Symptom Tracker

Project Story: SymptomTracker

Inspiration

We were inspired by real-world challenges in early disease detection and self-monitoring. Many people delay checkups until symptoms worsen. This project aims to bridge that gap with an AI-assisted tool.

What it does

SymptomTracker captures vitals and symptoms, applies a transparent rule-based engine, and flags risks for:

Type 2 Diabetes
Hypertension
Depression/Mood
Migraine
Sleep Apnea
Anemia

It also integrates a Hugging Face zero-shot classifier for optional enrichment.
Data is stored locally in SQLite, with support for trend visualization and CSV export for clinician review.

⚠️ Educational demo only — not medical advice.

How we built it

We used:

Python (Streamlit) for front-end UI
SQLite for local data storage
Pandas & Matplotlib for visualization
Hugging Face Transformers for AI classification
GitHub for Version Control

Challenges we ran into

Balancing explainability vs. AI complexity
Designing rules that were clinically relevant yet simple
Handling data visualization in a clear way
Keeping the app lightweight for local execution
Setting the UI in a nice format

Accomplishments that we're proud of

Built a working prototype in limited time
Achieved explainable risk detection with hybrid AI + rules
Created an intuitive UI that non-technical users can navigate

What we learned

We learned:

How to combine symbolic AI with machine learning
The importance of UX in healthcare demos
That even simple math, like BMI calculation,
can have a big impact when visualized for the user