This project addresses the need for earlier and more accessible detection of heart failure risk, a condition that often presents late and typically requires costly, specialised investigations. Delayed diagnosis limits opportunities for early intervention and increases long-term healthcare burden.

The project develops an AI-based binary stratification system that integrates both structured and unstructured clinical data. Structured data include routinely collected physiological and clinical parameters, while the unstructured component uniquely incorporates patient imaging in the form of chest X-rays. By combining these data streams, the system provides a more comprehensive assessment of heart failure susceptibility than conventional single-modality approaches.

A key innovation lies in the use of hybrid machine learning, where deep learning extracts image-based features from chest X-rays and merges them with clinical indicators into a unified model. Through this approach, the researchers have identified 13 key physiological parameters that serve as early indicators of heart failure and cardiac arrest risk. This hybrid, data-integrated strategy enables more nuanced and clinically meaningful risk stratification.

The system is designed for practical deployment in clinical workflows, enabling rapid preliminary assessment using standard chest X-rays and routine body check-up reports, without the need for specialised cardiac scans at the initial stage. This significantly reduces diagnostic cost and turnaround time, allowing clinicians to identify high-risk patients earlier and prioritise further investigation or intervention.

The project is supported by Ogawa Malaysia, strengthening its translational and industry relevance. It also contributes to human capital development, through the deployment of both local and international biomedical engineering talent working directly within a clinical environment. This interdisciplinary setting bridges engineering expertise with real-world clinical needs, supporting the development of clinically grounded AI solutions.

Looking ahead, the project is positioned for integration into clinical care pathways as a decision-support tool, supporting preventive cardiology, improving patient outcomes, and reducing downstream healthcare burden through earlier, lower-cost risk assessment.