HAKSAL26/AI-FOR-MEDICAL-IMAGING-AND-PRECISION-DIAGNOSTICS-FROM-THEORY-TO-PRACTICE
Domain:
healthcareCreator:
HAK
Host:
This study examines the role of artificial intelligence in improving precise and accurate diagnosis is low-resource settings where delayed or inaccurate diagnosis continue to increase morbidity and mortality rate
# AI-Powered Diagnostics Platform
React frontend presenting the abstract, methodology, and expected outcomes for an AI-powered diagnostics initiative in low-resource healthcare settings.
Python backend provides dataset-based model training and prediction APIs.
## Scripts
- `npm install`
- `npm start`
- `npm run build`
## Backend (Node.js)
### Setup
- Run `npm install` to install all dependencies (frontend + backend).
- Start the API server with `npm run server`.
### API Endpoints (Node.js)
- `GET /health` – health check
- `POST /train?target= ` – upload CSV and train a model
- `POST /predict` – send JSON records to get predictions
### Dataset Requirements
- CSV format.
- `target` query parameter must match a column in the CSV.
- Other columns are treated as features (numeric and categorical supported).
### Example Predict Body
```json
{
"records": [
{"age": 29, "symptom_score": 3.2, "region": "rural"},
{"age": 41, "symptom_score": 1.7, "region": "urban"}
]
}
```
### Example Chat Body
```json
{
"message": "I have a sore throat and fever for two days"
}
```
## Pages
- Overview: Abstract and impact highlights
- Methodology: Phased implementation approach
- Outcomes: Expected health and system benefits
- Assistant: Prototype chat experience for user interaction with an eventual trained model
## Local Model Training (Python)
This project includes a local text model for symptom → condition classification.
### 1) Download datasets
Download the Kaggle datasets and place the CSV files in a local folder such as `data/raw/`:
- kaggle.com
- kaggle.com
- kaggle.com
- kaggle.com
### 2) Train the model
Run the training script (uses scikit-learn locally):
- `python ml/train_symptom_model.py --da …