Physical health status defines an individual’s ability to perform normal activities of daily living and is usually assessed in clinical settings by questionnaires and/or by validated tests, e.g. timed walk tests. These measurements have relatively low information content and are usually limited in frequency. Wearable sensors, such as activity monitors, enable remote measurement of parameters associated with physical activity but have not been widely explored beyond measurement of daily step count. Here we report on results from a cohort of 22 individuals with Pulmonary Arterial Hypertension (PAH) who were provided with a Fitbit activity monitor (Fitbit Charge HR®) between two clinic visits (18.4 ± 12.2 weeks). At each clinical visit, a maximum of 26 measurements were recorded (19 categorical and 7 continuous). From analysis of the minute-to-minute step rate and heart rate we derive several metrics associated with physical activity and cardiovascular function. These metrics are used to identify subgroups within the cohort and to compare to clinical parameters. Several Fitbit metrics are strongly correlated to continuous clinical parameters. Using a thresholding approach, we show that many Fitbit metrics result in statistically significant differences in clinical parameters between subgroups, including those associated with physical status, cardiovascular function, pulmonary function, as well as biomarkers from blood tests. These results highlight the fact that daily step count is only one of many metrics that can be derived from activity monitors.

To evaluate the performance of machine-learning models based on multiple years of continuous data to predict incident diabetes among patients with metabolic syndrome. Patients and Methods:The dataset comprises the health records from 2008 to 2020 including 4510 nondiabetic participants with metabolic syndrome (MetS) at baseline and with at least 6 years of records. MetS was defined according to the International Diabetes Federation (IDF) criteria. Overall, 332 patients developed incident diabetes during the 7±1.4 years of follow-up. Three popular classification algorithms were evaluated on the dataset: logistic regression, random forest, and Xgboost. Five models including singleyear models (year 1, year 2, and year 3) and multiple-year models (year 1-2 and year 1-3) were developed for each algorithm. Results:The model performances improved with the increasing longitudinal dataset as the area under the receiver operating characteristic curve (AUROC) was boosted for both random forest (year 1-3: AUROC=0.893; year 3: AUROC=0.862; year 1-2: AUROC=0.847; year 2: AUROC=0.838) and Xgboost (year 1-3: AUROC=0.897; year 3: AUROC=0.833; year 1-2: AUROC=0.856; year 2: AUROC=0.823) model. In the multiple-year models, the highest fasting plasma glucose, followed by the mean or lowest level of HbA1c and BMI had the most important predictive value for the onset of diabetes. In the "1-3" year model, "delta weight" which reflects the fluctuations of yearly change of weight was the fourth-most important feature. Conclusion:This study demonstrated improved performance with the accumulation of longitudinal data when using machine learning for diabetes prediction in MetS patients. For individuals with similar clinical parameters, the variation trends of these parameters could change the risk of future diabetes. This result indicated that models based on longitudinal multiple years' data may provide more personalized assessment tools for risk evaluation.