Project Implementation

[ethanchristoff]

Project Implementation

Overview

This document outlines the steps and approach for implementing the project selected in Week 2. The project aims to develop a Generative AI-based chatbot for medical prescription issuing. The chatbot will utilize a large language model (LLM) trained on extensive datasets of documented symptoms, diseases, and treatments, allowing users to interact with the AI to receive advice or potential prescription suggestions for common conditions.

Step-by-Step Plan

1. Data Collection (Week 3-4)

• Objective: Gather and curate data related to major diseases and their corresponding symptoms, treatments, and medications. Focus will be on conditions such as Diabetes, Hypertension, Asthma, and Chronic Kidney Disease.
• Tasks:
  • Research medical databases and gather clinical data for the mentioned major disease.
  • Structure the data into categories: symptoms, treatments, medications, lifestyle advice.
  • Ensure the data covers common cases as well as edge cases for more comprehensive training.

2. Model Training (Week 5-6)

• Objective: Fine-tune a large language model (LLM) to specialize in handling medical data related to the selected diseases.
• Tasks:
  • Preprocess the gathered data and format it for LLM training.
  • Fine-tune the LLM on the curated dataset, emphasizing medical terminology and patient interaction.
  • Validate the chatbot’s responses using test cases and adjust the model parameters for accuracy.

3. User Interface Development (Week 7-8)

• Objective: Build an intuitive interface that users can interact with to consult the AI-based chatbot.
• Tasks:
  • Develop a web or mobile interface for user interaction.
  • Implement features such as symptom input, prescription suggestions, and condition explanations.
  • Ensure the design is user-friendly, accessible, and adheres to privacy/security standards.

4. Backend Implementation (Week 7-9)

• Objective: Set up the backend architecture to handle chatbot interactions, store user data securely, and maintain smooth system operations.
• Tasks:
  • Set up a server to run the trained chatbot model.
  • Alternatively, if we may utilize Discord's JDA to create a discord based chatbot
  • Besides that we can use a framework with django to make a simple web based chatbot..
  • Implement secure data handling for user inputs and chatbot responses.
  • Create a database for logging interactions and storing necessary information (while maintaining user anonymity).

5. Testing and Validation (Week 10-11)

• Objective: Ensure the chatbot provides accurate and reliable advice by running extensive tests.
• Tasks:
  • Use a variety of user queries to test the chatbot’s effectiveness in providing appropriate medical advice.
  • Validate the chatbot’s performance using medical professionals for feedback.
  • Optimize performance based on test results, adjusting model outputs and refining the dataset.

6. Launch and Evaluation (Week 12)

• Objective: Launch the chatbot and evaluate its performance in real-world scenarios.
• Tasks:
  • Deploy the chatbot for public use or within a controlled environment for beta testing.
  • Collect user feedback and monitor performance metrics.
  • Continue to update and improve the chatbot based on user experiences.

Conclusion

This project will be implemented in a structured, phased approach, starting with data collection and leading to model training, interface development, and testing. The goal is to deliver a reliable, AI-based medical chatbot that can assist users in symptom analysis and potential prescription suggestions for common medical conditions.

[ethanchristoff]

Issue Submissions

[!important] If there are any issues you need to submit related to a certain issue mentioned in the roadmap, please continue it as a discussion under the threads here. However, if there are any sub-issues under the main one mentioned here, it is to be made under that issue instead.

Sub-Issues:

For any sub-issues or specific concerns that fall under the main issue, please create a new submission within the relevant issue thread. This helps maintain clarity and allows for better tracking of the problem's scope.

Why is this important?

Keeping related discussions in the same thread enhances collaboration and reduces redundancy, allowing the development team to respond more efficiently.