Smart Digital Twin System for optimized Elderly Type 2 Diabetes through Personalized Management using Adaptive Predictive Control

Patients admitted with E-T2D have multiple comorbid conditions and often require polypharmacy atypical of a younger T2D patient who the management expects to be uncomplicated. This work proposes for a Smart Digital Twin (SDT) framework an advanced digital representation that caters individualized and improved diabetes care for the elderly through real-time shared data. Incorporating activities and profiles of patients, diagnostic matrices are used to predict steady and kinetic variations in glucose level, medication reactions, and physiological status of patients and the SDT applies system controls to adapt therapeutic treatments. This framework includes an adaptive, learning-based reality model predictive control (LB-MPC) algorithm to continuously adjust insulin and medication dosing in real-time, based on the unique patient’s metabolic response, drug interactions and geriatric considerations. Such an approach enables accurate, planned adjustments that fine-tune the treatment, as well as its risk profile. A breakthrough was observed from clinical trials and simulation based on 15 patients, where the time-in-range (TIR) glucose level raised from a baseline of 3-75 % up to the range of 88-96 % together with moderate reduction of insulin demand of 15-30 %. SDT framework has pointed out the new trends of the application of IoMT, artificial intelligence, and adaptive control in geriatric diabetes care, to target the precision medicine and releasing the potential of digital health for elderly patients.