Disease severity-based subgrouping of type 2 diabetes does not parallel differences in quality of life: the Maastricht Study.

AIMS/HYPOTHESIS: Type 2 diabetes is a highly heterogeneous disease for which new subgroups ('clusters') have been proposed based on disease severity: moderate age-related diabetes (MARD), moderate obesity-related diabetes (MOD), severe insulin-deficient diabetes (SIDD) and severe insulin-resistant diabetes (SIRD). It is unknown how disease severity is reflected in terms of quality of life in these clusters. Therefore, we aimed to investigate the cluster characteristics and cluster-wise evolution of quality of life in the previously defined clusters of type 2 diabetes. METHODS: We included individuals with type 2 diabetes from the Maastricht Study, who were allocated to clusters based on a nearest centroid approach. We used logistic regression to evaluate the cluster-wise association with diabetes-related complications. We plotted the evolution of HbA1c levels over time and used Kaplan-Meier curves and Cox regression to evaluate the cluster-wise time to reach adequate glycaemic control. Quality of life based on the Short Form 36 (SF-36) was also plotted over time and adjusted for age and sex using generalised estimating equations. The follow-up time was 7 years. Analyses were performed separately for people with newly diagnosed and already diagnosed type 2 diabetes. RESULTS: We included 127 newly diagnosed and 585 already diagnosed individuals. Already diagnosed people in the SIDD cluster were less likely to reach glycaemic control than people in the other clusters, with an HR compared with MARD of 0.31 (95% CI 0.22, 0.43). There were few differences in the mental component score of the SF-36 in both newly and already diagnosed individuals. In both groups, the MARD cluster had a higher physical component score of the SF-36 than the other clusters, and the MOD cluster scored similarly to the SIDD and SIRD clusters. CONCLUSIONS/INTERPRETATION: Disease severity suggested by the clusters of type 2 diabetes is not entirely reflected in quality of life. In particular, the MOD cluster does not appear to be moderate in terms of quality of life. Use of the suggested cluster names in practice should be carefully considered, as the non-neutral nomenclature may affect disease perception in individuals with type 2 diabetes and their healthcare providers.

Automatic Identification of Physical Activity Intensity and Modality from the Fusion of Accelerometry and Heart Rate Data.

BACKGROUND: Physical activity (PA) is essential to prevent and to treat a variety of chronic diseases. The automated detection and quantification of PA over time empowers lifestyle interventions, facilitating reliable exercise tracking and data-driven counseling. METHODS: We propose and compare various combinations of machine learning (ML) schemes for the automatic classification of PA from multi-modal data, simultaneously captured by a biaxial accelerometer and a heart rate (HR) monitor. Intensity levels (low / moderate / vigorous) were recognized, as well as for vigorous exercise, its modality (sustained aerobic / resistance / mixed). In total, 178.63 h of data about PA intensity (65.55 % low / 18.96 % moderate / 15.49 % vigorous) and 17.00 h about modality were collected in two experiments: one in free-living conditions, another in a fitness center under controlled protocols. The structure used for automatic classification comprised: a) definition of 42 time-domain signal features, b) dimensionality reduction, c) data clustering, and d) temporal filtering to exploit time redundancy by means of a Hidden Markov Model (HMM). Four dimensionality reduction techniques and four clustering algorithms were studied. In order to cope with class imbalance in the dataset, a custom performance metric was defined to aggregate recognition accuracy, precision and recall. RESULTS: The best scheme, which comprised a projection through Linear Discriminant Analysis (LDA) and k-means clustering, was evaluated in leave-one-subject-out cross-validation; notably outperforming the standard industry procedures for PA intensity classification: score 84.65 %, versus up to 63.60 %. Errors tended to be brief and to appear around transients. CONCLUSIONS: The application of ML techniques for pattern identification and temporal filtering allowed to merge accelerometry and HR data in a solid manner, and achieved markedly better recognition performances than the standard methods for PA intensity estimation.