Daily rhythm and metabolic syndrome: a study of 272 adults

Metabolic syndrome combines abdominal obesity, insulin resistance, high blood pressure, high triglycerides and low HDL. It is associated with a higher risk of type 2 diabetes and cardiovascular disease.

Explainable Ai (XAI): the model explains its results

Unlike many Ai models, XAI does not just give a result: it shows which variables contributed most, in which direction and with what weight. In medicine that matters, because whoever reads a result needs to understand why the model says what it says. This study used three different methods: SHAP, an explainable boosting machine and a neural network for tabular data.

Across all the models, the most important marker was a new one proposed by the authors: circadian rhythm energy derived from heart rate (CCE), which was lower in people with metabolic syndrome. Other heart-rate markers, such as relative amplitude and the low-activity period, also mattered. Sleep markers did not reach statistical significance and appeared only as secondary factors. CCE kept its weight after adjusting for age, sex and BMI. It is a cross-sectional study: it describes an association in a group, not a personal risk.

Heart rate and steps recorded by a watch or a wristband are the same kinds of data used in this study, but the study's circadian indicators are not a feature of the app. The Digital Twin lines up the data your devices write to your phone, such as steps, heart rate and sleep, and shows you how they change over time. It does not estimate your risk of metabolic syndrome and it does not diagnose: this research describes groups of people, not you. Reading your values is for your doctor.