Toward crises prediction and enhancing patient care in myasthenia gravis using telemonitoring and wearable-based data
Final Report Abstract
Myasthenia gravis (MG) is a rare, chronic autoimmune disease characterized by exercisedependent muscle weakness that can lead to life-threatening myasthenic crises. The course of the disease is fluctuating and highly variable both inter- and intra-individually. Therefore, long-term, often lifelong specialized care is usually required, which in practice is often made more difficult by limited access to specialized centers and long waiting times for appointments. There is a “black box” between regular medical appointments and there are no predictive markers that enable early detection of myasthenic crises. To address this gap in care, we have developed MyaLink - a certified telemedicine platform consisting of an app for patients and a web-based portal for physicians. The app enables the continuous recording of disease-specific symptoms using Patient-Reported Outcome Measures (PROMs) and the integration of sensor-based data from wearables (e.g., step count, heart rate, O2 saturation, spirometry). At the same time, MyaLink enables a bidirectional exchange of health data and secure patient-physician communication via a communication module, thus enabling realtime monitoring of the course of the disease, including remote therapy adjustments if necessary. For my Walter Benjamin Program, the effectiveness of this digital platform in improving care for patients with MG was examined based on the data from an initial randomized controlled trial alongside the development of machine-learning based models for crises prediction. The aim was to investigate whether additional telemedical care with MyaLink could reduce disease burden and improve quality of life for MG patients. Furthermore, communication patterns between patients and physicians were analyzed to explore what conclusions could be drawn for the design of telemedical interventions and the specific information needs involved. In addition, a predictive machine learning-based model was developed, utilizing multimodal telemonitoring data to identify digital predictive markers for the early detection of myasthenic crises (“digital phenotyping”). The goal is to identify high-risk patients at an early stage and to prevent avoidable myasthenic crises through timely, personalized support. The results of my Walter Benjamin Program aim to contribute to the development of data-driven, needsoriented care in MG and to pave the way for the sustainable integration of digital health solutions into clinical routine.
