Early Prognosis of Diabetes Harnessing Physiological Data and Artificial Intelligence

Abstract

This paper presents an artificial intelligence-based early detection technique for individuals with diabetes. According to the International Diabetes Federation (2024), the number of adults with diabetes is projected to exceed 850 million by 2050. Unhealthy food habits, physical inactivity, and family history are commonly blamed for diabetes. Improperly managed diabetes can lead to life-threatening complications, including cardiovascular diseases, kidney failure, nerve damage, and vision problems. Hence, early detection of diabetes is crucial and has become a primary focus of recent research. Computer-based disease detection, powered by artificial intelligence, can play a pivotal role here. With recent advances in algorithms and artificial intelligence, these technologies have become increasingly popular across diverse fields of biomedicine and bioinformatics, leading to rapid advancements in computer-based disease diagnosis. This work investigates the Pima Indian Diabetes Dataset and demonstrates that a shallow feedforward neural network (FFNN) can predict diabetes from critical biological data, achieving 79.1% accuracy. This population-based projection measure can effectively alert individuals to be vigilant and participate in recommended health screenings.