Innovative AI System Aims for Early Detection of Cardiovascular, Diabetic Eye Diseases, and Cancer

Researchers at Edith Cowan University (ECU) have developed an advanced Artificial Intelligence (AI) technology designed to assist healthcare professionals in the early diagnosis and staging of severe health conditions, including cardiovascular disease (CVD), diabetic retinopathy (DR), and various cancers. This innovative AI is called the Supervised Contrastive Ordinal Learning algorithm, and it leverages routine, noninvasive medical images such as ultrasounds and bone density scans to identify subtle disease-specific changes that aid in accurate detection and disease staging.

Dr. Afsah Saleem from ECU emphasized the critical need for noninvasive diagnostic tools for conditions like CVD and DR, which often progress silently. Globally, CVD affects over 640 million individuals and is a leading cause of death in Australia, accounting for one in four fatalities. Meanwhile, diabetic retinopathy, a primary cause of blindness, affects over 103 million people worldwide, with projections reaching 160 million by 2045. In Australia, nearly 1.9 million individuals have diabetes, and roughly one-third of these develop DR over time.

Traditional diagnostic approaches largely depend on manual analysis of medical scans, which can be time-consuming, costly, and subjective. The AI system aims to streamline this process, making early detection more accessible and reliable. Notably, this algorithm has achieved promising results, with accuracy rates of up to 91% in distinguishing different stages of breast cancer and high sensitivity in identifying early indicators of cardiovascular issues.

ECU researcher Dr. Saleem highlighted that the system's ability to learn and differentiate the unique features of healthy versus diseased individuals enhances its diagnostic precision. This approach offers significant potential for reducing the delays and expenses associated with conventional methods. Dr. Saleem plans to present her research on diabetic retinopathy at the upcoming Medical Image Computing and Computer Assisted Intervention Conference in Korea.

The development represents a step forward in incorporating machine learning into medical imaging, offering a less invasive, cost-effective solution for managing chronic diseases. As this technology continues to evolve, it promises to improve early diagnosis rates and ultimately save lives.

For more information, the research was published as: Afsah Saleem et al, Deep-Attention Feature Fusion Network for Automated Diagnosis of Diabetic Retinopathy Using Fundus Photographs, 2024 International Conference on Digital Image Computing: Techniques and Applications (DICTA) (2025). [DOI: 10.1109/DICTA63115.2024.00077]

Source: https://medicalxpress.com/news/2025-07-ai-early-cardiovascular-disease-diabetic.html