Are you interested in how the body responds to food and whether we can detect those responses from wearable data alone? In collaboration with the Department of Endocrinology and Diabetology at University Hospital Ulm, this project aims to improve our understanding of typical physiological responses to meals in free-living conditions. The ultimate goal is to identify data-driven indicators of meal events and reduce reliance on self-reported dietary information.
This thesis involves integrating and analyzing multimodal time-series data to capture physiological changes during and after food intake. You will work with continuous measurements of glucose levels, cardiac response, physical activity, and core body temperature to uncover consistent postprandial patterns. Key tasks include pre-processing and aligning diverse sensor data streams, extracting informative features, and developing statistical and machine learning models for meal detection.
The project offers the opportunity to contribute to the growing field of personalized digital health by developing tools that enhance our understanding of metabolic responses in real-world settings.
References
1.    Miyakoshi, T.; Ito, Y. M. (2024). Association of Blood Glucose Data with Physiological and Nutritional Data from Dietary Surveys Investigated Using Publicly Available Wearable-type Databases (Preprint). JMIR Diabetes, 9, e62831. doi.org/10.2196/62831
2.    Ali, H.; Niazi, I.K.; White, D.; Akhter, M.N.; Madanian, S. Comparison of Machine Learning Models for Predicting Interstitial Glucose Using Smart Watch and Food Log. Electronics 2024, 13, 3192. doi.org/10.3390/electronics13163192