In this study, adults with pre-diabetes will be prospectively enrolled for data collection to design prediction models that integrate electronic health record data and patient-generated activity data. Patients will be randomized to receive either a waist-worn or wrist-worn wearable device for 6 months to capture patient-generated activity data.

Patients with suboptimal glycemic control could be better managed if these higher risk patients could be identified and effective interventions were then targeted towards them. However, most practice settings perform infrequent laboratory testing every 3 to 6 months, if not at longer intervals. Current models to predict change in glycemic control perform poorly and do not take into account the behaviors that occur between these intervals. In this study, we will compare different methods to use data on daily health behaviors collected by wearable devices to enhance risk prediction models. Adults with pre-diabetes will complete a series of surveys and baseline assessments and then will be randomly assigned to use a waist-worn or wrist-worn wearable device for 6 months. Measures of HbA1c and LDL will be obtained at baseline and at 6 months.

Inclusion Criteria:

1. Be 18 years or older
2. Be able to provide informed consent
3. Have a smartphone or tablet compatible with the wearable device smartphone application and be willing to use the wearable device for 6 months
4. Baseline hemoglobin A1c of 5.7 to 6.4

Exclusion Criteria:

Participants will no be eligible if they have any medical condition or other reason that will likely prohibit them from completing the 6-month study

The participants will be aware of the two different study arms and the wearable devices used in both. The study investigator will not have not have knowledge of the participants assigned to each arm.