A Machine Learning Paradigms for Smart Health Monitoring System

Healthcare of patients in society today faces several problems. In society, patients are unaware about the extent and analysis of their medical condition and whether it is critical or not, mainly due to lack of the mortality prediction rate as the prediction is usually not done by the lab testing facilities or the medical facilities. With the help of 10 different LightGBM models, our AI-based health monitoring system effectively takes patient’s medical test data as input and thus calculates and analyzes the mortality prediction rate of every patient’s uploaded data. Hence, doctors can now treat patients based on a highest to lowest priority order and treat patients who have a critical condition first. This model achieved an AUC score of 0.912 which is better than the Acute Physiology and Chronic Health Evaluation (APACHE) IV model which achieved an AUC score of 0.88, and thus our model outperforms the APACHE IV model and gives patients an efficient system to predict their fatality prediction rate.