Application of Deep Learning and Biomarkers in the Prediction of Gastrointestinal Cancer and Chronic Kidney Disease in Patients with Type 2 Diabetes

Abstract

Type 2 diabetes mellitus (DM2) is associated with a high risk of developing chronic complications such as chronic kidney disease (CKD) and different types of cancer, especially of the gastrointestinal tract. In this context, advances in deep learning and biomarker identification allow the development of highly accurate and non-invasive predictive tools. This article presents a critical review and methodological proposal for the application of deep learning models based on clinical data and molecular biomarkers to predict the occurrence of gastrointestinal cancer and CKD in patients with T2D. Computational approaches are integrated with recent clinical evidence, highlighting the usefulness of convolutional neural network (CNN) models and attention models, as well as the integration of omics and clinical-biochemical parameters in prediction. Preliminary results show high diagnostic accuracy, suggesting a promising approach to improve prognosis and personalization of treatments in preventive medicine.