Prof. Dr.-Ing. Bert Arnrich


Development of Nutritional-Behavior-Maps from Wearables

Funded by the Federal Ministry for Economic Affairs and Energy.

Arpita Kappattanavar, Nico Steckhan, Pascal Hecker

The number of people with diabetes rose from 108 million in 1980 to 422 million in 2014 [1]. The global prevalence of diabetes among adults above 18 years has risen from 4.7% in 1980 to 8.5% in 2014 [1]. The leading risk factor for type 2 diabetes is obesity [1]. Obesity is usually caused by eating too much and moving too little. According to research in the addiction and nutrition field, unregulated ‘binge-eating’ can cause obesity, particularly when eating energy-dense (‘addictive’) food [2]. Stress is suspected to drive excessive food intake and weight gain [3]. Affective states such as anxiety, anger, depressive mood and other negative emotions correlate with eating behavior [4]. Moreover, the exact cause of the corresponding ‘emotional eating’ remains unclear and requires methodological investigation [4]. 

EatMaps aims to transfer knowledge on how affective states and user contexts correlate with eating habits in everyday life. Here we collect physiological signals from the sensors present in smart wearables and we classify the emotional state of the person by integrating his/her present context (activity (e.g. walking, sitting, eating), location (e.g. at frequently visited places or unknown environment) and social interactions (e.g. alone, in a group)) using Machine learning and Neural Networks. Further, the context related emotions are correlated with food images classified by Oviva application, to understand the eating behavior. Figure 1 presents the graphical representation of EatMaps.

We are conducting this study along with Oviva AG, which is our partnered institute for this funding. Oviva has several years of expertise in the topic of monitoring eating habits of diabetic and obese patients. Oviva provides and supports digital solutions for patient-oriented therapy to change dietary behavior via a mobile application. Therein, patients can take pictures of their meals, which are analyzed and categorized with state-of-the-art machine learning methods. Based on the analysis, nutrition experts assigned to the patients, supervise the patient’s eating behavior and provide intervention if needed.

[1] Roglic, G. (2016). WHO Global report on diabetes: A summary. International Journal of Noncommunicable Diseases, 1(1), 3.
[2] Pursey, K. M., Collins, C. E., Stanwell, P., & Burrows, T. L. (2015). Foods and dietary profiles associated with ‘food addiction’ in young adults. Addictive behaviors reports, 2, 41-48.
[3] Sinha, R. (2018). Role of addiction and stress neurobiology on food intake and obesity. Biological psychology, 131, 5-13.
[4] Nyklíček, I., Vingerhoets, A., & Zeelenberg, M. (Eds.). (2010). Emotion regulation and well-being. Springer Science & Business Media.