Type 1 diabetes (T1D) is a metabolic disease affecting insulin production, necessitating continuous monitoring, and numerous daily interventions for effective management. In pediatric patients, these challenges are amplified by their evolving physiology. Although advancements in technology, such as continuous glucose monitoring (CGM) sensors and smart pens for multiple daily insulin injections, have improved T1D management, achieving a good glycemic control remains a difficult task. To address this challenge, we are developing TWIN, a new clinical decision support system for pediatric patients. TWIN is designed to enhance the effectiveness of therapy-adjustment routine visits by providing clinicians with personalized therapeutic suggestions through advanced analysis and visualization tools. The system consist of a mobile application for data collection from wearable sensors, a cloud-based therapy optimization algorithm leveraging digital twins, and a web interface for clinicians to visualize recommendations and analyze patient’s data. The use of digital twins enables the system to directly optimize and personalize the insulin therapy, keeping in consideration the ever-changing physiology typical of the pediatric population. The clinical effectiveness of TWIN’s optimization algorithm is being evaluated through a simulated 24-week clinical trial using data extracted from the Tidepool dataset. In this trial, basal insulin injections are adjusted bi-weekly, and preliminary results indicate improvements in glucose control metrics over the course of the trial. The same dataset has been employed to prototype the web interface that will deliver TWIN recommendations. This prototype was developed in collaboration with expert clinicians and employing Health Design Thinking techniques to identify the clinical workflow during routine visits and to understand open challenges. Furthermore, data has been used to develop and test advanced data analysis tools, including a data summarization tool that generates a concise free text paragraph containing clinically relevant information on patient glycemic control.
Enhancing pediatric type 1 diabetes therapy using TWIN, a clinical decision support system based on digital twins
Cossu L.;Pellizzari E.;Cappon G.;Sparacino G.;Facchinetti A.
2025
Abstract
Type 1 diabetes (T1D) is a metabolic disease affecting insulin production, necessitating continuous monitoring, and numerous daily interventions for effective management. In pediatric patients, these challenges are amplified by their evolving physiology. Although advancements in technology, such as continuous glucose monitoring (CGM) sensors and smart pens for multiple daily insulin injections, have improved T1D management, achieving a good glycemic control remains a difficult task. To address this challenge, we are developing TWIN, a new clinical decision support system for pediatric patients. TWIN is designed to enhance the effectiveness of therapy-adjustment routine visits by providing clinicians with personalized therapeutic suggestions through advanced analysis and visualization tools. The system consist of a mobile application for data collection from wearable sensors, a cloud-based therapy optimization algorithm leveraging digital twins, and a web interface for clinicians to visualize recommendations and analyze patient’s data. The use of digital twins enables the system to directly optimize and personalize the insulin therapy, keeping in consideration the ever-changing physiology typical of the pediatric population. The clinical effectiveness of TWIN’s optimization algorithm is being evaluated through a simulated 24-week clinical trial using data extracted from the Tidepool dataset. In this trial, basal insulin injections are adjusted bi-weekly, and preliminary results indicate improvements in glucose control metrics over the course of the trial. The same dataset has been employed to prototype the web interface that will deliver TWIN recommendations. This prototype was developed in collaboration with expert clinicians and employing Health Design Thinking techniques to identify the clinical workflow during routine visits and to understand open challenges. Furthermore, data has been used to develop and test advanced data analysis tools, including a data summarization tool that generates a concise free text paragraph containing clinically relevant information on patient glycemic control.
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
