Development and validation of a risk nomogram model for predicting superficial fungal infections in patients with type 2 diabetes mellitus : a cross-sectional study

Abstract

Objective

To develop and validate a nomogram for the individualized prediction of superficial fungal infections (SFI) risk in patients with type 2 diabetes mellitus (T2DM).

Methods

This cross-sectional study enrolled patients with T2DM from the Affiliated Anning First People’s Hospital of Kunming University of Science and Technology between December 2023 and December 2024. Risk factors were identified using multivariable logistic regression, and a nomogram was developed subsequently for predicting SFI in T2DM patients. The model’s performance was evaluated using receiver operating characteristic (ROC) curves, calibration curves, decision curve analysis (DCA) and clinical impact curve (CIC) in the training and validation sets.

Results

Among 308 hospitalized T2DM patients screened by multiplex quantitative polymerase chain reaction (qPCR), 220 (71.40%) were diagnosed with SFI. Trichophyton rubrum (107 cases) was the most common pathogen, and monoinfection was the most frequent presentation (90 patients). Patients were randomly divided into a training (n = 216) and a validation cohort (n = 92) in a 7:3 ratio. Multivariable logistic regression analysis identified eight key variables: body mass index (BMI), blood glucose (Glu), hemoglobin A1c (HbA1c), urinary albumin-to-creatinine ratio (UACR), serum potassium (K⁺), sodium (Na⁺), hyperlipidemia (HLP) and hypertension (HTN). The nomogram demonstrated excellent predictive ability. The ROC analysis indicated good discrimination in the training cohort (area under the curve (AUC) = 0.966; 95% CI, 0.945–0.987) and the validation cohort (AUC = 0.931; 95% CI, 0.877–0.985). The optimal cut-point of the nomogram was 0.624 with a sensitivity of 92.3% and specificity of 86.7% (Youden’s index: 0.79). Calibration curves showed good agreement. DCA confirmed the clinical utility of the nomogram.

Conclusion

This nomogram effectively predicts the risk of SFI in T2DM patients and provides an objective tool to facilitate early identification and intervention by clinicians.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-026-12613-2.

Introduction

Diabetes mellitus is a chronic metabolic disease characterized by hyperglycemia, polydipsia, polyphagia, and polyuria resulting from defects in insulin secretion or action [1, 2]. It represents a critical global public health challenge, characterized by rising prevalence and the broad spectrum of complications that exact a heavy burden on individuals, healthcare systems, and economies [3]. Diabetes mellitus is primarily classified into four types: T1DM, T2DM, other specific types (e.g., genetic forms) [4], and gestational diabetes mellitus. T2DM accounts for 90–95% of all diabetes cases worldwide, affecting an estimated 589 to 828 million individuals [5]. Under current trends, the number of people living with diabetes is projected to exceed 1.31 billion by 2050. China now has the largest population of individuals with diabetes, representing nearly one-quarter of the global prevalence [6]. Addressing the growing challenges of prevention and effective management of the disease and its complications remains a central priority for healthcare systems worldwide [1]. Diabetes mellitus is associated with a range of metabolic and immunological complications, such as diabetic nephropathy and cutaneous disorders [7, 8]. Cutaneous diseases are present in 79.20% of people with diabetes and are more common in T2DM than T1DM, and they may appear as the initial presentation or at any stage of the disease [9].

Diabetes impairs the skin’s innate barrier, which predisposes patients to lesions, ulcers and makes fungal infections one of the most common dermatologic disorders in this population [10, 11]. In diabetes, elevated HbA1c levels, peripheral neuropathy, and impaired peripheral perfusion increase susceptibility to SFI [12]. Furthermore, SFI can increase the risk of diabetic foot ulcers (DFU) and diabetic foot infections contributing to substantial morbidity and mortality [12]. Early screening in high-risk populations facilitates timely interventions that mitigate the onset and progression of SFI. However, there is currently a lack of clinical prediction models for the early identification of T2DM patients at risk of SFI. To address this gap, we aimed to identify risk factors for SFI and to develop and validate a nomogram to predict and prevent SFI in patients with T2DM.

Yunnan’s complex topography and diverse climate make it an ideal region for fungal propagation. The extremely high fungal diversity in this area has attracted significant attention from mycologists both domestically and internationally [13]. However, studies on cutaneous fungal infections in diabetic patients in Yunnan remain limited. SFI, commonly referred to as “tinea”, is conventionally diagnosed using microscopy and fungal culture which remain the gold standard for confirmed SFI [14]. However, these methods have several limitations, including low sensitivity, prolonged growth time and the need for trained mycologists to differentiate species [15]. In response to evolving clinical needs, the field of fungal diagnostics has shifted from reliance on traditional microscopy and fungal culture toward more advanced non-culture-based methods, such as novel PCR assays, next-generation sequencing, and artificial intelligence-based models, which are revolutionizing fungal diagnostics and identification [16]. Multiplexed qPCR can simultaneously detect multiple species, with high sensitivity and specificity, while also reducing costs, shortening turnaround time, and decreasing labor requirements [16, 17]. Several systematic reviews and studies have identified Trichophyton rubrum [18], Candida albicans [19], Malassezia spp. [20], Candida parapsilosis [21], Trichophyton mentagrophytes [22], Epidermophyton floccosum [23], Microsporum canis [24] and Scopulariopsis brevicaulis [25] as the primary etiological agents of superficial mycoses in the diabetic population. Based on these findings, we developed a multiplex qPCR assay capable of detecting these major pathogenic fungi.

Materials and methods

Study design and patients

This cross-sectional study was conducted by successfully 308 consecutive T2DM patients hospitalized at The Affiliated Anning First People’s Hospital of Kunming University of Science and Technology from December 2023 to December 2024.

Inclusion criteria: (1) Diagnosis of T2DM confirmed by the World Health Organization criteria [26]. defined as fasting plasma glucose ≥ 7.0 mmol/L, and/or random blood glucose ≥ 11.1 mmol/L, and/or 2-hour post-oral glucose tolerance test glucose ≥ 11.1 mmol/L, and/or HbA1c ≥ 6.5%. (2) Provision of written informed consent and ability to complete the sampling and testing protocol.

Exclusion criteria: (1) Diagnosis of diabetes mellitus other than T2DM (e.g., type 1 diabetes, gestational diabetes). (2) Human immunodeficiency virus infection, solid organ transplantation history, prolonged corticosteroid or immunosuppressant use. (3) Use of any systemic or topical antifungal medication within the preceding 3 months. (4) Pregnancy, lactation, or severe hepatic or renal dysfunction. (5) Cognitive impairment or any condition preventing study completion or compromising data quality.

Sample collection and multiplex qPCR assay construction

The sampling procedure was explained to all participants to ensure their cooperation. Prior to sampling, the skin surface was gently wiped with physiological saline and allowed to dry completely. Skin samples were collected from all infected body surfaces excluding nails and hair, including the hands (palmar and dorsal aspects), feet (plantar and dorsal aspects), trunk (anterior, posterior, and lateral surfaces), and head (scalp and face, where applicable). Each contiguous, visibly demarcated infected skin area was designated as a “target sampling region”. To ensure systematic and representative coverage, the target region was conceptually partitioned into four quadrants by establishing two perpendicular imaginary axes intersecting at the center of the infection zone. This method minimizes spatial sampling bias and ensures thorough sampling of all areas within the infected site, including the central zone, peripheral margins, and adjacent transitional skin. The quadrants were labeled as upper-left, upper-right, lower-left, and lower-right relative to the anatomical orientation of the patient. A sterile swab was held by the handle to avoid contamination, applied with moderate pressure to the skin, and moved in a back-and-forth zigzag pattern at least five times per quadrant. The swab was rotated 90° axially between quadrants. Separate sterile swabs were used for each specimen, and environmental control samples were simultaneously collected to exclude potential contamination. A sterile swab was first moistened with physiological saline (consistent with the pre-sampling preparation for patient samples) and then exposed to the ward environment during the patient sampling period. After completion of patient sample collection, the environmental swab was immediately placed into a dedicated sample collection buffer and labeled separately to avoid cross-contamination. Thus, a single sample collection process was completed.

Specific primers and probes targeting unique gene fragments of the eight pathogens were designed and validated by singleplex qPCR, demonstrating high specificity (no cross-reactivity) and sensitivity (detection limit: 1 copy/µL). For multiplex qPCR validation, the primers and probes were organized into two reaction panels to evaluate specificity, sensitivity, repeatability, and accuracy. The multiplex assays maintained high specificity without cross-reactivity within or between groups, a detection limit of 10 copies/µL, and excellent repeatability (intra- and inter-assay coefficient of variation < 5%). No false positives/negatives were observed, and multiplex qPCR results showed high concordance with Sanger sequencing, confirming robust accuracy.

Data collection

The dependent variable was defined as the presence of SFI. Independent variables were obtained from the electronic medical record system and categorized as follows:

Clinical indicators: sex, age, BMI, and occupation.

Biochemical indicators: HbA1c, Glu, UACR, K⁺, Na⁺, and pH.

Diabetic complications and comorbidities: Coronary atherosclerotic heart disease (CAD), diabetic peripheral vascular disease (DPVD), diabetic peripheral neuropathy (DPN), diabetic kidney disease (DKD), obesity, diabetic retinopathy (DR), DFU, diabetic ketoacidosis (DKA), HLP, HTN, vitamin D deficiency (VD).

For logistic regression–based prediction models, it is generally recommended that cross-sectional cohorts include approximately 5–10 patients per predictor to limit prediction error [27]. This study included 21 variables as potential risk factors; therefore, at least 105 participants with SFI in T2DM were required to ensure predictive accuracy. Ultimately, 308 participants were successfully enrolled and randomly divided into training and validation cohort at a 7:3 ratio. Of these participants, 156 were diagnosed with SFI and allocated to the training cohort, and this sample size met the minimum requirement. The predictive model was developed in the training cohort and its predictive performance was evaluated in the validation cohort [28]. In the training cohort, multivariable logistic regression analyse was performed to identify independent predictors of SFI [29]. All variables with P < 0.05 in univariate analysis were considered candidate predictors for the multivariable model and were used to construct the nomogram (Table 1).

Table 1.

Risk factors for SFI in patients with T2DM in the training cohort

Variable	SFI (156)	Non-SFI (60)	P-value
Sex
Male	86 (55.10%)	29 (48.30%)	0.370
Female	70 (44.90%)	31 (51.70%)
Age (years)	60 (53,73)	59 (50,67)	0.112
BMI (kg/m2)	24.64 (22.37, 27.10)	22.50 (21.12, 24.95)	< 0.001
HbA1c (mmol/L)	10.20 (8.53, 12.50)	6.50 (6.20, 6.98)	< 0.001
Glu (mmol/L)	13.20 (10.53, 18.83)	6.35 (5.90, 7.30)	< 0.001
UACR	36.50 (10.83, 113.43)	12.85 (7.83, 21.33)	< 0.001
< 30	73 (46.80%)	54 (90.00%)	< 0.001
30–300	63 (40.40%)	6 (10.00%)
> 300	20 (12.80%)	0 (0.00%)
Na+ (mmol/L)	137.25 (135.00,140.68)	139.00 (137.25, 140.63)	0.021
K+ (mmol/L)	4.02 (3.72, 4.30)	3.90 (3.60, 4.10)	0.011
PH	7.41 (7.39, 7.44)	7.42 (7.40, 7.45)	0.301
Occupation
Employee	11 (7.10%)	5 (8.30%)	0.845
Farmer	54 (34.60%)	17 (28.30%)
Freelancer	12 (7.70%)	6 (10.00%)
Retiree	63 (40.40%)	25 (41.70%)
Student	1 (0.60%)	2 (3.30%)
Small business owner	3 (1.90%)	1 (1.70%)
Unemployed people	4 (2.60%)	1 (1.70%)
Worker	8 (5.10%)	3 (5.00%)
DPVD
Yes	43 (27.6%)	8 (13.3%)	0.027
No	113 (72.4%)	52 (86.7%)
DPN
Yes	127 (81.40%)	28 (46.70%)	< 0.001
No	29 (18.60%)	32 (53.30%)
DKD
Yes	60 (38.50%)	7 (11.70%)	< 0.001
No	96 (61.50%)	53 (88.30%)
CAD
Yes	35 (22.40%)	5 (8.30%)	0.017
No	121 (77.60%)	55 (91.70%)
Adiposity
Yes	10 (6.40%)	8 (13.30%)	0.099
No	146 (93.60%)	52 (86.70%)
VD
Yes	39 (25.00%)	5 (8.30%)	0.006
No	117 (75.00%)	55 (91.70%)
DR
Yes	31 (19.90%)	3 (5.00%)	0.007
No	125 (80.10%)	57 (95.00%)
DFU
Yes	10 (6.4%)	1 (1.70%)	0.297
No	146 (93.6%)	59 (98.30%)
HLP
Yes	71 (45.50%)	12 (20.00%)	0.001
No	85 (54.50%)	48 (80.00%)
HTN
Yes	86 (55.10%)	15 (25.00%)	< 0.001
No	70 (44.90%)	45 (75.00%)
DKA
Yes	41 (26.30%)	12 (20.00%)	0.337
No	115 (73.70%)	48 (80.00%)

Abbreviations: SFI superficial fungal infections, BMI body mass index, Glu blood glucose, HbA1c hemoglobin A1c, UACR urinary albumin-to-creatinine ratio, K⁺ serum potassium, Na⁺ sodium, HLP hyperlipidemia, HTN hypertension, DFU diabetic foot ulcers, CAD Coronary atherosclerotic heart disease, DPVD diabetic peripheral vascular disease, DPN diabetic peripheral neuropathy, DKD diabetic kidney disease, DR diabetic retinopathy, DKA diabetic ketoacidosis, VD vitamin D deficiency

Statistical analysis

Data were analyzed with GraphPad Prism 9.5.0, SPSS 26.0 and R version 4.5.1. The “pROC” and “rms” packages in R were used to build ROC curves, Calibration curves and nomogram. DCA and CIC were built using “rmda” package. The Missing values for PH and UACR (n = 2; 0.6%) were imputed using median substitution [30]. Continuous variables were expressed as mean ± standard deviation or as median (interquartile range), whereas categorical variables were summarized as frequencies and percentages. Normality was assessed using the Shapiro-Wilk test. Group comparisons used the Wilcoxon rank-sum test or chi-square tests, as appropriate. Independent factors identified by multivariable logistic regression using a backward stepwise selection (P < 0.05) were incorporated into a nomogram model (Table 2). The model’s performance was assessed using ROC, CIC, DCA, and calibration curves. The UACR variable displayed a highly skewed distribution with multiple outliers and was initially categorized into three clinically relevant groups (< 30, 30–300, and > 300 mg/g; Supplemental Fig. 1). To avoid quasi-complete separation in logistic regression, UACR was further dichotomized (< 30 vs. ≥ 30 mg/g) to ensure model stability (Supplemental Table 2).

Table 2.

Prediction factors for SFI in patients with T2DM in multivariate logistics regression analysis

Variable	OR (95% CI)	P-value
BMI	1.194 (1.020, 1.397)	0.027
Glu	1.593 (1.287, 1.972)	< 0.001
HbA1c	1.909 (1.402, 2.599)	< 0.001
K+	8.578 (1.752, 41.996)	0.008
Na+	1.360 (1.099, 1.683)	0.005
UACR (mg/g)
< 30	Reference
≥ 30	8.140 (1.926, 34.404)	0.04
DR
No	Reference
Yes	11.507 (0.781, 169.626)	0.075
HLP
No	Reference
Yes	7.632 (1.907, 30.540)	0.004
HTN
No	Reference
Yes	3.956 (1.138, 13.755)	0.031

Abbreviations: OR odds ratio, BMI body mass index, Glu blood glucose, HbA1c hemoglobin A1c, UACR urinary albumin-to-creatinine ratio, K⁺ serum potassium, Na⁺ sodium, HLP hyperlipidemia, HTN hypertension, DR diabetic retinopathy

Results

Clinical characteristics between SFI and non-SFI groups

Among 308 patients with T2DM, 220 (71.43%) were diagnosed with SFI. Trichophyton rubrum was the most frequently isolated pathogen (107 cases), followed by Candida albicans (72 cases). Monoinfection was the most common presentation (90 patients), followed by dual infection (78 patients), as shown in Fig. 1. No significant difference was observed in sex distribution. However, patients with SFI were slightly older and had a higher BMI, as well as elevated levels of Glu, HbA1c, and UACR, suggesting poorer glycemic control, more severe metabolic and renal dysregulation (all P < 0.05). Compared to those without SFI, the SFI group showed higher incidences of diabetic complications including DPVD, DPN, DKD, CAD, VD, DR, DFU, HLP, and HTN (all P < 0.05) (Supplemental Table 1).

Fig. 1.

Spectrum and multiplicity of fungal infections. (A) Cases distribution by fungal species. (B) Profile of mono and poly infections

Construction of the predicting nomogram model for SFI in T2DM patients

The univariate analysis was first performed on the training set. Variables with P < 0.05 in the univariate analysis were subsequently entered into the multivariable analysis. These variables included BMI, Glu, HbA1c, UACR, K⁺, Na⁺, DPVD, DPN, DKD, CAD, VD, DR, HLP, HTN (Table 1). Based on the results of the multivariable logistic regression analyses (Table 2), a prediction model for SFI was developed based on variables significantly associated in the multivariable analysis and clinically relevant, including BMI, Glu, HbA1c, K, Na, UACR, HLP, and HTN. The nomogram comprised twelve lines; lines 2–9 represented the model variables. Scores for each variable were summed to yield a total score, which was then projected onto the “Total Points” line to obtain the predicted probability at the bottom of the chart (Fig. 2).

Fig. 2.

Nomogram predicting the probability of SFI in patients with T2DM. Clinicians can apply the nomogram as follows: (1) Obtain the assigned points for each predictor (BMI, UACR, K⁺, Na⁺, Glu, HbA1c, HLP, HTN); (2) Sum the points of all predictors to calculate a total score; (3) Locate the total score on the corresponding axis to derive the predicted SFI probability. Abbreviations: BMI body mass index, Glu blood glucose, HbA1c hemoglobin A1c, UACR urinary albumin-to-creatinine ratio, K⁺ serum potassium, Na⁺ sodium, HLP hyperlipidemia, HTN hypertension

Evaluation of nomogram

The nomogram exhibited excellent discriminative ability, in the training cohort, the AUC was 0.966 (95% CI: 0.945–0.987; Fig. 3A). In the validation cohort, the AUC was 0.931 (95% CI: 0.877–0.985; Fig. 3B), indicating strong predictive performance in both cohorts. The optimal cut-point of the nomogram was 0.624 with a sensitivity of 92.3% and specificity of 86.7% (Youden’s index: 0.79). The model’s stratification of patients into three risk tiers (low: < 0.312; medium: 0.312–0.624; high: > 0.624) demonstrated robust capacity, as evidenced by a marked gradient in infection rates from 4.3% to 17.4%, then to 45.5%–58.8% and finally to 94.7%–94.8% (Fig. 4). The calibration curve, constructed with 1,000 bootstrap resamples, demonstrated good agreement between the predicted probabilities and the actual outcomes. The Brier scores were 0.063 and 0.062 in the training and validation cohorts, respectively, which demonstrating the model’s excellent calibration in both cohorts (Fig. 5). The DCA showed that the nomogram model provided higher net benefit than the treat-all or none strategies across threshold probabilities between 0.1 and 1.0, demonstrating strong clinical applicability in both the training and validation cohorts (Fig. 6). Furthermore, the CIC revealed a consistent ability in identifying at-risk populations while maintaining a favorable balance between true positive predictions and the overall number of individuals classified as high risk (Fig. 7). Collinearity diagnostics were performed using the variance inflation factor (VIF). The results indicated that there was no significant collinearity among all predictors incorporated in the model, with all VIF values less than 5, confirming the stability of the regression model (Supplemental Table 3).

Fig. 3.

ROC curves of the prediction model for the training and validation cohorts. (A) training cohort (AUC = 0.966), (B) internal validation cohort (AUC = 0.931) with an optimal cut-off value of 0.624 for both cohorts, respectively. Abbreviations: ROC receiver operating characteristic, AUC Area Under the Curve

Fig. 4.

Distribution of infection rates in the training and validation cohorts across different risk stratifications. Infection rate distribution across risk stratifications (Low: < 0.312; Medium: 0.312–0.624; High: > 0.624) in training (blue) and validation (orange) cohorts. Infection rates rise sharply with increasing risk, and results are highly consistent between datasets, confirming stable model performance in risk stratification

Fig. 5.

Calibration curves of the prediction model for the training and internal validation cohorts. (A) training cohort, (B) internal validation cohort. The plots display the agreement between predicted and observed probabilities of infection in the training and validation cohorts. The dashed line (“Ideal”) represents perfect calibration, the dotted line (“Apparent”) shows the uncorrected performance, and the solid line (“Bias-corrected”) indicates the bootstrap-corrected calibration curve. Abbreviations: B bootstrap

Fig. 6.

DCA of the prediction model for the the training and internal validation cohorts. (A) training cohort, (B) internal validation cohort. In both cohorts, the nomogram (red line) showed greater net benefit than the treat-all (dashed line) and treat-none (solid line) strategies across most threshold probabilities

Fig. 7.

CIC of the prediction model. (A) training cohort, (B) internal validation cohort. The curves depict the number of patients identified as high risk (solid red line) and the number of true positive cases (dashed blue line) across a range of risk thresholds (0.0–1.0) and corresponding cost: benefit ratios in the cohorts

Discussion

SFI can be challenging to treat in diabetic patients. Therefore, early prevention is particularly important [31]. This study systematically demonstrates that several variables (BMI, Glu, HbA1c, UACR, K⁺, Na⁺, HLP and HTN) are the independent predictors of SFI risk (P < 0.05 for all) in patients with T2DM. Furthermore, we developed and successfully validated a novel nomogram to assess the risk of SFI in individuals with T2DM. The ROC and calibration curves demonstrated robust discrimination and calibration of our nomogram. Additionally, the DCA and CIC confirmed its clinical utility, supporting its use as a simple, practical clinical tool to assist clinicians in effectively predicting the risk of SFI. Diabetic patients often exhibit impaired immune function, increasing their susceptibility to infections [32]. In our study cohort, the prevalence of SFI was 71.43%, notably higher than rates reported in previous studies [31, 33]. Trichophyton rubrum was the most common dermatophyte isolated, and mixed fungal infections were predominated among infected patients, a finding consistent with other domestic and international studies [34–36]. T2DM is a metabolic disorder characterized by disturbances in glucose, protein, water, and electrolyte homeostasis [37]. BMI, Glu, and HbA1c were identified as significant predictors of SFI. These results are consistent with previous research [38–40]. Hyperglycemia and elevated HbA1c levels have been identified as key drivers of SFI pathogenesis. High tissue glucose concentrations not only provide an abundant carbon source that promotes microbial growth and virulence but also directly impair leukocyte functions (e.g., chemotaxis and phagocytosis), thereby creating a microenvironment conducive to fungal infection [41, 42]. Our findings, together with existing evidence, reinforce the association between poor glycemic control and SFI [43]. Therefore, maintaining a normal BMI is essential for mitigating the risk of SFI in patients with T2DM. HLP has also been identified as a risk factor for SFI pathogenesis [44]. Furthermore, while current evidence only indirectly links K⁺, Na⁺, and UACR to SFI in T2DM, our study provides direct evidence confirming these associations. Potential reasons may include metabolic disturbances that impair host defenses, as reflected by dysregulated K ⁺ and Na ⁺ levels. Elevated UACR also signals systemic microvascular and immune dysfunction; however, these findings require further corroboration. While the relationship between HTN and fungal infections remains uncertain, our study identified HTN as an independent risk factor for SFI. However, DR showed a marginally association with SFI (P = 0.075), a trend that may reflect the shared pathological basis of microangiopathy between DR and SFI. As a borderline predictive factor, DR may offer supplementary value for SFI risk stratification, though its predictive efficacy requires validation in larger cohorts. The model’s stratification into low-, moderate-, and high-risk tiers, corresponding to infection rates of 4.3%–17.4%, 45.5%–58.8%, and 94.7%–94.8%, respectively, effectively informs the following tiered intervention strategy. This strategy ranges from intensive management (immediate antifungal therapy and glycemic control for high-risk patients) to proactive monitoring (monthly follow-up for moderate-risk individuals) and routine care (annual examination for low-risk patients), thereby enabling precise resource allocation and personalized prevention.

Limitations

This study has several limitations. First, selecting hospitalized patients with severe disease may introduce selection bias and inflate SFI rates, limiting generalizability to outpatients. Second, the single-center, cross-sectional design restricts causal inference, while the modest sample size limits statistical power. Third, sampling was restricted to lesional areas without paired non-lesional controls or head-to-head comparison with microscopy and fungal culture. Fourth, UACR dichotomization—necessary to resolve complete separation—may reduce predictive precision for severe albuminuria. Finally, data availability limited the inclusion of other potential predictors, and multiplex qPCR protocols require further standardization. Future prospective, multicenter studies are needed to validate these findings across broader populations.

Conclusion

In summary, in this study we developed an easy-to-use and internally validated nomogram containing 8 baseline variables for SFI in T2DM patients. This may be a useful complement to the current identification of SFI, The risk model provides valuable insights into the identification of SFI risk and prevention.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

SFI

Superficial fungal infections

T2DM

Type 2 diabetes mellitus

ROC

Receiver operating characteristic

DCA

Decision curves analysis

CIC

Clinical impact curve

qPCR

Quantitative polymerase chain reaction

BMI

Body mass index

Glu

Blood glucose

HbA1c

Hemoglobin A1c

UACR

Urinary albumin-to-creatinine ratio

K⁺

Serum potassium

Na⁺

Sodium

HLP

Hyperlipidemia

HTN

Hypertension

DFU

Diabetic foot ulcers

CAD

Coronary atherosclerotic heart disease

DPVD

Diabetic peripheral vascular disease

DPN

Diabetic peripheral neuropathy

DKD

Diabetic kidney disease

DR

Diabetic retinopathy

DKA

Diabetic ketoacidosis

VD

Vitamin D deficiency

Fig.

Figure

VIF

Variance inflation factor

Author contributions

Ying Yang: conceptualization, supervision, methodology, writing–review & editing. Nan Chen: writing – review & editing, resources, funding acquisition. Yu Li: Writing–original draft, Writing–review & editing, Formal analysis. Guozhong Zhou: conceptualization, supervision, methodology, resources. Feifei Yang:​ investigation, data curation, resources. Rong long: Resources, investigation, funding acquisition. Wei Shi: Resources, investigation. Yan Dong: resources, investigation. Yuanyuan Zhou: resources, investigation.

Funding

This work was supported by Yunnan Provincial Science and Technology Plan Project (202301BE070001-039), Kunming University of Science and Technology (KUST) Medical Joint Research Program-Young Scholars Project(KUST-AN2023010Q).

Data availability

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request but are not publicly available due to privacy restrictions.

Declarations

Ethics approval and consent to participate

The study protocol complied with the Declaration of Helsinki and was approved by the Institutional Review Board of The Affiliated Anning First People’s Hospital of Kunming University of Science and Technology (Ethics Approval No.2023-045-SCI-02). Written informed consent was obtained from all participants prior to their enrollment.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.