A project I worked on that took different symptoms and provided a diagnosis. The model currently has 92% accuracy and is deployed to Heroku!
Engineering an AI-based algorithm for reliable medical diagnosis given a set of symptoms for those without access to doctors.
Trained an MLP on ~4500 records of symptoms and diseases, with a validation accuracy of 92%.
Visualized and performed a prediction distribution analysis using Matplotlib and Seaborn.
Spearheaded user ease-of-use by building an interactive GUI + deploying the App on Heroku!
Hosted a workshop detailing this project to a group of 20+ professionals and those attempting to learn more about the field.
For more information, check out this article I wrote detailing EXACTLY how this works - and, a presentation I hosted for this very project:
⚙️ Workshop: https://www.youtube.com/watch?v=qwliferwZNk 📝 Article: https://aditya-dewan124.medium.com/making-an-ai-that-can-diagnose-your-symptoms-483468833fcd