Scientists have been trying to develop a vaccine ever since COVID-19 became a pandemic. But, now that it is here one of the biggest challenges is the logistics. The solution our team came up with employs machine learning to make the best prediction for prioritization and also eliminates natural human bias. We have also made a website that uses the trained model to act as the frontend for our project. We use publicly available individual-level data of all age groups including people with various pre-existing diseases to train our model to get the best possible performance and to generalize it for a larger population. On our website, users can enter their details such as sex, age, weight, height, smoking history, disease history, COVID history. We then send a request to the backend i.e., fast API with these details. The trained model then predicts a risk factor using the given data. The model was trained using a Random Forest Regressor. The predicted risk factor is sent back to the website and displayed as one of 4 categories(Low risk, Medium Risk, High Risk, Very High Risk) to the user. The vaccine distribution can be done in accordance with the risk factor predicted thereby streamlining the process.
Problems we faced while building our project Finding a medical dataset that fit with our problem statement and pre-processing it for use was very time-consuming. Setting up the backend using fast API to interact with the website took us some time to learn and implement. Collaborating with teammates overall was a challenge. But we are proud to have completed the project successfully.
Future Prospects of our project: One of the first improvements would be to make this a mobile app for easy access. The model can be improved with more individual-level data from local health sources. The model can be made more relevant to India by collecting data from Indian health clinics. The app could also be linked with government documents such as Aadhaar to validate user’s details.
Log in or sign up for Devpost to join the conversation.