Predicting the meal macronutrient composition from continuous glucose monitors

Abstract

Sustained high levels of blood glucose in type 2 diabetes (T2DM) can have disastrous long-term health consequences. An essential component of clinical interventions forT2DM is monitoring dietary intake to keep plasma glucose levels within an acceptable range. Yet, current techniques to monitor food intake are time intensive and error prone. To address this issue, we are developing techniques to automatically food intake and the composition of those foods using continuous glucose monitors (CGMs). This article presents the results of a clinical study in which participants consumed nine standardized meals with known macronutrients amounts (carbohydrate, protein, and fat) while wearing a CGM. We built a multitask neural network to estimate the macronutrient composition from the CGM signal, and compared it against a baseline linear regression. The best prediction result comes from our proposed neural network, trained with subject-dependent data, as measured by root mean squared relative error and correlation coefficient. These findings suggest that it is possible to estimate macronutrient composition from CGM signals, opening the possibility to develop automatic techniques to track food intake.

Innovative Solution or Cause for Concern? The Use of Continuous Glucose Monitors in People Not Living with Diabetes: A Narrative Review.

“…Enhanced accuracy in dietary predictions is plausible when merging diverse datasets with CGM metrics. 85 Yet, prior to the mainstreaming of these advanced glycaemic data analysis techniques, robust validation is mandated. Current reproducibility challenges attributable to variation in approaches for metric calculation and lack of algorithms validated on public datasets, result in inconsistent values of the same CGM metrics across software applications.…”

“…While similar calorie and fitness monitoring has been linked to the development of anxiety, compulsive behaviours, or disordered eating patterns, the consequences of CGM data on exacerbating these issues has not been extensively explored. 85 Studies measuring the effect of CGMs on the quality of life, disordered eating and other potential non-clinical factors would provide guidance for PNLD should they want to use the device. 13 Finally, by presenting the findings in a narrative format this review aims to allow for a broad perspective on the developments in the field.…”

Digital Biomarkers for Personalized Nutrition: Predicting Meal Moments and Interstitial Glucose with Non-Invasive, Wearable Technologies

“…In T1D patients, meal detection is applied to control insulin administration, whereas, in a healthy or a (pre-)T2D population, meal detection can be used to provide individuals with more insight into their eating behavior and may provide opportunities for personalized feedback on frequency or timing of eating moments. In the future, it may even be possible to predict both meal timing and dietary composition from CGM data [ 44 ], which would provide even more opportunities for personalized advice to stimulate behavior change. Meal detection algorithms could also play a role in improving the quality of food diary applications.…”

Towards The Development of Subject-Independent Inverse Metabolic Models

“…kernels out of this range did not have significant impact on the results 3. We also experimented using Neural networks[10] and Adaboost models, but there was no significant difference in the results.…”

“…A preliminary version of this study, involving 9 subjects, showed poor results due to individual variability[10]. Here, we increase the number of subjects, address individual variability, and propose an updated IMM.…”