Inspiration

Lack of centralized healthcare services and models being used in India

What it does

Easy diagnosis, prediction and monitoring of healthcare of all the patients and doctors on the portal

How we built it

A blockchain and machine learning-powered centralized electronic healthcare records management system, with key features such as Post-Recovery Comorbidities Prediction using Knowledge Graphs, Drug Recommendation system with the least side effects using Logistic Regression, Medical Named Entity Recognition using Microsoft Azure Machine Learning Services, and Blockchain Medical Certificate Issuing system hosted on Ethereum Ropsten Network.

Challenges we ran into

Getting Approval for Real Patient Records

Accomplishments that we're proud of

Real patient records such as MIMIC – III and Synthea were used for training Machine Learning Models. The average Spearman’s Rank Correlation of the proposed comorbidity prediction algorithm was between 0.5 and 1, Drug Recommendation model had a precision of 0.93, F1 of 0.96, AUC score of 0.903, and accuracy of 91.33%.

What we learned

Learnt new tools such as Neo4j and Microsoft Azure services

What's next for Panacea

Make the project open-source and let people integrate Panacea as a service

Share this project:

Updates