Information and Analytical Support of Telemedicine Services for Predicting the Risk of Cardiovascular Diseases

Abstract

The article is devoted to the identification and study of predictors for the personalized multivariate predictive models of the cardiovascular diseases’ risk based on the patient's digital footprint in the context of information and analytical support of telemedicine services. To build predictors - features of machine learning models, the methods for depersonalizing and extracting data from electronic medical records were developed. The paradigm “5P Medicine” -prevention, prediction, personalization, participation, practicality, formed the basis for the comparative analysis of models and obtaining the estimates of the degree of the cardiovascular diseases’ risk for personalized prediction. The created service prototype, using data from medical information systems, generates lists of problem patients who need an in-depth preventive examination. The developed prototype of a telemedicine information system ensures safe collection and analysis of medical data received, among other things, from mHealth devices. This makes it possible to determine additional predictors for assessing the patient's condition according to the information system of the home hospital. The original technology implemented in the system is based on attributive encryption to protect both the transmission and storage of personal health information in the cloud.