Project

This project aims to develop and implement artificial intelligence (AI) techniques to automatically summarize electronic medical records (EMRs). The primary goal is to streamline the medical data processing by making it faster and more precise, thus supporting health professionals, including clinical coders, in making informed decisions quickly and accurately. The significance of this project lies in its potential to reduce the time and effort required for clinicians and coders to review extensive patient records, ultimately improving patient care and operational efficiency within healthcare institutions. The expected outcomes include the creation of an AI-based summarization tool that can condense vast amounts of medical data into concise and relevant summaries, the validation of this tool's effectiveness through rigorous testing, and the demonstration of its utility in real-world clinical settings. This project will pave the way for more advanced AI applications in healthcare, contributing to the ongoing digital transformation of the medical field.

Aims

The aims of this project are:

1. To develop AI Techniques for Summarization: Implement advanced artificial intelligence algorithms to automatically summarize electronic medical records (EMRs) effectively.
2. To validate and Demonstrate AI Tool Effectiveness: Rigorously test the AI-based summarization tool to validate its effectiveness and demonstrate its utility in real-world clinical settings, ensuring it meets the practical needs of healthcare professionals.

The problem is trying to solve: 

1. Improve Operational Efficiency in Healthcare Institutions: Reduce the time and effort required for clinicians and coders to review extensive patient records, ultimately improving operational efficiency within healthcare institutions.
2. Enhance Decision-Making for Health Professionals: Support clinical coders and other health professionals in making more informed and timely decisions by providing concise and relevant summaries of patient records.

Design

To address the task of summarizing electronic medical records (EMRs), we wil adapt large language models (LLMs) specifically for this purpose. This involves using pretrained LLMs on a vast dataset of EMRs to ensure they can accurately and concisely summarize medical data. We will employ a combination of techniques, including fine-tuning and prompt engineering.

Given that evaluating summarization quality is an open problem, we will develop our own evaluation metrics tailored to the specific needs of clinical coders and other health professionals. Collaboration with clinical coders and health professionals will be integral to our methodology, as they will provide feedback on the summaries generate by the AI tool to ensure accuracy and relevance. 

Data will be collected from various medical records, including publicily available sources such as MIMIC, and Australian datasets like the Cardiac Analytics and Innovation (CardiacAI) Data Repository. The performance of the summarization tool will be assessed through rigorous testing and validation processes.