A Data Driven Machine Learning Approach for Predictive Analytics in Healthcare Domain

Abstract

Machine Learning and Deep Learning are transforming the healthcare domain so as to aid medical practitioners with effective analytical tools which can find pattern in copious amounts of data and augment the diagnosis and potential health risks. While the use of statistical techniques may be shrouded with skepticism for the healthcare sector, its inevitability came into light recently with the Covid-19 pandemic wherein the number of cases saw an exponential growth stressing the healthcare infrastructure. The sudden outbreak of such a global medical emergency stressed upon the need to develop intelligent computational models which can analyze medical datasets and render results pertaining to the onset or existence of a disease, potential cases and hotspots in the future. Such systems would offload the existing medical infrastructure and also give medical practitioners valuable inputs as a strong second opinion. Such systems would also render primary screening in remote location where necessary medical facilities are unavailable. Additionally, controlling and arresting potential pandemic like situations, by identifying potential hotspots or estimating future number of cases would also allow government agencies. This paper presets a data driven approach for classifying medical images, as well as a regression model for forecasting future cases. A comparative analysis with benchmark models in the domain of research clearly indicates that the proposed model outperform exiting research in terms of classification and forecasting accuracy