. 2022 Nov 23;84:104956. doi: 10.1016/j.amsu.2022.104956

Table 2.

Summaries of machine learning algorithms development.

Study ID	Data		ML Technique	Model Type	Model Threshold	Model Training ^(REF)^a			Model Validation ^(REF)
Study ID	Type	Reference Standard	ML Technique	Model Type	Model Threshold	Sample Size	No. of SSI	SSI Incidence Rate (%)	Sample Size	No. of SSI	SSI Incidence Rate (%)
^b Atti/2020 [43]	S-EMR, FT	Hospital surveillance program	RE	D	NR	T (2,944)	T [18]	T (0.61)	NR	NR	NA
^b Azimi/2020 [46]	S-EMR	NR	BN, DT, SVM, ANNs, RF	P	NR	T (208)	T [18]	T (8.65)	NR	NR	NA
^b Bucher/2020 [10]	S-EMR	NR	NER	D	NR	4574	255	5.57	17,210	793	4.61
^b Chen/2020 [39]	S-EMR, FT	National surveillance data (NNIS)	LR, RF, DT, ANNs	P	NR	17,597	202	1.15	4014	43	1.07
^b Hopkins/2020 [44]	NR	Chart review	ANNs	P	NR	3034	T [60]	T (1.48)	1012	NR	NA
^b Karhade/2020 [38]	FT	Chart review	BC	D	0.05, 0.1, 0.5	4483	46	1.03	1377	16	1.16
Merath/2020 [29]	S-EMR	ACS-NSQIP	DT	P	NR	15,657	NR	NA	NR	NR	NA
^b Petrosyan/2020 [47]	ADMIN	ACS-NSQIP	RF, LR	P	NR	10,046	556	5.53	4305	239	5.55
^b Skube/2020 [50]	S-EMR	ACS-NSQIP	LR	D	0.04, 0.06	6188	398	6.43	5132	161	3.14
^b Song/2020 [49]	ADMIN,	National surveillance data (NIC-HAI)	LR, DT, SVM	P	NR	7419	T (205)	T (2.21)	1855	NR	NA
^b Song/2020 [49]	S-EMR	National surveillance data (NIC-HAI)	LR, DT, SVM	P	NR	7419	T (205)	T (2.21)	1855	NR	NA
Gowd/2019 [23]	S-EMR	NR	LR	P	NR	13,697	NR	NA	3422	NR	NA
^b Quérouéa/2019 [48]	S-EMR, FT	Hospital surveillance program	LR	D	NR	T (2,133)	T [22]	T (1.03)	NR	NR	NA
Shen/2019 [32]	FT	Chart review	DT, SVM, RF	D	NR	T (1,178)	T (80)	T (6.79)	NR	NR	NA
^b Shi/2019 [40]	S-EMR, FT	Chart review	RF, SVM, LR	P	NR	T (5,795)	T (291)	T (5.02)	NR	NR	NA
^b Silva/2019 [7]	S-EMR, FT	Hospital surveillance program	RF, LR, SVM, BN, NC, SGD	P/D	NR	15,479	188	1.21	12,637	202	1.60
Thirukumaran/2019 [35]	ADMIN,	Hospital surveillance program	LR	P	NR	1263	172	13.62	316	36	11.39
Thirukumaran/2019 [35]	S-EMR, FT	Hospital surveillance program	LR	P	NR	1263	172	13.62	316	36	11.39
^b Tunthanathip/2019 [41]	S-EMR	NR	DT, BN, ANNs, KNN	P	NR	T (1,471)	T (67)	T (4.55)	295	NR	NA
Grundmeier/2018 [24]	ADMIN,	Chart review	RF, LR	P	NR	6871	209	3.04	1039	25	2.41
Grundmeier/2018 [24]	S-EMR, FT	Chart review	RF, LR	P	NR	6871	209	3.04	1039	25	2.41
^b Kocbek/2018 [42]	S-EMR, FT	NR	LR, BC	P	Range: 0.171-0.245	909	183	20.13	228	50	21.93
Strauman/2018 [34]	S-EMR	ICD-10 and Procedure codes	ANNs	D	NR	T (883)	T (232)	T (26.27)	NR	232	NA
Weller/2018 [36]	S-EMR, FT	NR	LR, RF, SVM, BN, BC	P	NR	1051	102	9.71	232	18	7.76
Chapman/2017 [37]	S-EMR, FT	Chart review	SVM	D	NR	565	NR	NA	100	NR	NA
Sohn/2017 [8]	S-EMR, FT	Chart review	BN	D	NR	T (751)	T (67)	T (8.92)	NR	NR	NA
Hu/2016 [26]	S-EMR	ACS-NSQIP	LR	D	NR	5280	336	6.36	3629	157	4.33
Ke/2016 [27]	S-EMR, FT	NR	Linear regression, SVM	P	NR	652	T (167)	T (20.49)	163	NR	NA
Mandagani/2016 [28]	S-EMR	NR	LR, DT	P	NR	T (879)	T (181)	T (20.59)	NR	NR	NA
Sanger/2016 [31]	S-EMR	NR	BN	D	NR	T (851)	T (167)	T (19.62)	NR	229	NA
Hu/2015 [25]	S-EMR	ACS-NSQIP	LR	D	NR	3996	278	6.96	2262	127	5.61
Soguero-Ruiz/2015 [33]	S-EMR	ICD-10 and Procedure codes	SVM	P	NR	T [1,005]	T (101)	T (10.05)	NR	NR	NA
Esbroeck/2014 [22]	S-EMR, FT	ACS-NSQIP	LR	P	NR	602,089	NR	NA	350,545	NR	NA
Michelson/2014 [30]	S-EMR, FT	Hospital surveillance program	LR	P	NR	T (2,407)	T [59]	T (2.45)	NR	NR	NA
^b Campillo-Gimenez/2013 [45]	S-EMR, FT	Chart review	VSM	D	NR	3785	42	1.11	1225	NR	NA

Abbreviation: IQR, interquartile range; NA, not available; NR, not reported; SSI, surgical site infection.

Data type: FT, Free text data; ADMIN, Administrative data; S-EMR, Structured Electronic Medical Records. Reference standard, ACS-NSQIP, American College of Surgeons-National Surgical Quality Improvement Program; NIC-HAI, Nursing Intensity of Patient Care Needs and Rates of Healthcare-Associated Infections; NNIS, National Nosocomial Infections Surveillance. ML type: ANNs, Artificial Neural Networks and its variations; BC; Boosted Classifiers (e.g., AdaBoost, XGBoost); BN, Bayesian Network; DT, Decision Tree; KNN, k-nearest neighbors; LR, Logistic Regression and its variations; NC, Nearest Centroid; NER, Named Entity Recognizer; SGD, Stochastic Gradient Descent; SVM, Support Vector Classification; RE, Regular Expression; RF, Random Forest; VSM, Vector Space Model. Model type: P, predictive; D, detective.

^(REF): Data extracted from reference standard (e.g., chart review).

T (number indicates total number of SSI/procedures (when training/testing sample size were not specified in article).

Articles included for Meta-analysis.