Meal and Physical Activity Detection from Free-Living Data for Discovering Disturbance Patterns of Glucose Levels in People with Diabetes

Abstract

Objective: The interpretation of time series data collected in free-living has gained importance in chronic disease management. Some data are collected objectively from sensors and some are estimated and entered by the individual. In type 1 diabetes (T1D), blood glucose concentration (BGC) data measured by continuous glucose monitoring (CGM) systems and insulin doses administered can be used to detect the occurrences of meals and physical activities and generate the personal daily living patterns for use in automated insulin delivery (AID). Methods: Two challenges in time-series data collected in daily living are addressed: data quality improvement and the detection of unannounced disturbances of BGC. CGM data have missing values for varying periods of time and outliers. People may neglect reporting their meal and physical activity information. In this work, novel methods for preprocessing real-world data collected from people with T1D and the detection of meal and exercise events are presented. Four recurrent neural network (RNN) models are investigated to detect the occurrences of meals and physical activities disjointly or concurrently. Results: RNNs with long short-term memory (LSTM) with 1D convolution layers and bidirectional LSTM with 1D convolution layers have average accuracy scores of 92.32% and 92.29%, and outperform other RNN models. The F1 scores for each individual range from 96.06% to 91.41% for these two RNNs. Conclusions: RNNs with LSTM and 1D convolution layers and bidirectional LSTM with 1D convolution layers provide accurate personalized information about the daily routines of individuals. Significance: Capturing daily behavior patterns enables more accurate future BGC predictions in AID systems and improves BGC regulation.

Detection of Physical Activity Using Machine Learning Methods Based on Continuous Blood Glucose Monitoring and Heart Rate Signals

“…We have shown that when using the blood glucose data of real patients, we need additional features, such as the ones extracted from heart rate signals. The Logistic Regression, AdaBoost, Random Forest, and Multi-Layer Perceptron with the ReLU and Tanh activation function were found to be the best five models from our tests; they provide better or comparable results to those reported in similar studies, e.g., [ 61 , 62 , 63 , 64 ]. The other models require a post-processing of the predicted data to be more precise.…”

A lightweight deep learning with feature weighting for activity recognition

“…Advancements in pervasive computing have seen an increase in the interest of researchers in human activity recognition. Human activity recognition has numerous application areas, such as; elder healthcare, 1 child monitoring, 2 rehabilitation monitoring, and general well‐being, 3 among other areas. Generally, activity data can be collected using the vision‐based method or sensor‐based method.…”

New Contributions Towards Meal and Exercise Announcement-Free Artificial Pancreas Systems

“…Other methods use information from behavioral meal patterns to confirm a meal occurrence (Cameron et al 2012;Villeneuve et al 2020). Lastly, classification algorithms have also been used to discern the meal events, such as logistic regression (Garcia-Tirado et al 2021c;Garcia-Tirado et al 2021b;Corbett et al 2022), linear discriminant analysis (Kölle et al 2017;Kölle et al 2020), extended isolation forest (Zheng et al 2020), fuzzy logic (Samadi et al 2017), or recursive neural networks (Askari et al 2022).…”

“…Food intake usually follows daily patterns (Askari et al 2022). Some predictionbased controllers, namely MPC, have exploited this property to enhance the prediction quality against unannounced meals and consequently improve the control performance.…”

“…Mahmoudi et al (2019) consider the cumulative sum of the error between CGM and the glucose estimated by a KF;Zheng et al (2019) construct residuals from the predictions of multiple models; andAskari et al (2022) combine CGM-derived features (product of derivatives, correlation between CGM and its derivatives, frequency of dominant peaks) with estimations of plasma concentration and gut absorption rate.…”