Diabetes mellitus has become a widespread metabolic disease in contemporary medicine. In 2019, an estimated 463 million people were living with diabetes, representing 9.3% of the global adult population (ages 20-79). This number is projected to increase to 578 million (10.2 percent) in 2030 and 700 million (10.9 percent) in 2045. According to the etiology and pathology of diabetes, there are three main clinical types: type 1 diabetes, type 2 diabetes and gestational diabetes. T3cDM (pancreatic diabetes, type 3c diabetes) is a secondary diabetes mellitus resulting from endocrine and exocrine insufficiency caused by disease or injury of the pancreas. At present, the specific pathogenesis of T3cDM is not clear, and there is still a lack of effective diagnostic methods and predictive indicators in clinical practice. It is often misdiagnosed as type 2 diabetes mellitus. Therefore, it is of great significance to find a more accurate, simple and feasible diagnostic method for the treatment of T3cDM. In the past few years, machine learning and deep learning technologies have been widely applied in the medical field. There have been many published studies using machine learning and deep learning in the diagnosis of type 1 and type 2 diabetes mellitus, indicating that the diagnosis of T3cDM by using machine learning and deep learning has a broad development prospect.