eagrawal519-del/Build-a-Solution-for-a-Sustainable-Development-Goal-SDG-
Domain:
environment and energyCreator:
eag
Host:
AI-powered satellite image analysis for deforestation detection and forest monitoring in Africa.
# MediCare AI
An AI-assisted medication adherence tracker built for **SDG 3 — Good Health and Well-being**.
Patients (or caregivers) add their medications, log each dose as taken or
missed, and an AI assistant (powered by the Claude API) turns that adherence
history into a plain-language insight, plus answers free-text questions
about medications and adherence.
Stack: **HTML5 + CSS + vanilla JavaScript** (frontend) · **Python / Flask +
SQLite** (backend) · **Claude API** (AI layer).
---
## 1. Project structure
```
medicare_flask/
├── app.py # Flask app: routes, DB, AI calls
├── requirements.txt
├── .env.example # copy to .env and add your API key
├── data/
│ └── reference_medications.json # reference dataset (see "Data sources")
├── templates/
│ └── index.html
└── static/
├── css/style.css
└── js/app.js
medicare.db # created automatically on first run
```
## 2. Data sources
This app uses two kinds of data:
1. **User-generated data (primary).** Every medication and dose log you
enter is written to a local SQLite database, `medicare.db`, created
automatically the first time you run the app. This is the app's real
dataset — the adherence history the AI insight is generated from.
2. **Reference data (secondary).** `data/reference_medications.json` is a
small, hand-curated sample of common generic medication names, typical
strengths, and frequencies used only to power autocomplete suggestions
in the "Add medication" form. It's modeled on the kind of fields found
in public drug-reference sources such as the
openFDA Drug Label dataset and
RxNorm (U.S. National
Library of Medicine), but is bundled as a static file so the project
runs fully offline for grading/demo purposes. `GET /api/reference-medications`
is the single place you'd swap in a live call to one of those APIs.
3. **AI-generated content.** Adherence insights and chat replies are
generated at request time by the Claude API from you …