AI-Based Risk Prediction Model Helps Prevent Foot Amputations in Diabetic Patients
Singapore, (ANTARA/PRNewswire) - Patient-centric health monitoring is set to become the new standard in managing diabetic patients at high risk of lower extremity amputation (LEA). Developed by Singapore General Hospital (SGH), SingHealth, and the MOH Office for Healthcare Transformation (MOHT), the artificial intelligence model known as LEA – Neural Network Model (LEA-Net) assists healthcare professionals in detecting amputation risks earlier, enabling preventative measures to be taken before a patient’s condition deteriorates.
LEA-Net can predict the risk of lower extremity amputation three to five years before patients develop foot ulcers and infections. This allows medical teams to intervene before permanent tissue damage occurs. The model categorises patients into low and high-risk groups, streamlining targeted intervention procedures and reducing waiting times for specialist vascular care. In Singapore, nearly nine out of ten patients undergoing lower extremity amputations are known to have diabetes. Approximately 85% of amputation cases also begin with foot ulcers.
Lower extremity amputation has serious consequences on a patient’s mobility, independence, social life, and overall health. Diabetic or vascular disease patients who have undergone amputation are also at higher risk of complications from further wounds and loss of other limbs. Financially, the impact is significant. Early-stage interventions average around US$25,000, while advanced-stage treatment can exceed US$40,000 to US$50,000 per patient.
Diabetic patients typically undergo annual check-ups for foot, eye, and kidney conditions. However, compliance rates for diabetic foot examinations are lower compared to the other two. Many patients only seek medical help when ulcers or infections have already appeared, increasing the risk of amputation, as well as other complications such as surgical complications and death.
LEA-Net was developed using anonymised data consisting of over 830,000 patient records from SingHealth, including demographic data, clinical conditions, and medical examination results. Of these, approximately 250,000 records were used for model validation. The model demonstrates a sensitivity rate of nearly 80% and a specificity approaching 90% in predicting lower extremity amputation risk. This performance is significantly better than several other models for both parameters.
The model’s testing results show that high-risk patients can be identified during the critical period before ulcers and infections appear. The model also reflects a shift towards data-driven, preventative healthcare services within the Healthier SG programme.
LEA-Net won the “People’s Choice Award” at the International Consortium for Health Outcomes Measurement Conference 2025 (ICHOM) 2025 in Dublin, Ireland. The LEA-Net development team is now working to further validate its clinical effectiveness through a pilot study involving patients from SingHealth’s Diabetes Registry.
SOURCE Singapore General Hospital