Tillmann Int-Veen, Supervisor: Nico Steckhan

Evidences suggest that meal time-based strategies can be employed to prevent obesity, diabetes and associated diseases [1]. The ongoing ChronoFast study, recruiting since January 2020 and conducted by the German Institute for Nutritonal Research (DIfE), is a cross-over trial that serves as the data foundation of this thesis. The study investigates the effects of early or late time-restricted eating on the glucose metabolism of overweight, non-diabetic women. The study intervention lasts fourteen days. Each participant includes two weeks of early time-restricted eating (eTRE) and late time-restricted eating (lTRE). At the beginning of the study, four weeks are reserved as a run-in phase, then participants complete two weeks of one fasting phase, two weeks of washout phase, then two weeks of the other fasting phase. eTRE subjects may eat from 7:00 - 15:00. lTRE subjects may eat from 12:00 - 20:00. Hence, each of the hereafter mentioned data points includes fourteen days worth of data at the mentioned sampling rate, unless stated otherwise. For each participant, this includes the data from actigraphy, continuous glucose monitoring, metabolic tests like the oral glucose tolerance test, as well as clinical and nutritional data. 

McDonnell et al. [2] have created new, clinically useful parameters, namely continuous overall glycemic variability, by analyzing CGM data. Yu et al. [3] have further shown how the measurement of existing clinical reference assessments can be improved with an algorithmic approach [4-6]. This master's thesis aims to explore the degree and granularity to which a person's intermittent fasting schedule can be personalized. And therefore optimizing the health benefit of intermittent fasting with the help of circadian and glucose prediction models that use further clinical and nutritional data.

[1] Michael J Wilkinson et al. “Ten-hour time-restricted eating reduces weight, blood pressure,and atherogenic lipids in patients with metabolic syndrome”. In:Cell metabolism31.1 (2020),pp. 92–104.

[2] CM McDonnell et al. “A novel approach to continuous glucose analysis utilizing glycemic variation”. In:Diabetes technology & therapeutics7.2 (2005), pp. 253–263.

[3] Xuefei Yu et al. “Calculating the mean amplitude of glycemic excursions from continuousglucose data using an open-code programmable algorithm based on the integer nonlinearmethod”. In:Computational and mathematical methods in medicine2018 (2018).

[4] Chul-Hyun Cho et al. “Mood prediction of patients with mood disorders by machine learn-ing using passive digital phenotypes based on the circadian rhythm: prospective observa-

tional cohort study”. In:Journal of medical Internet research21.4 (2019), e11029.

[5] Julia E Stone et al. “Generalizability of a neural network model for circadian phase prediction in real-world conditions”. In:Scientific reports9.1 (2019), pp. 1–17.

[6] Chiara Zecchin et al. “A new neural network approach for short-term glucose prediction using continuous glucose monitoring time-series and meal information”. In:2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE. 2011,pp. 5653–5656