The Gen AI Hackathon project aims to utilize machine learning for healthcare by developing a web application that can detect various diseases such as asthma, cancer, diabetes, and stroke. The application provides real-time diagnosis based on predictive modeling. If a disease is detected, the user receives preventive measures and can engage in further conversation with a llm chatbot specialized in that disease.
- Real-time disease detection using machine learning
- Personalized preventive measures for detected diseases
- Interactive llm chatbot for further consultation
- User authentication with sign-up and sign-in functionality
- Easy-to-use interface with intuitive navigation
- User signs up or signs in to the platform.
- On the home page, the user can navigate to the diagnosis page.
- In the diagnosis page, the user fills out a form providing necessary information.
- Based on the input, the application predicts whether the user has any of the specified diseases.
- If a disease is detected, preventive measures are provided to the user.
- The user can engage in further conversation with a disease-specific llm chatbot on the chat page.
We have worked on both aspects of the problem statement, i.e, Disease Diagnosis , treatment Recommendation.
The disease diagnosis part consists of the identification of disease based on information received from user through form-filling. Then ML model specific to that disease is used for the diagnosis of the particular disease.
The recommendation process plans out a detailed report regarding causes, symptoms, medical prescription and changes to be made in living style.
There is a LLM based chat-doctor finetuned on dataset (link mentioned at last), which can given guidance on general medical related questions.
-
Clone the repository:
https://github.com/shivam6862/Gen-AI-Hackathon.git cd Gen-AI-Hackathon -
Install dependencies and Run the application:
cd frontend npm install npm run dev -
Install dependencies and Run the application:
cd backend python -m venv env [for windows] pip install -r requirements.txt cd genaimechbackend python manage.py runserver
-
Give .env in
backend/genaimechbackend/genaimechbackend/.envHOST= PROJECT_NAME= DB_USERNAME= PASSWORD= SECRET_KEY=
-
Access the application at port
-
The finetuning is done on Intel/Mistral-7B-v0.1-int4-inc, the code for fine-tuning is written under medical_finetune.py. The file can be run using below command
python medical_finetune.py --bf16 True --use_ipex True --max_seq_length 512- Dataset
- Model for finetuning
- scikit-learn-intelex
Team name - genaimech