https://github.com/shinichishi/adrcheck
Medify is an AI-powered platform for early ADR detection, integrating FDA reports, patient feedback, and social media data.
https://github.com/shinichishi/adrcheck
adr analysis drug fda-api ml reddit-api
Last synced: 2 months ago
JSON representation
Medify is an AI-powered platform for early ADR detection, integrating FDA reports, patient feedback, and social media data.
- Host: GitHub
- URL: https://github.com/shinichishi/adrcheck
- Owner: ShinichiShi
- License: mit
- Created: 2025-03-29T14:04:34.000Z (7 months ago)
- Default Branch: main
- Last Pushed: 2025-03-30T07:45:00.000Z (6 months ago)
- Last Synced: 2025-06-24T00:44:21.998Z (4 months ago)
- Topics: adr, analysis, drug, fda-api, ml, reddit-api
- Language: TypeScript
- Homepage:
- Size: 450 KB
- Stars: 0
- Watchers: 1
- Forks: 2
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Medify
## Overview
Medify is an advanced AI-powered platform for the early and accurate detection of Adverse Drug Reactions (ADRs). It integrates multiple heterogeneous data sources, including FDA reports, unstructured patient feedback, and real-time social media discourse. Medify provides explainable AI insights to help users and healthcare professionals make informed decisions about drug safety.
## Features
- **Drug Search & Analysis**: Users can search for a drug and retrieve FDA reports, label composition, and known side effects.
- **Reddit-Based Sentiment & ADR Analysis**: Scrapes Reddit for anonymous yet legitimate user experiences regarding drug effects.
- **ADR Count & Statistics**: Aggregates patient reports to show the number of individuals affected by specific side effects.
- **Personal Health Record Analysis**: Users can upload past medical records to detect if a drug was responsible for previous adverse reactions.
- **Hospital Risk Analysis**: Healthcare institutions can assess patient susceptibility to diseases based on drug interaction data and historical trends.
## How It Works
1. **Data Aggregation**:
- FDA reports are fetched and analyzed.
- Social media (Reddit) data is scraped and classified using NLP models.
- User-provided medical records are processed for potential ADR detection.
2. **AI-Powered Causal Detection**:
- Uses Explainable AI (XAI) models to determine causal relationships between drugs and adverse reactions.
- Overcomes data sparsity and bias through robust ML techniques.
3. **User Interaction**:
- Simple search interface for retrieving drug-related information.
- Insights into how many users report specific side effects.
- Custom alerts for hospitals regarding at-risk patients.
## Tech Stack
- **Backend**: FastAPI / NextJS for API handling.
- **Database**: MongoDB.
- **AI & NLP**: Various models like logistic regression and NLP is used.
- **Frontend**: NextJS for visualization.
- **Scraping**: Reddit API for real-time ADR discussions.
## Future Enhancements
- Integrate real-time notifications for hospitals when a drug is flagged as high-risk.
- Expand social media analysis to platforms like Twitter and patient forums.
- Implement a patient-community portal for verified ADR reports.
## Contributing
We welcome open-source contributions! Feel free to submit issues, feature requests, or pull requests to help improve Medify.
## License
This project is licensed under the [MIT License](LICENSE).