Abstract Background This study evaluated a remote monitoring program for IBD patients, using a mobile app to track disease activity and improve patient-provider communication. The app included disease activity questionnaires, quality of life assessments, and home-based calprotectin tests. we aimed to develop a data-driven, patient-reported care management system for early detection of clinical flares in IBD patients, enabling timely treatment adjustments. Methods A 12-month prospective cohort study included 100 adult IBD patients. Participants completed monthly online questionnaires via the “DATOS” platform, assessing disease activity (PRO2/SCCAI), quality of life (PROMIS-10, SIBDQ), emotional state (DASS-21), and perceived disease control (IBD CONTROL). Home calprotectin tests were performed at baseline, 6 months, and 12 months. Data from 533 monthly observations were used to develop a machine learning model to predict flares. Results Among 100 patients (74% Crohn’s, 26% ulcerative colitis; median age 31 years), 94% were on biological therapy. Adherence to the remote platform declined over time (86% at 0 months, 53% at 6 months, 30% at 12 months). Over the study period, 40 flares, 14 hospitalizations, 6 surgeries, and 31 changes in biological therapy occurred. Patients experiencing a flare had significantly higher fecal calprotectin levels (mean: 901.8 vs. 330.6 mg/kg, p=0.044), as well as increased anxiety, depression, and stress (DASS-21, p<0.05), poorer quality of life (PROMIS-10, p=0.005; SIBDQ, p<0.001), and reduced perceived disease control (IBD-Control, p<0.001). A machine learning model incorporating calprotectin and patient-reported outcomes demonstrated a 0.85 accuracy in predicting flares. Conclusion This study demonstrates the feasibility of a remote monitoring program for IBD patients. Although the program facilitated streamlined patient follow-up, long-term adherence was challenging, highlighting the need for improvements in system usability and patient engagement. The machine-learning model, which combined fecal calprotectin and patient-reported outcomes, showed good accuracy in identifying patients at risk of flares. These findings suggest that remote monitoring, supported by machine learning, could enhance IBD management.
Read full abstract