US Department of Veteran Affairs x Stanford x Andover
It's important to consider more than just individual data and how they differ from the "norm" when developing certain treatment strategies. For example, the Department of Veteran Affairs will calculate the risk of certain veterans of opioid overdose or depression—by knowing these they can preemptively provide more resources or better help, but they are not the most accurate because they aren't personalized.
- economic distress
- US census bureau
- social capital and connections
- lethal means / access to firearms (FBI)
- healthcare quality and access (CMS, substance abuse, and so on)
- national mental health services survey (2018) -> veterans?
- drug use and health can be interesting (really robust data!!)
- environmental determinants of health (noise, pollution, UV)
- modeling / outcome
curated -> estimate and standard deviations + numbers, so there are some sense of how reliable they are
- Okra dataset project (?)
- targeted prevention programs -> targeted outreach to those at risk based on models
- more likely to engage, less ER, less suicide attempts
- also for those using opioids and at risk of overdose
- these models only use patient medical record info (treating patients as if they are at equal risk regardless of their environment, context, and so on)
- first efforts are using modeling
- deep dive into how some of these community factors interplay
- visualization tools
- understanding health outcome data (i.e. death, census…)
- 10% of us adult pop is veterans, shows how important this work is
- longitudinal data!!
- CEPC / forward facing project
- POLICY PAPER WRITING (cool) vs. academic
Preview into some of the possible datasets: