Mobile Data by Orson Xu
Add full notes here... appear to have lost them. Great lecture.
Patient-Generated Health Data is (a) created, recorded, and gathered (b) by or from patients (c) or their family members or caregivers. We use this data to address a medical/health-related concern. Your Apple Watch (or the Android equivalent) provides plenty of commonsensical examples of this data: heart rate, when and how long you walked or exercised, your calories if you choose to track them, your engagement with mindfulness, sleep patterns, and so on. It's very different than your EHR data in where (patient's environment) and how (continuous, longitudinal, multi-modal) it's captured. The smart-watch remains a great source of this kind of data.
Discussion on the history of PGHD and the renewed modern drive for patient-empowered healthcare in the context of ubiquitous computing. All manner of stakeholders you can guess commonsensically (patients, clinicians, researchers, money people, epidemiologists, etc)
Discussion on Active and Passive methods of collecting data. Active methods involve questions and feedback that patient must enter (logging things on your Apple Watch). Passive methods just collect data in the background (you can ask for a weekly report of your heartrate on your watch). There are important user-centered considerations and up/downsides of each.
Finally, a discussion on what to actually do with this data. The high-level steps are (1) Cleaning it and prepping it for (2) Fusing it with other relevant data so you can (3) Create meaningful variables to (4) Create models and find associations and patterns. Discussion of the biggest problem: Model validation. E.g. What exactly is "stress" and how does my model know to tell me? Am I "stressed" when running (high HR) or if my blood glucose is high or if I had poor-quality sleep?