Predictive Telemedicine Model for Early Heart Attack Detection in Dialysis Patients Using Multimodal Data and Machine Learning

Authors- Assistant Professor Mr J.P Pramod, Boddupally Pravalika, Gyajangi Nandini

Abstract-The dialysis patients are prone to developing heart attacks, as fluctuations in blood pressure and electrolytes occur very often. Such an early diagnosis is extremely important to ensure timely interventions. The current work presents a predictive telemedicine model for continuous monitoring of dialysis patients, thus providing them with early warning for a heart attack. Patients on dialysis, more often are liable for heart attacks due to frequent fluctuation of blood pressure, electrolyte balance, and intrinsic medical illness. Early detection of this event is the prerequisite treatment for patients. In light of this need, there is a proposed predictive model aimed at predicting heart attacks or chances of a heart attack in patients who have come for dialysis; remote monitoring and early intervention over telemedicine channels will surely be beneficial. In applying these various machine learning algorithms, logistic regression, random forest, SVM, KNN, Gradient Boosting, and finally XGBoost classifiers have been used. Using relevant clinical features like vital signs, lab findings, and other demographic data in a given database, the models will be trained to identify certain patterns that are correlated to the risk of having an attack. XGBoost had the highest score at 92.6% with excellent precision, recall, and F1-score, which shows that the model was pretty robust in predicting the events.

DOI: /10.61463/ijset.vol.13.issue2.208