Can a machine learning model be used to successfully identify risk factors for stroke recurrence?

Stroke patient information was extracted from electronic health records and statistically analyzed to build a machine learning model for the identification of risk factors for stroke recurrence.
The Shapley Additive exPlanations (SHAP) method was used to determine whether the machine learning model could be easily interpreted.
The top 10 risk factors identified by the SHAP method (severe carotid artery stenosis, weak, homocysteine, glycosylated hemoglobin, sex, lymphocyte percentage, neutrophilic granulocyte percentage, urine glucose, fresh cerebral infarction, and red blood cell count) were used in the model which produced results that were deemed not significantly different from that of a model with 20 risk factors.