Ge J. Panvascular medicine: an emerging discipline focusing on atherosclerotic diseases. Eur Heart J. 2022;43(43):4528–4531. doi: 10.1093/eurheartj/ehac448 [DOI] [PubMed] [Google Scholar]
19.Meijer WT, Grobbee DE, Hunink MG, Hofman A, Hoes AW. Determinants of peripheral arterial disease in the elderly: the Rotterdam study. Arch Intern Med. 2000;160(19):2934–2938. doi: 10.1001/archinte.160.19.2934 [DOI] [PubMed] [Google Scholar]
20.Newman AB, Siscovick DS, Manolio TA, Polak J, Fried LP, Borhani NO, et al. Ankle-arm index as a marker of atherosclerosis in the Cardiovascular Health Study. Cardiovascular Heart Study (CHS) Collaborative Research Group. Circulation. 1993;88(3):837–845. doi: 10.1161/01.cir.88.3.837 [DOI] [PubMed] [Google Scholar]
21.Haas AV, McDonnell ME. Pathogenesis of Cardiovascular Disease in Diabetes. Endocrinol Metab Clin North Am. 2018. Mar;47(1):51–63. doi: 10.1016/j.ecl.2017.10.010 . [DOI] [PubMed] [Google Scholar]
22.Adler AI, Stevens RJ, Neil A, Stratton IM, Boulton AJM, Holman RR. UKPDS 59: hyperglycemia and other potentially modifiable risk factors for peripheral vascular disease in type 2 diabetes. Diabetes Care. 2002;25(5):894–899. doi: 10.2337/diacare.25.5.894 [DOI] [PubMed] [Google Scholar]
23.Althouse AD, Abbott JD, Forker AD, Bertolet M, Barinas-Mitchell E, Thurston RC, et al. Risk factors for incident peripheral arterial disease in type 2 diabetes: results from the Bypass Angioplasty Revascularization Investigation in type 2 Diabetes (BARI 2D) Trial. Diabetes Care. 2014;37(5):1346–1352. doi: 10.2337/dc13-2303 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Norman PE, Davis WA, Bruce DG, Davis TME. Peripheral arterial disease and risk of cardiac death in type 2 diabetes: the Fremantle Diabetes Study. Diabetes Care. 2006;29(3):575–580. doi: 10.2337/diacare.29.03.06.dc05-1567 [DOI] [PubMed] [Google Scholar]
25.Mohammedi K, Woodward M, Hirakawa Y, Zoungas S, Williams B, Lisheng L, et al. Microvascular and Macrovascular Disease and Risk for Major Peripheral Arterial Disease in Patients With Type 2 Diabetes. Diabetes Care. 2016;39(10):1796–1803. doi: 10.2337/dc16-0588 [DOI] [PubMed] [Google Scholar]
26.Yaribeygi H, Farrokhi FR, Butler AE, Sahebkar A. Insulin resistance: Review of the underlying molecular mechanisms. J Cell Physiol. 2019;234(6):8152–8161. doi: 10.1002/jcp.27603 [DOI] [PubMed] [Google Scholar]
27.Welsh GI, Hale LJ, Eremina V, Jeansson M, Maezawa Y, Lennon R, et al. Insulin signaling to the glomerular podocyte is critical for normal kidney function. Cell Metab. 2010;12(4):329–340. doi: 10.1016/j.cmet.2010.08.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Libby P. The changing landscape of atherosclerosis. Nature. 2021;592(7855):524–533. doi: 10.1038/s41586-021-03392-8 [DOI] [PubMed] [Google Scholar]
29.Bertrand C, Saulnier PJ, Potier L, Croyal M, Blanchard V, Gand E, et al. Plasma concentrations of lipoproteins and risk of lower-limb peripheral artery disease in people with type 2 diabetes: the SURDIAGENE study. Diabetologia. 2021;64(3):668–680. doi: 10.1007/s00125-020-05326-x [DOI] [PubMed] [Google Scholar]
30.Eshcol J, Jebarani S, Anjana RM, Mohan V, Pradeepa R. Prevalence, incidence and progression of peripheral arterial disease in Asian Indian type 2 diabetic patients. J Diabetes Complications. 2014;28(5):627–631. doi: 10.1016/j.jdiacomp.2014.04.013 [DOI] [PubMed] [Google Scholar]
31.Yeo TC, Chan YH, Low LP, Venketasubramanian N, Lim SC, Tay JC, et al. Risk factor profile and treatment patterns of patients with atherothrombosis in Singapore: insight from the REACH Registry. Ann Acad Med Singapore. 2008;37(5):365–371. [PubMed] [Google Scholar]
32.Criqui MH, Matsushita K, Aboyans V, Hess CN, Hicks CW, Kwan TW, et al. Lower Extremity Peripheral Artery Disease: Contemporary Epidemiology, Management Gaps, and Future Directions: A Scientific Statement From the American Heart Association. Circulation. 2021;144(9):e171–e191. doi: 10.1161/CIR.0000000000001005 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Gerhard-Herman MD, Gornik HL, Barrett C, Barshes NR, Corriere MA, Drachman DE, et al. 2016 AHA/ACC Guideline on the Management of Patients With Lower Extremity Peripheral Artery Disease: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Circulation. 2017;135(12):e686–e725. doi: 10.1161/CIR.0000000000000470 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Heffron SP, Dwivedi A, Rockman CB, Xia Y, Guo Y, Zhong J, et al. Body mass index and peripheral artery disease. Atherosclerosis. 2020;292:31–36. doi: 10.1016/j.atherosclerosis.2019.10.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Mann JFE, Gerstein HC, Pogue J, Bosch J, Yusuf S, for the HOPE Investigators. Renal Insufficiency as a Predictor of Cardiovascular Outcomes and the Impact of Ramipril: The HOPE Randomized Trial. Ann Intern Med. 2001;134(8):629–636. doi: 10.7326/0003-4819-134-8-200104170-00007 [DOI] [PubMed] [Google Scholar]
36.Checheriţă IA, Smarandache D, David C, Ciocâlteu A, Ion DA, Lascăr I. Vascular calcifications in chronic kidney disease—clinical management. Romanian J Morphol Embryol Rev Roum Morphol Embryol. 2012;53(1):7–13. [PubMed] [Google Scholar]
37.Stabouli S, Papakatsika S, Kotsis V. Hypothyroidism and hypertension. Expert Rev Cardiovasc Ther. 2010;8(11):1559–1565. doi: 10.1586/erc.10.141 [DOI] [PubMed] [Google Scholar]
38.Biondi B, Klein I. Hypothyroidism as a risk factor for cardiovascular disease. Endocrine. 2004;24(1):1–13. doi: 10.1385/ENDO:24:1:001 [DOI] [PubMed] [Google Scholar]
39.Ferriere A, Tabarin A. Cushing’s syndrome: Treatment and new therapeutic approaches. Best Pract Res Clin Endocrinol Metab. 2020;34(2):101381. doi: 10.1016/j.beem.2020.101381 [DOI] [PubMed] [Google Scholar]
40.Lassila R. Role and management of coagulation disorders in peripheral arterial disease. Scand J Surg SJS Off Organ Finn Surg Soc Scand Surg Soc. 2012;101(2):94–99. doi: 10.1177/145749691210100204 [DOI] [PubMed] [Google Scholar]
41.Salhiyyah K, Forster R, Senanayake E, Abdel-Hadi M, Booth A, Michaels JA. Pentoxifylline for intermittent claudication. Cochrane Database Syst Rev. 2015;9(9):CD005262. doi: 10.1002/14651858.CD005262.pub3 [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Hampe N, Wolterink JM, van Velzen SGM, Leiner T, Išgum I. Machine Learning for Assessment of Coronary Artery Disease in Cardiac CT: A Survey. Front Cardiovasc Med. 2019;6. doi: 10.3389/fcvm.2019.00172 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Santos MS, Abreu PH, García-Laencina PJ, Simão A, Carvalho A. A new cluster-based oversampling method for improving survival prediction of hepatocellular carcinoma patients. J Biomed Inform. 2015;58:49–59. doi: 10.1016/j.jbi.2015.09.012 [DOI] [PubMed] [Google Scholar]
44.Wang X, Yang Y, Zhang J, Zang S. Development and validation of a prediction model for the prolonged length of stay in Chinese patients with lower extremity atherosclerotic disease: a retrospective study. BMJ Open. 2023. Feb 9;13(2):e069437. doi: 10.1136/bmjopen-2022-069437 [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Zhang L, Ma Y, Li Q, Long Z, Zhang J, Zhang Z, et al. Construction of a novel lower-extremity peripheral artery disease subtype prediction model using unsupervised machine learning and neutrophil-related biomarkers. Heliyon. 2024. Jan 6;10(2):e24189. doi: 10.1016/j.heliyon.2024.e24189 ; PMCID: PMC10827514 [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Pearce J, Ferrier S. Evaluating the predictive performance of habitat models developed using logistic regression. Ecol Model. 2000;133(3):225–245. doi: 10.1016/S0304-3800(00)00322-7 [DOI] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0314862.r001

Decision Letter 0

Adeel Ahmad Khan

28 Aug 2024

PONE-D-24-26075Risk Factors and Predictive Model Construction for Lower Extremity Arterial Disease in Diabetic PatientsPLOS ONE

Dear Dr. Kuang,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript within Oct 12 2024 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Adeel Ahmad Khan

Academic Editor

PLOS ONE

Journal Requirements:

1. When submitting your revision, we need you to address these additional requirements.Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf 2. Please note that PLOS ONE has specific guidelines on code sharing for submissions in which author-generated code underpins the findings in the manuscript. In these cases, we expect all author-generated code to be made available without restrictions upon publication of the work. Please review our guidelines at https://journals.plos.org/plosone/s/materials-and-software-sharing#loc-sharing-code and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse. 3. Thank you for stating the following financial disclosure: "This work was supported by the Shandong Provincial Natural Science Foundation (ZR2022MH268)."Please state what role the funders took in the study. If the funders had no role, please state: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript." If this statement is not correct you must amend it as needed. Please include this amended Role of Funder statement in your cover letter; we will change the online submission form on your behalf. 4. For studies involving third-party data, we encourage authors to share any data specific to their analyses that they can legally distribute. PLOS recognizes, however, that authors may be using third-party data they do not have the rights to share. When third-party data cannot be publicly shared, authors must provide all information necessary for interested researchers to apply to gain access to the data. (https://journals.plos.org/plosone/s/data-availability#loc-acceptable-data-access-restrictions) For any third-party data that the authors cannot legally distribute, they should include the following information in their Data Availability Statement upon submission:1) A description of the data set and the third-party source2) If applicable, verification of permission to use the data set3) Confirmation of whether the authors received any special privileges in accessing the data that other researchers would not have4) All necessary contact information others would need to apply to gain access to the data.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Partly

Reviewer #4: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

Reviewer #4: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

Reviewer #4: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: No

Reviewer #4: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors analyzed 3000 diabetic patients from the Diabetes Complications Warning Dataset of the China National Population Health Science Data Center to identify risk factors of LEAD and construct its risk prediction model. With 476 patients who developed LEAD, they identified several risk factors and their logistic regression-based prediction model has moderate discrimination performance with AUCs of 0.73 and 0.72.

1. Treatment factors were excluded in the prediction model. Please clarify what treatment and also provide more detail about the rational for this decision.

2. Authors only said “missing values were handled” but no further information. How were they handled should be provided.

3. Also, testing collinearity is good. But what authors deal with collinear variables should be provided.

4. All significant predictors from univariate analysis were included in multivariate analysis. How about those non-significant in univariate analysis? If not included, will you be concerned about the potential confounding?

5. Please present the sample characteristic comparison between the training and testing sample

Reviewer #2: It needs a minor revision , of the statistical methods with better description of the patients and methods.

The results are not well detailed and needs more refinement.

The discussions are weakly describing the current literature.

Reviewer #3: The authors used data from the Diabetes Complications Warning Dataset of the China National Population Health Science Data Center and Logistic regression, decision tree, random forest, k-nearest neighbors, and neural network algorithms to test the accuracy of LEADDP risk prediction models. Several suggestions were listed as follows.

1. Please make a brief introduction and clarification of the dataset and study design the authors used. It seems to be a cross-sectional design. If so, it is inappropriate to call it incidence.

2. Table 1 should be included into supplementary.

3. The figures are very vague in the combined PDF. After downloading the original figures, it's hard to see the words alongside the bars. Please add figure legend.

4. BP high and BP low, should be changed into systolic blood pressure and diastolic blood pressure.

5. No unit is indicated in Table 2.

6. Why listing so many variables in the table 2? Are these variables all associated with LEAD? It's suggested to make clear the clinical significance of these variables with LEAD first before doing the analyses.

7. In the first part of the discussion, it's suggested to make a brief introduction of the main findings first. Strengths and limitation should be clarified in the discussion part also.

8. The key finding is the prediction model in this study. So, please clarify the main findings compared with other studies instead of talking about each risk factor in the model.

9. It's suggested to reorganize the manuscript to highlight the main findings from this study.

Reviewer #4: This paper presents a highly interesting study that offers valuable "big data" insights into a particularly challenging disease. I have the following comments :

1. In Table 1, you present the characteristics of all patients included in the dataset. However, it is unclear who is responsible for collecting this data and by what methods. It is essential to provide more detailed information about the Diabetes Complications Warning Dataset. Specifically, who is responsible for validating the accuracy of the data?

2. Several parameters listed in Table 1 (e.g., arterial hypertension, renal failure, fatty liver, etc.) have definitions established by published standards. How were these parameters defined within your dataset? In the absence of such definitions, how can you ensure consistency in the diagnosis of conditions like renal failure across different patients?

3. Could you elaborate on the rationale behind choosing a 7:3 ratio for the training and validation sets?

4. Your statement regarding atmospheric changes appears to be contentious. The paper you cite associates atmospheric temperature with acute limb ischemia, rather than LEAD. How does your paper’s definition of LEAD compare with the definition used in the study by Chen et al.?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Aram Baram

Reviewer #3: No

Reviewer #4: Yes: Theodosios Bisdas

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2024 Dec 30;19(12):e0314862. doi: 10.1371/journal.pone.0314862.r002

Author response to Decision Letter 0

12 Sep 2024

PONE-D-24-26075

Risk Factors and Predictive Model Construction for Lower Extremity Arterial Disease in Diabetic Patients

PLOS ONE

Dear Editors and Reviewers,

Thank you for your letter and for the reviewers' comments concerning our manuscript entitled "Risk factors and predictive model construction for lower extremity arterial disease in diabetic patients" (ID: PONE-D-24-26075). The comments provided are valuable and very helpful for revising and improving our paper, as well as having significant guiding implications for our future research. We have carefully considered the comments and made the necessary corrections, which we hope will meet with your approval. The revisions have all been marked in the revised manuscript.

Responding to the journal's requirements:

We sincerely apologize for overlooking PLOS ONE's style requirements. In this revision, we have carefully followed the journal's style guidelines and made the necessary adjustments to our manuscript.

Regarding code sharing, we were unable to access the PLOS Computational Biology author guidance provided by the journal, so the code used in our study was not uploaded in this revision. Our research, which utilizes the Python-based scikit-learn package, focuses on analyzing risk factors for LEAD in diabetic patients and constructing a risk prediction model to help prevent LEAD in this population. This study is more oriented towards medical objectives rather than the improvement of machine learning code. Therefore, we are not currently eager to share our code. However, we would be happy to provide the code upon request.

We update our financial disclosure as follows: "This work was supported by the Shandong Provincial Natural Science Foundation (ZR2022MH268). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

Regarding data sharing, we will update the data availability statement in this revision to comply the journal's requirements in our manuscript.

Thank you very much for your valuable feedback and assistance in improving our manuscript.

Responds to the reviewer's comments:

Reviewer #1:

1. Response to comment: Treatment factors were excluded in the prediction model. Please clarify what treatment and also provide more detail about the rational for this decision.

Response:

Thank you very much for your valuable feedback. Most diabetic patients use medications such as insulin to control their blood glucose levels, and there are significant variations in both the types and dosages of these medications. If these patients also suffer from other conditions, the treatment regimen becomes even more complex. Additionally, considering the healthcare environment in China, many patients may have already begun treatment before being admitted, but their prior treatment plans may not have been appropriate. Given these practical conditions, we decided not to include treatment factors in this study design. The main reasons are as follows:

a. Including treatment factors would inevitably make the model overly complex, requiring a sample size beyond our capacity.

b. Collecting data on patients' previous treatment plans would largely rely on the accounts of the patients and their families, who lack a professional medical background. These accounts may contain inaccuracies that could affect the experimental results.

c. We aimed to make this predictive model accessible to non-professionals, so it was important to simplify the model as much as possible without compromising its discriminative ability. This would lower the threshold for use and help more LEAD patients receive treatment earlier.

d. In the future, to enable the automated operation of this predictive model, we need to rely as much as possible on data already available in the medical system, such as diagnoses and laboratory test results. Avoiding the need for secondary data entry would enhance the model's clinical usability.

e. The effects of treatment measures are somewhat reflected in patients' various clinical indicators, so we believe that excluding treatment factors would have an acceptable impact on the model.

For these reasons, we decided not to include treatment factors in this study.

2. Response to comment: Authors only said "missing values were handled" but no further information. How were they handled should be provided.

Response:

We sincerely apologize for not providing the details on how we handled missing values. For the missing data in our study, we addressed it by removing incomplete cases. We have revised the resubmitted manuscript to include these details. Specifically, in lines 111-115, we changed "Missing values were handled before model training" to "Before training the model, we excluded incomplete cases from the dataset, resulting in the removal of 15 cases. After this preprocessing step, a total of 2,985 cases were included in the model."

3. Response to comment: Also, testing collinearity is good. But what authors deal with collinear variables should be provided.

Response:

Thank you very much for your feedback. We examined the multicollinearity of the relevant risk factors using the df.corr() method from the pandas library in Python.

4. Response to comment: All significant predictors from univariate analysis were included in multivariate analysis. How about those non-significant in univariate analysis? If not included, will you be concerned about the potential confounding?

Response:

We apologize for not providing a detailed explanation on this issue. Since the differences observed in the results of univariate analysis may not accurately reflect the effect of each factor on the outcome event, we decided to include variables with statistical significance (P < 0.05) from the univariate analysis into the multivariate analysis. Additionally, to avoid omitting important variables, we also extended the inclusion criteria to P < 0.1. This approach is commonly used in medical research, and many well-regarded studies have employed this method.

Therefore, in our study, we included factors with P < 0.1 from the univariate analysis in the multivariate analysis. To clarify this issue, we have added this explanation to the revised manuscript. In lines 107-110, we changed "All significant predictors identified in the univariate analysis were incorporated into the multivariate analysis" to "Based on the results of the univariate analysis, factors with a p-value less than 0.1 were included in the multivariate analysis." In lines 158-162, we revised "In the multivariate analysis, all variables that showed a significant association with LEADDP in the univariate analysis were included" to "In the multivariable analysis, all variables that showed a strong association with LEADDP in the univariate analysis (P < 0.1) were included."

5. Response to comment: Please present the sample characteristic comparison between the training and testing sample.

Response:

Thank you very much for your suggestion. In this study, we overlooked providing the sample characteristics for the training and testing samples. After this revision, we have included a detailed description of the sample characteristics in the text. For example, in lines 117-119, we have updated it to: "The training set included 2,089 cases, with 296 cases of LEADDP and 1,793 cases of non-LEADDP. The validation set included 896 cases, with 127 cases of LEADDP and 769 cases of non-LEADDP." To make it easier for readers to understand the sample characteristics of both the training and testing samples, we have created a table that directly presents these characteristics. Please refer to Table 1 for more details.

Table 1. Characteristics of Samples in the Training and Validation Sets

NO LEADDP LEADDP Total

Training set 1793 296 2089

Validation set 769 127 896

Total 2562 423 2985

The above content constitutes our responses to your comments and the revisions made to the manuscript based on your suggestions. We must express our gratitude to you for your valuable feedback. Your recommendations, which address aspects such as study design, data preprocessing, and statistical analysis, have significantly contributed to improving the readability of our manuscript and enhancing our research. Thank you once again.

Reviewer #2:

1. Response to comment: It needs a minor revision , of the statistical methods with better description of the patients and methods.

Response:

Thank you very much for your comments. We have revised the description of the statistical methods, further explained the considerations and reasons for including risk factors in our study, supplemented the criteria for including risk factors in the multivariate analysis, and provided details on the handling of missing values. We have also presented the characteristics of the training and testing samples.

In lines 79-82, we added: "Due to the complex pathogenesis of LEADDP and the unclear associated risk factors, we also included more information on comorbidities and additional laboratory indicators in order to identify new risk factors and avoid omissions" to further explain the rationale for including risk factors in this study.

The original Table 1 has been moved to the supplementary materials for easier reading.

We also supplemented the criteria for including risk factors in the multivariate analysis, stating: "Based on the results of the univariate analysis, factors with a p-value less than 0.1 were included in the multivariate analysis" (lines 109-110).

We added details on the handling of missing values, as follows: "Before training the model, we excluded incomplete cases from the dataset, resulting in the removal of 15 cases. After this preprocessing step, a total of 2,985 cases were included in the model. The collinearity of the included factors was examined" (lines 113-115).

We described the sample characteristics of the training and validation sets. You can see in lines 117-119: "The training set included 2,089 cases, with 296 cases of LEADDP and 1,793 cases of non-LEADDP. The validation set included 896 cases, with 127 cases of LEADDP and 769 cases of non-LEADDP (Table 1)." This information can be visually assessed from Table 1.

Table 1. Characteristics of Samples in the Training and Validation Sets

NO LEADDP LEADDP Total

Training set 1793 296 2089

Validation set 769 127 896

Total 2562 423 2985

2. Response to comment: The results are not well detailed and needs more refinement.

Response:

We apologize for the lack of detail in the results provided. In this revision, we have addressed your comments and further enhanced the research results. We have added the units for the risk factors included in this study, which were overlooked in the initial draft. Additionally, we have supplemented the 95% confidence intervals for the multivariate analysis results.

In line 130, we added general information about the patients, such as: "The average age was 57.8 years, with 37.5% being female."

In Table 2, we have included the units for each risk factor.

In Table 3, we have refined and completed the results of the multivariate analysis. The revised Table 3 is as follows.

Table 3. Results of the multivariate analysis

P value OR ( 95% CI )

Sex 0.018 1.422 (1.062, 1.904)

AS ＜0.001 19.911 (12.349, 32.103)

Carotid Artery Stenosis 0.002 2.198 (1.345, 3.591)

FLD 0.004 1.518 (1.140, 2.023）

Hematonosis ＜0.001 2.094 (1.467, 2.989）

Endocrine Disease ＜0.001 2.155 (1.623, 2.862）

GSP 0.014 1.002 (1.000, 1.004）

Abbreviations: AS, Atherosclerosis; FLD, Fatty Liver Disease; GSP, Glycated Serum Protein.

3. Response to comment: The discussions are weakly describing the current literature.

Response:

Thank you very much for your comments. Following your feedback, we have reorganized the manuscript and revised the discussion section.

At the beginning of the discussion, we added the following content to summarize the findings of our study: "In our study, approximately 15.9% of Chinese diabetic patients had LEAD. Male gender, atherosclerosis, carotid artery stenosis, fatty liver, hematologic diseases, other endocrine disorders, and glycated serum protein were found to be independently associated with the prevalence of LEADDP. The risk prediction model constructed using Logistic regression and MLP algorithms achieved the best performance, with moderate discriminative ability" (lines 174-179).

To more accurately explain the differences between our study and a previous study, we included additional details about the study population's regional characteristics. In lines 180-182, we modified "In our study, approximately 15.9% of Chinese diabetic patients had LEADDP, whereas another study reported a prevalence of about 4.9%" to "In our study, approximately 15.9% of Chinese diabetic patients had LEADDP, whereas another study from the southern coastal region of China reported a prevalence of about 4.9%." To explain the differences in prevalence between the two studies, we added the following content in lines 183-187: "The reason for this difference is not yet clear but may be related to regional variations. Our study data are from a medical center in northern China, where the climate is colder compared to the south. The diagnosis of LEAD primarily relies on ABI (Ankle-Brachial Index). At low temperatures, the pressure in the distal arteries of the lower limbs significantly decreases, which reduces the ABI value [11]." Additionally, in lines 189-194, we added: "Furthermore, there are significant differences in dietary habits, lifestyles, and economic levels among people in different provinces of China, which may, to some extent, affect the development of diabetes and its complications. Existing research data indicate that the prevalence of diabetes varies across different provinces in China [12]. However, there have been no epidemiological reports specifically on LEADDP in China." These modifications help clarify the differences in prevalence between our study and the previous research on LEAD among Chinese diabetic patients.

In lines 199-201, we revised "However, in our study, we found that the incidence of LEADDP was higher in males" to "However, in our study, we found that there were more men than women among diabetic patients with concurrent LEAD." In line 202, we added the following content and reference: "A study on LEAD in the diabetic population also reached similar conclusions [16]." These changes help explain the gender differences in LEAD prevalence among diabetic versus non-diabetic patients.

In lines 216-218, we added: "However, in our study, there were no significant differences in the prevalence of complications such as coronary heart disease, myocardial infarction, and chronic heart failure," to further discuss our results.

To explain the mechanism by which hyperglycemia damages blood vessels, we added in lines 224-226: "Hyperglycemia promotes oxidative stress, glycoxidation, and systemic inflammation, damaging the endothelial cells of the arterial wall, leading to lipid deposition and the development of atherosclerosis [21]." To clarify the relationship between blood glucose control and LEAD risk in our study, we included in lines 231-234: "In our study, LEAD patients had poorer blood glucose control compared to those without LEAD. Although fasting blood glucose levels did not show significant differences, the postprandial 2-hour blood glucose levels were significantly elevated." To explain our findings, we added in lines 236-238: "Studies have also shown that arterial vascular damage can begin before blood glucose levels increase, with elevated insulin levels potentially being a contributing factor [21]."

To clarify our results regarding lipids, fatty liver, and BMI, we added in lines 252-254: "Elevated low-density lipoprotein levels indicate dyslipidemia. In our study, patients with both LEAD and fatty liver had an increased risk of LEAD, and t

Attachment

Submitted filename: Response to Reviewers.docx

pone.0314862.s002.docx^{(507.7KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0314862.r003

Decision Letter 1

Jincheng Wang

14 Oct 2024

PONE-D-24-26075R1Risk Factors and Predictive Model Construction for Lower Extremity Arterial Disease in Diabetic PatientsPLOS ONE

Dear Dr. Kuang,

Please address comments from reviewer 3.

Please submit your revised manuscript by Nov 28 2024 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

We look forward to receiving your revised manuscript.

Kind regards,

Jincheng Wang

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Additional Editor Comments :

Please address comments from reviewer 3.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

Reviewer #3: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #1: (No Response)

Reviewer #2: Yes

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: (No Response)

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #1: (No Response)

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #1: (No Response)

Reviewer #2: Yes

Reviewer #3: No

**********

6. Review Comments to the Author

Reviewer #1: (No Response)

Reviewer #2: All points are revised; it's a benefit for the specialists in this field, while needs more syntax refinements

Reviewer #3: The authors have addressed most of the comments left. However, there still exist some main concerns here. The authors stated that they opted to include more influencing factors in their analysis to identify risk factors associated with LEAD and avoid omissions by using machine learning algorithms. I agree with the authors on this view, but the following points still need to be revised.

First, this study found that “Male gender, atherosclerosis, carotid artery stenosis, fatty liver, hematologic diseases, other endocrine disorders, and glycated serum protein were found to be independently associated with the prevalence of LEADDP.” It makes little sense to include so many comorbidities in developing the risk prediction model for LEAD. Due to the nature of the cross-sectional design, it cannot be determined whether LEAD cause these diseases or vice versa. At first, it’s okay to include various variables as the candidate for risk factor selection. However, the clinical significance of these variables to be selected as variables should be made clear.

Second, the authors said that there existed gender differences in LEAD distribution and the associations of hematologic diseases with LEAD as new findings. “Currently, no studies have analyzed the correlation between hematologic diseases and LEADDP. This finding opens new avenues for research into LEADDP.” So, is there any clinical significance here to study the association between hematologic diseases and LEAD in people with diabetes? And, what’s the clinical or public health significance here to use this model among those with diabetes? For the discussion part, it’s wordy to present discussion with each risk factor found in this study as a single paragraph. It makes no highlight or focus. And, the key problem is still the finding of this study. So many comorbidities with just gender and glycated serum protein as a risk prediction model makes little clinical or public health significance. Or, the focus of the paper is to study the association between cormobidites with LEAD among those with diabetes?

Besides, is there other risk prediction model not using machine learning algorithms? The authors should discuss this in the discussion part. Is there any added value here in this study by using machine learning algorithms? Or, is this study the first one to study the risk prediction model of LEAD in those with diabetes?

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Aram Baram

Reviewer #3: No

**********

PLoS One. 2024 Dec 30;19(12):e0314862. doi: 10.1371/journal.pone.0314862.r004

Author response to Decision Letter 1

17 Oct 2024

Dear Editors and Reviewers,

Responding to the journal's requirements:

Thank you for your reminder; we have reviewed the reference list again.

Responds to the reviewer's comments:

Reviewer #3:

Thank you very much for your comments. Your feedback is thorough and rigorous, which has been very helpful to us. In response to your remarks, we would like to address the following points.

Significance and Public Health Value of the Study

This study aims to predict the risk of LEAD in diabetic patients and to improve the diagnostic rate of LEAD within the diabetic population, facilitating early detection and intervention to prevent the further progression of LEAD.

As stated in our manuscript, "Statistics indicate that over 60% of LEAD patients may be asymptomatic, leading to early stages of the disease often being overlooked. Compared to non-diabetic LEAD patients, diabetic peripheral neuropathy may mask LEAD symptoms, making the diagnosis of LEADDP more challenging. Current guidelines recommend initial screening for LEAD based on patient interviews and clinical examinations, using the Fontaine or Rutherford scales for assessment. This diagnosis heavily relies on the expertise of specialists and examination conditions, and the diagnostic performance is not ideal." (line 49-56)

Due to a lack of specialized knowledge, diabetic patients may not pay attention to symptoms such as reduced skin temperature in the lower extremities or mild pain, leading them to conceal their medical history and miss timely specialist examinations. They might seek care for diabetes or its complications in endocrinology or other departments, rather than choosing vascular surgery or peripheral vascular disease specialties to address LEAD. This could further result in LEAD being overlooked, delaying treatment. Therefore, we aim to address this issue through our study.

Our model predicts the risk of LEAD at a specific moment by integrating the clinical data of patients, allowing for the identification of high-risk populations. To illustrate this point further, we would like to provide an example. Risk assessment for venous thromboembolism (VTE) has been widely implemented in clinical practice, where risk prediction incorporates not only laboratory test indicators but also factors such as comorbidities. While these comorbidities are associated with VTE, this does not imply that they cause VTE to occur. By evaluating the risk of VTE, high-risk patients can be identified early, enabling timely treatment to prevent further progression of the condition. Similarly, our study aims to identify high-risk LEAD patients to facilitate early detection and intervention.

In our research, we have incorporated machine learning algorithms. Compared to traditional risk prediction methods, machine learning algorithms are better equipped to handle complex datasets containing multiple variables and features. More importantly, the risk prediction models constructed using machine learning have the potential for full automation, thereby reducing the need for healthcare personnel intervention, lowering clinical workloads, and improving diagnostic efficiency. This prospect holds significant social value.

Therefore, we believe that this study holds significant clinical importance and public health value.

Relationship Between Risk Factors and LEAD

LEAD is part of a broader spectrum of vascular diseases, with complex pathophysiological mechanisms. Although numerous studies exist, the detailed pathogenic mechanisms remain unclear. In the diabetic population, LEAD may have unique mechanisms of onset. While these pathogenic mechanisms are causally related to LEAD, their ambiguity prevents us from incorporating them into our risk prediction for LEAD.

By analyzing the relationship between various risk factors and LEAD, we can identify comorbidities that share similar pathogenic mechanisms with LEAD. In our study, we found an association between hematologic diseases and LEAD. Although the common mechanisms underlying this relationship remain unclear, this finding may provide insights for future research. To illustrate this point, we can use the example of patients with polycythemia vera, who may simultaneously experience headaches and lower limb venous thrombosis. While there is an association between headaches and venous thrombosis, this does not imply a causal relationship. By analyzing these associations and conducting further research, we may uncover changes in the patients' blood cells and ultimately identify the mutated genes.

Such findings can help us better understand the pathophysiological processes of LEAD, laying the groundwork for developing more effective medical interventions. Considering these factors, we intend to dedicate a section of our discussion to exploring the potential associations between various risk factors and LEAD in the diabetic population.

Comparison of This Study with Similar Research

You pointed out that "the discussion section should include relevant studies on LEAD risk prediction models," and we agree with your observation. This supplementary section will be outlined in the following paragraph. While we believe that more attention is needed for research on LEAD in diabetic patients, it is important to note that some studies on LEAD risk prediction models already exist. Some of these studies have employed machine learning algorithms, while others have not. Compared to studies that did not use machine learning algorithms, the introduction of machine learning can effectively handle a greater number of variables and complex features, yielding better performance. Furthermore, by continuously updating data and retraining the models, the predictive capabilities can be further enhanced. Additionally, improving machine learning algorithms to establish diagnostic platforms can lower the barrier to use and enhance diagnostic efficiency, offering broader application prospects. Compared to similar studies that utilized machine learning algorithms, the risk prediction model constructed in this study demonstrates certain performance advantages. Moreover, to our knowledge, this is the first study to develop a LEAD risk prediction model using machine learning algorithms specifically for the diabetic population in China.

In response to this issue, we have made additions in the revised manuscript. In lines 258-263, we included the following content: "Previous studies have employed machine learning algorithms to develop risk prediction models for LEAD, achieving relatively good predictive performance. However, these studies did not specifically focus on diabetic patients, and thus could not capture the unique characteristics of LEAD occurrence in this population. The application of machine learning algorithms in LEAD risk prediction models specifically targeting diabetic patients remains relatively limited."

Other changes:

We have added two references in lines 437-444.

Finally, we would like to express our gratitude once again for your assistance in improving this study.

Attachment

Submitted filename: Response to Reviewers.docx

pone.0314862.s003.docx^{(18.7KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0314862.r005

Decision Letter 2

Jincheng Wang

19 Nov 2024

Risk Factors and Predictive Model Construction for Lower Extremity Arterial Disease in Diabetic Patients

PONE-D-24-26075R2

Dear Dr. Kuang,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. If you have any questions relating to publication charges, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Jincheng Wang

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Authors have addressed all comments.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #3: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #3: Yes

**********

6. Review Comments to the Author

Reviewer #3: The authors of this manuscript have addressed all the comments. No further questions to the authors.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #3: No

**********

PLoS One. doi: 10.1371/journal.pone.0314862.r006

Acceptance letter

Jincheng Wang

17 Dec 2024

PONE-D-24-26075R2

PLOS ONE

Dear Dr. Kuang,

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now being handed over to our production team.

At this stage, our production department will prepare your paper for publication. This includes ensuring the following:

* All references, tables, and figures are properly cited

* All relevant supporting information is included in the manuscript submission,

* There are no issues that prevent the paper from being properly typeset

If revisions are needed, the production department will contact you directly to resolve them. If no revisions are needed, you will receive an email when the publication date has been set. At this time, we do not offer pre-publication proofs to authors during production of the accepted work. Please keep in mind that we are working through a large volume of accepted articles, so please give us a few weeks to review your paper and let you know the next and final steps.

Lastly, if your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

If we can help with anything else, please email us at customercare@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Jincheng Wang

Academic Editor

PLOS ONE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 Table. Patient characteristics in the dataset.

(PDF)

pone.0314862.s001.pdf^{(125KB, pdf)}

Attachment

Submitted filename: Response to Reviewers.docx

pone.0314862.s002.docx^{(507.7KB, docx)}

Attachment

Submitted filename: Response to Reviewers.docx

pone.0314862.s003.docx^{(18.7KB, docx)}

Data Availability Statement

[pone.0314862.ref001] 1.Fowkes FGR, Aboyans V, Fowkes FJI, McDermott MM, Sampson UKA, Criqui MH. Peripheral artery disease: epidemiology and global perspectives. Nat Rev Cardiol. 2017;14(3):156–170. doi: 10.1038/nrcardio.2016.179 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref002] 2.Guariguata L, Whiting DR, Hambleton I, Beagley J, Linnenkamp U, Shaw JE. Global estimates of diabetes prevalence for 2013 and projections for 2035. Diabetes Res Clin Pract. 2014;103(2):137–149. doi: 10.1016/j.diabres.2013.11.002 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref003] 3.Jarrett RJ. Type 2 (non-insulin-dependent) diabetes mellitus and coronary heart disease-chicken, egg or neither? Diabetologia. 1984;26(2):99–102. doi: 10.1007/BF00281114 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref004] 4.Stern MP. Diabetes and cardiovascular disease. The “common soil” hypothesis. Diabetes. 1995;44(4):369–374. doi: 10.2337/diab.44.4.369 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref005] 5.Li J, Li Y, Ma W, Liu Y, Yin X, Xie C, et al. Association of Time in Range levels with Lower Extremity Arterial Disease in patients with type 2 diabetes. Diabetes Metab Syndr. 2020;14(6):2081–2085. doi: 10.1016/j.dsx.2020.09.028 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref006] 6.American Diabetes Association. 11. Microvascular Complications and Foot Care: Standards of Medical Care in Diabetes-2021. Diabetes Care. 2021;44(Suppl 1):S151–S167. doi: 10.2337/dc21-S011 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref007] 7.Rutherford RB, Baker JD, Ernst C, Johnston KW, Porter JM, Ahn S, et al. Recommended standards for reports dealing with lower extremity ischemia: revised version. J Vasc Surg. 1997;26(3):517–538. doi: 10.1016/s0741-5214(97)70045-4 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref008] 8.Nativel M, Potier L, Alexandre L, Baillet-Blanco L, Ducasse E, Velho G, et al. Lower extremity arterial disease in patients with diabetes: a contemporary narrative review. Cardiovasc Diabetol. 2018;17(1):138. doi: 10.1186/s12933-018-0781-1 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref009] 9.Buso G, Aboyans V, Mazzolai L. Lower extremity artery disease in patients with type 2 diabetes. Eur J Prev Cardiol. 2019;26(2_suppl):114–124. doi: 10.1177/2047487319880044 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref010] 10.Chen T, Xiao S, Chen Z, Yang Y, Yang B, Liu N. Risk factors for peripheral artery disease and diabetic peripheral neuropathy among patients with type 2 diabetes. Diabetes Res Clin Pract. 2024;207:111079. doi: 10.1016/j.diabres.2023.111079 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref011] 11.Carter SA. The effect of cooling on toe systolic pressures in subjects with and without Raynaud’s syndrome in the lower extremities. Clin Physiol. 1991;11(3):253–261. doi: 10.1111/j.1475-097x.1991.tb00456.x [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref012] 12.Zuo H, Shi Z, Hussain A. Prevalence, trends and risk factors for the diabetes epidemic in China: a systematic review and meta-analysis. Diabetes Res Clin Pract. 2014. Apr;104(1):63–72. doi: 10.1016/j.diabres.2014.01.002 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref013] 13.Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al. Global Burden of Cardiovascular Diseases and Risk Factors, 1990–2019: Update From the GBD 2019 Study. J Am Coll Cardiol. 2020;76(25):2982–3021. doi: 10.1016/j.jacc.2020.11.010 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref014] 14.Aboyans V, Criqui MH, McClelland RL, Allison MA, McDermott MM, Goff DC Jr, et al. Intrinsic contribution of gender and ethnicity to normal ankle-brachial index values: the Multi-Ethnic Study of Atherosclerosis (MESA). J Vasc Surg. 2007;45(2):319–327. doi: 10.1016/j.jvs.2006.10.032 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref015] 15.Collins TC, Suarez-Almazor M, Bush RL, Petersen NJ. Gender and peripheral arterial disease. J Am Board Fam Med JABFM. 2006;19(2):132–140. doi: 10.3122/jabfm.19.2.132 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref016] 16.Mohammedi K, Woodward M, Hirakawa Y, Zoungas S, Williams B, Lisheng L, et al. Microvascular and Macrovascular Disease and Risk for Major Peripheral Arterial Disease in Patients With Type 2 Diabetes. Diabetes Care. 2016. Oct;39(10):1796–803. doi: 10.2337/dc16-0588 Epub 2016 Jul 25. [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref017] 17.Li Y, Liu Y, Liu S, Gao M, Wang W, Chen K, et al. Diabetic vascular diseases: molecular mechanisms and therapeutic strategies. Signal Transduct Target Ther. 2023;8(1):152. doi: 10.1038/s41392-023-01400-z [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref018] 18.Zhou X, Yu L, Zhao Y, Ge J. Panvascular medicine: an emerging discipline focusing on atherosclerotic diseases. Eur Heart J. 2022;43(43):4528–4531. doi: 10.1093/eurheartj/ehac448 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref019] 19.Meijer WT, Grobbee DE, Hunink MG, Hofman A, Hoes AW. Determinants of peripheral arterial disease in the elderly: the Rotterdam study. Arch Intern Med. 2000;160(19):2934–2938. doi: 10.1001/archinte.160.19.2934 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref020] 20.Newman AB, Siscovick DS, Manolio TA, Polak J, Fried LP, Borhani NO, et al. Ankle-arm index as a marker of atherosclerosis in the Cardiovascular Health Study. Cardiovascular Heart Study (CHS) Collaborative Research Group. Circulation. 1993;88(3):837–845. doi: 10.1161/01.cir.88.3.837 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref021] 21.Haas AV, McDonnell ME. Pathogenesis of Cardiovascular Disease in Diabetes. Endocrinol Metab Clin North Am. 2018. Mar;47(1):51–63. doi: 10.1016/j.ecl.2017.10.010 . [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref022] 22.Adler AI, Stevens RJ, Neil A, Stratton IM, Boulton AJM, Holman RR. UKPDS 59: hyperglycemia and other potentially modifiable risk factors for peripheral vascular disease in type 2 diabetes. Diabetes Care. 2002;25(5):894–899. doi: 10.2337/diacare.25.5.894 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref023] 23.Althouse AD, Abbott JD, Forker AD, Bertolet M, Barinas-Mitchell E, Thurston RC, et al. Risk factors for incident peripheral arterial disease in type 2 diabetes: results from the Bypass Angioplasty Revascularization Investigation in type 2 Diabetes (BARI 2D) Trial. Diabetes Care. 2014;37(5):1346–1352. doi: 10.2337/dc13-2303 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref024] 24.Norman PE, Davis WA, Bruce DG, Davis TME. Peripheral arterial disease and risk of cardiac death in type 2 diabetes: the Fremantle Diabetes Study. Diabetes Care. 2006;29(3):575–580. doi: 10.2337/diacare.29.03.06.dc05-1567 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref025] 25.Mohammedi K, Woodward M, Hirakawa Y, Zoungas S, Williams B, Lisheng L, et al. Microvascular and Macrovascular Disease and Risk for Major Peripheral Arterial Disease in Patients With Type 2 Diabetes. Diabetes Care. 2016;39(10):1796–1803. doi: 10.2337/dc16-0588 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref026] 26.Yaribeygi H, Farrokhi FR, Butler AE, Sahebkar A. Insulin resistance: Review of the underlying molecular mechanisms. J Cell Physiol. 2019;234(6):8152–8161. doi: 10.1002/jcp.27603 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref027] 27.Welsh GI, Hale LJ, Eremina V, Jeansson M, Maezawa Y, Lennon R, et al. Insulin signaling to the glomerular podocyte is critical for normal kidney function. Cell Metab. 2010;12(4):329–340. doi: 10.1016/j.cmet.2010.08.015 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref028] 28.Libby P. The changing landscape of atherosclerosis. Nature. 2021;592(7855):524–533. doi: 10.1038/s41586-021-03392-8 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref029] 29.Bertrand C, Saulnier PJ, Potier L, Croyal M, Blanchard V, Gand E, et al. Plasma concentrations of lipoproteins and risk of lower-limb peripheral artery disease in people with type 2 diabetes: the SURDIAGENE study. Diabetologia. 2021;64(3):668–680. doi: 10.1007/s00125-020-05326-x [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref030] 30.Eshcol J, Jebarani S, Anjana RM, Mohan V, Pradeepa R. Prevalence, incidence and progression of peripheral arterial disease in Asian Indian type 2 diabetic patients. J Diabetes Complications. 2014;28(5):627–631. doi: 10.1016/j.jdiacomp.2014.04.013 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref031] 31.Yeo TC, Chan YH, Low LP, Venketasubramanian N, Lim SC, Tay JC, et al. Risk factor profile and treatment patterns of patients with atherothrombosis in Singapore: insight from the REACH Registry. Ann Acad Med Singapore. 2008;37(5):365–371. [PubMed] [Google Scholar]

[pone.0314862.ref032] 32.Criqui MH, Matsushita K, Aboyans V, Hess CN, Hicks CW, Kwan TW, et al. Lower Extremity Peripheral Artery Disease: Contemporary Epidemiology, Management Gaps, and Future Directions: A Scientific Statement From the American Heart Association. Circulation. 2021;144(9):e171–e191. doi: 10.1161/CIR.0000000000001005 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref033] 33.Gerhard-Herman MD, Gornik HL, Barrett C, Barshes NR, Corriere MA, Drachman DE, et al. 2016 AHA/ACC Guideline on the Management of Patients With Lower Extremity Peripheral Artery Disease: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Circulation. 2017;135(12):e686–e725. doi: 10.1161/CIR.0000000000000470 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref034] 34.Heffron SP, Dwivedi A, Rockman CB, Xia Y, Guo Y, Zhong J, et al. Body mass index and peripheral artery disease. Atherosclerosis. 2020;292:31–36. doi: 10.1016/j.atherosclerosis.2019.10.017 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref035] 35.Mann JFE, Gerstein HC, Pogue J, Bosch J, Yusuf S, for the HOPE Investigators. Renal Insufficiency as a Predictor of Cardiovascular Outcomes and the Impact of Ramipril: The HOPE Randomized Trial. Ann Intern Med. 2001;134(8):629–636. doi: 10.7326/0003-4819-134-8-200104170-00007 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref036] 36.Checheriţă IA, Smarandache D, David C, Ciocâlteu A, Ion DA, Lascăr I. Vascular calcifications in chronic kidney disease—clinical management. Romanian J Morphol Embryol Rev Roum Morphol Embryol. 2012;53(1):7–13. [PubMed] [Google Scholar]

[pone.0314862.ref037] 37.Stabouli S, Papakatsika S, Kotsis V. Hypothyroidism and hypertension. Expert Rev Cardiovasc Ther. 2010;8(11):1559–1565. doi: 10.1586/erc.10.141 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref038] 38.Biondi B, Klein I. Hypothyroidism as a risk factor for cardiovascular disease. Endocrine. 2004;24(1):1–13. doi: 10.1385/ENDO:24:1:001 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref039] 39.Ferriere A, Tabarin A. Cushing’s syndrome: Treatment and new therapeutic approaches. Best Pract Res Clin Endocrinol Metab. 2020;34(2):101381. doi: 10.1016/j.beem.2020.101381 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref040] 40.Lassila R. Role and management of coagulation disorders in peripheral arterial disease. Scand J Surg SJS Off Organ Finn Surg Soc Scand Surg Soc. 2012;101(2):94–99. doi: 10.1177/145749691210100204 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref041] 41.Salhiyyah K, Forster R, Senanayake E, Abdel-Hadi M, Booth A, Michaels JA. Pentoxifylline for intermittent claudication. Cochrane Database Syst Rev. 2015;9(9):CD005262. doi: 10.1002/14651858.CD005262.pub3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref042] 42.Hampe N, Wolterink JM, van Velzen SGM, Leiner T, Išgum I. Machine Learning for Assessment of Coronary Artery Disease in Cardiac CT: A Survey. Front Cardiovasc Med. 2019;6. doi: 10.3389/fcvm.2019.00172 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref043] 43.Santos MS, Abreu PH, García-Laencina PJ, Simão A, Carvalho A. A new cluster-based oversampling method for improving survival prediction of hepatocellular carcinoma patients. J Biomed Inform. 2015;58:49–59. doi: 10.1016/j.jbi.2015.09.012 [DOI] [PubMed] [Google Scholar]

[pone.0314862.ref044] 44.Wang X, Yang Y, Zhang J, Zang S. Development and validation of a prediction model for the prolonged length of stay in Chinese patients with lower extremity atherosclerotic disease: a retrospective study. BMJ Open. 2023. Feb 9;13(2):e069437. doi: 10.1136/bmjopen-2022-069437 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref045] 45.Zhang L, Ma Y, Li Q, Long Z, Zhang J, Zhang Z, et al. Construction of a novel lower-extremity peripheral artery disease subtype prediction model using unsupervised machine learning and neutrophil-related biomarkers. Heliyon. 2024. Jan 6;10(2):e24189. doi: 10.1016/j.heliyon.2024.e24189 ; PMCID: PMC10827514 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0314862.ref046] 46.Pearce J, Ferrier S. Evaluating the predictive performance of habitat models developed using logistic regression. Ecol Model. 2000;133(3):225–245. doi: 10.1016/S0304-3800(00)00322-7 [DOI] [Google Scholar]

PERMALINK

Risk factors and predictive model construction for lower extremity arterial disease in diabetic patients

Yingjie Kuang

Zhixin Cheng

Jun Zhang

Chunxu Yang

Yue Zhang

Roles

Abstract

Objective

Methods

Results

Conclusion

Introduction

Data and methods

Fig 1. Heatmap of variable correlations.

Table 1. Characteristics of samples in the training and validation sets.

Results

Table 2. Baseline characteristics according to the presence of LEADDP.

Table 3. Results of the multivariate analysis.

Fig 2. AUC curves of various risk prediction models.

Discussion

Conclusion

Supporting information

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Adeel Ahmad Khan

Roles

Author response to Decision Letter 0

Decision Letter 1

Jincheng Wang

Roles

Author response to Decision Letter 1

Decision Letter 2

Jincheng Wang

Roles

Acceptance letter

Jincheng Wang

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases