State-of-the-art in exploring Integration of Machine Learning and Deep Learning Techniques with Medical Parameters for Diabetic Clinical Decision Support

Diabetes is a prevalent global health condition, and diabetic neuropathy stands out as a frequently encountered and significant consequence associated with this disease. Diabetic autonomic neuropathy (DAN) is a form of diabetic neuropathy characterized by dysfunction of the autonomic nerve system, leading to potential impairment of several organs like the heart and kidneys. DAN is frequently subject to underdiagnosis, mainly due to factors such as the financial burden and limited accessibility of diagnostic equipment, the challenges associated with conducting cardiovascular assessments, and the sometimes silent nature of the disease during its initial phases. In light of this, the purpose of this research is to leverage information retrieved from electronic health records to employ machine learning to detect DAN, particularly an emphasis on diabetic cardiac autonomic neuropathy (DCAN) in diabetic patients. The machine learning strategy yields a maximum accuracy of 85.5% through the implementation of the Random Forest (RF) and one-dimensional Convolutional Neural Network (1D CNN) model. Conversely, the deep learning approach, employing the Long Short-Term Memory (LSTM) model, achieves an accuracy of 95.7%.