Opportunity
Type 2 diabetes (T2D) is a global epidemic affecting 496 million people globally. Currently, T2D is diagnosed 5 to 7 years after the onset of the first symptoms. Often, at the time of diagnosis, patients have already developed diabetes complications that negatively impact their vision, kidneys, nerves, and cardiovascular system, and contribute to excess mortality.
We aim to significantly reduce the time-lag between the onset of diabetes symptoms and diagnosis by adapting an artificial intelligence (AI) model capable of analysing a voice “signature” (characteristics of voice) to predict the risk of T2D for Australians. This approach is rooted in emerging evidence suggesting that vocal biomarkers can provide insights into various health conditions, including metabolic (T2D), mental health, heart failure, and neurological disorders. Our project involves using participant voice data and correlating it with several metabolic and cardiovascular risk factors for a comprehensive analysis using AI, thus enhancing the accuracy and effectiveness of the predictive model.
This project could pave the way for an innovative, non-invasive, and cost-effective risk prediction tool to alert individuals who may potentially have type 2 diabetes to seek medical advice in a timely manner. By integrating the AI-driven tool into health platforms or mobile services, Australians will have access to timely, efficient, and accessible disease detection solutions through innovative and sustainable digital health initiatives.
Project Objectives
- Refine an AI system to analyse voice data for predicting the risk of T2D in an Australian setting
- Validate an AI T2D risk prediction model using chronic glycaemic control (HBA1c) and continuous glucose monitoring data
- Co-design a T2D risk prediction tool that integrates with existing healthcare services to facilitate early detection and management
