Health - AI

Health - AI uses probabilistic models to understand the variation of health metrics and informs patients about their health. The application uses dynamic visualizations and probabilistic multinomial regression to model historical datasets and make predictive analysis and then identify anomalies. It also makes intelligent suggestions to the patients and notifies users of abnormalities based on historical data and predictive values.

The application is now live at .

You may use the following details to log in: (You can also sign up on the application because the login system is functional)

Email: Password: letsdothis


The application is built using Flask/Python and was deployed using the Oracle Cloud. The data pre-processing and analysis was carried out using Python(sci-kit learn), R(Caret). The front-end has been built using HTML, CSS, and JS.

The Motivation

There is a lack of affordable and accessible healthcare providers to give constant care and attention to all patients in need of intensive care and regular monitoring. Our application is a basic model that notifies and updates users to changes in their health. This helps make the diagnosis of these patients much easier for medical professionals since the AI constantly monitors the vitals of the user. The user/patient will also be more aware of their health and activities. Being a Clinical Psychologist's son, Sohit and our team is fully aware of the inefficiencies in the healthcare system, the burden of demand on our clinical professionals, and the need of people around the world for health monitoring. We empathize with these people and understand that there is a lot to be done to make the healthcare industry more efficient.


We faced many challenges in the process of creating this software. The most difficult challenge was building a model that considered all real-world factors that affect the health of patients. There are an enormous amount of factors to consider when it comes to building a personalized health model. The amount of factors is too great to consider properly in the scope of a hackathon, so we worked on a small scale model considering only three basic metrics: heart-rate, calorie intake, and step count. The dataset was sourced from Kaggle which comprised of data collected from a smartwatch worn by an individual over a period of time. We decided to utilize a group of models to predict our metrics. We forecasted the future heart-rate of the user based on the multinomial regression approach. Using the predictive models, we notified users of health abnormalities based on certain thresholds, such as a massive spike in heart rate.

Another issue was stack integration and making sure that our notifications and visualizations of data were dynamic, which we accomplished. We needed to make sure that our python code had access to our processed data obtained from R scripts.

The Team

Our team accomplished this project because of the diverse and unique skills of each of the team members. Sayane Shome is a graduate student pursuing a Ph.D. in bioinformatics and is experienced in biological data and machine learning. Sohit Miglani is a data scientist with experience in web development and backend integration. Yusuf Quddus is a computer science student with experience in web development and front end design. Adrian is pursuing a Ph.D. in computational neuropsychology with expertise in biomedical data. Our team worked together and combined our unique knowledge to create an excellent team capable of producing a base model for AI-powered medical assistants.


Step 1. Clone the repository using the command:

git clone

Step 2. CD into the cloned repository and run the command:

cd health_ai

The application is now functioning on your local computer server. Go to a browser and log into 'localhost:5000' to start working with your application.

Share this project: