Isolation forest-voting fusion-multioutput: A stroke risk classification method based on the multidimensional output of abnormal sample detection

Hai He, Haibo Yang, Francesco Mercaldo, Antonella Santone, Pan Huang

Abstract: Highlights•This article is the first to explore the impact of abnormal samples of stroke screening data on the prediction of the stroke risk grade and stroke occurrence.•For the first time, an isolation forest-voting fusion-multi output (IF-VF-MO) predictive classification model, which can comprehensively and accurately predict the various stroke risk levels and stroke occurrence, also provide multidimensional auxiliary diagnostic information to medical staff, is built.•The characteristic composite score index is used to analyze the importance of different risk factors in the screening data for identifying all stroke risk levels.