Stroke Prediction in Elderly Persons using Remote Health Monitoring.

Stroke is a condition that effects the arteries leading to and within the brain, causing paralysis or death. Early detection or prediction of stroke can save patient's life. For this purpose, the medical condition of suspected patient needs to be continuously monitored which is not manually possible. The research results presented in this paper aim prediction of stroke in the elderly people by continuously monitoring their condition through non-invasive sensor-based devices and using Machine Learning (ML) techniques. The medical attributes that may cause stroke are given as input to the proposed XGBoost-based model that predicts the probability of occurrence of stroke with an accuracy of 94%. This paper presents development of the proposed ML model and discussion of obtained results. As perspectives, the model will be deployed on fog nodes of SAFE-RH and patient readings collected through sensors will be given as input to the model for stroke prediction.