In the paper, a probabilistic model of the eating behavior of subjects with type 1 diabetes is proposed and validated against two extensive datasets collected during experiments in free-living conditions in Padova and Amsterdam clinical centers. Meals are modeled as a discrete-time marked point process. The random meal events are associated to a Markov Chain whose state is the fasting period and the transition probabilities depend on variables such as daytime and carbohydrate intake of the previous meal. The random marks associated to each meal represent the carbohydrate intake and their distribution possibly depends on daytime, fasting period and carbohydrate intake of the previous meal. Logistic and Gaussian Mixture models are used to identify two models from the Padova and Amsterdam datasets, respectively. In order to validate it, the proposed model is used to simulate synthetic datasets, whose statistical properties appear to be in good agreement with those of the experimental datasets. The availability of a probabilistic model of eating behavior is expected to be valuable for several purposes ranging from the optimization and customization of automatic insulin dosing systems to decision supporting tools for insulin dosing and the alert management for missing meal announcements.

Model-based identification of eating behavioral patterns in populations with type 1 diabetes

Aiello E. M.
;
Toffanin C.;Magni L.;De Nicolao G.
2022-01-01

Abstract

In the paper, a probabilistic model of the eating behavior of subjects with type 1 diabetes is proposed and validated against two extensive datasets collected during experiments in free-living conditions in Padova and Amsterdam clinical centers. Meals are modeled as a discrete-time marked point process. The random meal events are associated to a Markov Chain whose state is the fasting period and the transition probabilities depend on variables such as daytime and carbohydrate intake of the previous meal. The random marks associated to each meal represent the carbohydrate intake and their distribution possibly depends on daytime, fasting period and carbohydrate intake of the previous meal. Logistic and Gaussian Mixture models are used to identify two models from the Padova and Amsterdam datasets, respectively. In order to validate it, the proposed model is used to simulate synthetic datasets, whose statistical properties appear to be in good agreement with those of the experimental datasets. The availability of a probabilistic model of eating behavior is expected to be valuable for several purposes ranging from the optimization and customization of automatic insulin dosing systems to decision supporting tools for insulin dosing and the alert management for missing meal announcements.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11571/1452180
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 6
  • ???jsp.display-item.citation.isi??? 5
social impact