Cardiac autonomic neuropathy (CAN) is a diabetes-related condition that is becoming more common but still difficult to diagnose in clinical settings. Using clinical data, machine learning (ML) techniques have the ability to predict CAN. For a study, researchers sought to create and analyze the efficacy of a machine learning model for predicting early CAN incidence in diabetic patients. They utilized the data set from the Diabetes Complications Screening Research Initiative, which included 200 CAN-related tests on over 2000 people with type 2 diabetes. Peripheral nerve functions, Ewing’s tests, blood biochemistry, demography, and medical history were all gathered. The ML model was verified using 10-fold cross-validation, with 90% of the data utilized for training and 10% for evaluating the model’s performance. The area under the receiver operating curve, as well as sensitivity, specificity, positive predictive value, and negative predictive value, were used to evaluate predictive accuracy.

The study comprised 237 individuals, 105 of whom had been diagnosed with CAN in their early stages while the rest 132 were healthy. With a receiver operating characteristic curve of 0.962 [95% CI=0.939–0.984], 87.34% accuracy, and 87.12% sensitivity, the ML model performed quite well for CAN prediction. The ML model and the incidence of CAN had a strong and positive relationship (P<0.001). Using Ewing’s tests, the ML model has the ability to diagnose CAN at an early stage. The approach might help healthcare practitioners forecast the incidence of CAN in diabetic patients, track their course, and give prompt management.