All of Us Research Program

Preparation

Evidence quality

Where it comes from

Highlighted claims

• The final analytic sample contained 205 participants with 3 to 11 repeated measurements each. — Precision Physical Activity Prescription via Reinforcement Learning for Functional Actions
• The analytic dataset came from the All of Us Registered Tier Dataset v8 and included Fitbit-derived daily step counts. — Precision Physical Activity Prescription via Reinforcement Learning for Functional Actions
• The study uses All of Us Research Program data linking Fitbit step counts with health records and biospecimens. — Precision Physical Activity Prescription via Reinforcement Learning for Functional Actions
• The study used an 11-interval, 990-day observation window beginning at each participant's first glucose measurement. — Precision Physical Activity Prescription via Reinforcement Learning for Functional Actions