The project aims to forecast an episode of a heart disease and attack based on different clinical parameters. This forecast can supplement the doctors’ evaluation. In the past few months, I have collaborated with many cardiologists and hospitals in my region to collect historic medical data of patients with and without heart disease in every age bracket. This helped me achieve realistic results rather than using dummy data to train the model. The dataset consists of fourteen different attributes like age, sex, fasting blood sugar levels, exercise induced angina, resting blood pressure, chest pain type, and resting electrocardiographic results just to name a few. I investigated how methods like convolution neural network based multimodal disease risk prediction algorithm and support vector machine can be applied to achieve a prediction accuracy of 91% for a congestive heart failure.

Wrote scripts to extract and index over 20,000+ questions written using AiML. Used a web interface to build an AI bot that interprets the input questions by splitting the content in keywords. Used NLP techniques on text input to find semantics and search services via context matching.