Table 2. Characteristics of included studies (author, year, country, design, population, sample size, data source, ML algorithms, revascularization type, and outcome definition).

ML, machine learning; MALE, major adverse limb events; ACS, American College of Surgeons; NSQIP, National Surgical Quality Improvement Program; VQI, Vascular Quality Initiative; SVM, support vector machine; MLP, multilayer perceptron; ANN, artificial neural network; PAD, peripheral arterial disease; TEG, thromboelastography.

Author (year)	Country	Design	Population	N	Data Source	ML algorithms	Revascularization type	Outcome definition
Li et al. (2024) [1]	USA	Retrospective cohort (database)	Patients undergoing lower-extremity endovascular revascularization (angioplasty/stent/atherectomy) for PAD	21,886	ACS NSQIP Targeted Vascular Database (2011-2021)	XGBoost; random forest; Naïve Bayes; SVM; MLP ANN; logistic regression (assumed similar to above)	Endovascular (angioplasty, stent, atherectomy)	30-day post-procedure MALE or death (MALE = major reintervention, untreated loss of patency, or major amputation)
Perkins et al. (2020) [2]	USA/UK	Retrospective development + external validation	Patients with lower-extremity arterial trauma undergoing attempted revascularization	508 (US) + 51 (UK)	US Joint Trauma System data (development); UK Joint Theatre Trauma Registry (validation)	Bayesian network (10-predictor model)	Open limb revascularization (trauma)	Failed revascularization (limb loss requiring amputation)
Ghandour et al. (2025) [3]	USA	Prospective cohort	PAD patients undergoing lower-extremity revascularization (single-/multicenter)	308	Prospective clinical registry (2020-2024, including viscoelastic TEG data)	Logistic regression; XGBoost; decision tree	Lower-extremity revascularization (bypass or stent)	1-year graft/stent thrombosis (arterial thrombotic event) following revascularization
Li et al. (2024)[4]	USA	Retrospective cohort (database)	Patients undergoing lower-extremity open bypass for chronic PAD	24,309	ACS NSQIP Targeted Vascular Database (2011-2021)	XGBoost; random forest; Naïve Bayes; SVM (RBF kernel); MLP ANN; logistic regression	Open surgical bypass (chronic PAD)	30-day MALE or death (MALE = untreated graft occlusion, reintervention, or major amputation)
Li et al. (2024) [5]	USA	Retrospective prognostic study	PAD patients undergoing endovascular intervention (VQI Registry) .	235,677	Vascular Quality Initiative Registry (2004-2023)	XGBoost; random forest; Naïve Bayes; SVM; MLP ANN; logistic regression	Endovascular (all PAD interventions)	1-year MALE or death (MALE = thrombectomy/thrombolysis, surgical reintervention, or major amputation)