A Machine Learning Tool to Predict Survival After First Surgery in Peripheral Artery Disease Patients.

The aim of this study was to develop and validate a machine learning tool for predicting survival in PAD patients who received surgical treatment. We used the data from 1,615 patients who underwent PAD surgery from 2005 to 2020. Gradient boosted decision trees (GBDTs) were used to predict mortality at one, three and five years after the first surgery, while predictor importance was assessed using the SHAP values method. The area under the curve (AUC) of the receiver operating characteristic curve of the one-, three and five-year prediction models were 0.86, 0.84 and 0.80, respectively. Disease stage was the most important predictor, along with age, chronic kidney disease status, hospital length-of-stay and total number of comorbidities. Presence of dyslipidemia was slightly predictive of one- and three-year mortality. Simple clinical and demographic parameters can be used to train a GBDT model capable of predicting PAD follow-up mortality.