OS-Koala

This is team OS-Koala's project repository

There are some technology tools link

• Trello: https://trello.com/b/tBUPhXal/koala-kanban-board

• Confluence: https://unimelb-team-qjufxaqfyxcv.atlassian.net/wiki/spaces/COMP9008221/overview?homepageId=131406

• Monday: https://student795653.monday.com/boards/1854808094

• All the documentation will be stored in docs/, the source about sprint one are in docs/Sprint 1 Content

• All code file will be stored in src/

1.0 Project Overview

Title: Development of a Virtual Agent for Osteoarthritis Self-Management

Introduction

The Centre for Health Exercise and Sports Medicine (CHESM) at the University of Melbourne, dedicated to the prevention and management of musculoskeletal conditions, particularly osteoarthritis (OA), seeks to enhance its digital patient resources. Following the success of our “Taking Control of Your Hip and Knee Osteoarthritis” eLearning course on the FutureLearn platform, our aim is to create a virtual agent. This agent will deliver on-demand, reliable, relevant, and personalized information for people with OA, using a curated dataset of open access resources.

Background

Amidst widespread misinformation about OA online, there's an urgent need for a reliable source that offers accurate information. Existing AI models often source their information from non-reputable sources, leading to inaccuracies and potential harm. Our project intends to address this gap by providing a virtual agent capable of offering trustworthy advice based on the latest scientific understanding of OA.

Project Goals

• Develop a virtual agent that uses a friendly, empathetic, and professional tone to answer patient questions about OA.
• Ensure the agent draws information exclusively from pre-selected, reputable sources.
• Enable easy future updates to the content database by CHESM staff.
• Address privacy and security considerations by minimizing personal data storage and adhering to data protection regulations.
• Implement fast response times, multilingual support, accessibility features, and the ability to collect user feedback for continuous improvement.

Essential Features and Requirements

• Dataset Management: Utilize a preset dataset with the capability for CHESM to add new sources as necessary.
• Tone and Language: Maintain a human-like interaction style without causing unnecessary concern or anxiety.
• Medical Disclaimer: Avoid offering specific medical advice, instead encouraging users to consult healthcare professionals.
• Privacy and Security: Limit personal data storage and clearly communicate data handling practices.
• Performance: Ensure minimal lag for quick user interactions.

Desired Features

• Multilingual Support: To cater to users speaking different languages.
• Accessibility: Including text-to-speech and screen reader compatibility.
• Feedback Mechanism: Allowing user feedback to inform ongoing refinement of the service.

Usability Testing

The project includes a phase of usability testing through online surveys with academics and consumers within our networks, designed to refine the virtual agent based on user experience.

Final Product Vision

Our goal is to produce a virtual agent that embodies the principles of friendly and empathetic interaction, based on solid, scientific knowledge and capable of being integrated across multiple digital platforms. It will offer a backend that is straightforward for CHESM staff to manage, ensuring the agent remains an up-to-date and valuable resource for individuals managing OA.

Data Sources

The virtual agent will utilize information from a range of reputable sources, including the NICE guidelines, RACGP OA guidelines, OARSI Guidelines, EULAR guidelines, and various patient resources and videos, ensuring the information provided is accurate and current.

This comprehensive approach aims not only to enhance the support for individuals with OA but also to set a new standard for digital health resources in the field.

2.0 Project Development

We divided the develop process into two parts UI and develop

In sprint 2, we finished the parts of UI and make the project connected with backend server.

We achieve all the functions on the branch UI in sprint 2.

All the documentation of sprint 2 are stored in docs/Sprint 2 Content, if you want to visit more, please check our Confluence page.

In sprint 3, we will finish the parts of backend database connection and project deployment.

We will achieve all functions of database connection in branch of develop.

In the end of the project, we will merge the branch UI and develop into the main branch.

3.0 Code Review

For currently development of the project, we will use https://github.com/agogear/chatgpt-pr-review as our code review method.

Considering sometimes it does not work, we decide to also use Jekyll site which is a baseline github action as our backup code review method.