1. Be/RLHF
                      \
   2. Do/Langugage -> 4. Be/Compute -> 5. Do/Training -> 6. Hide/Environment
                      /
                       3. Hide/Agent

1. Personalized medicine
   • Literature has thus far been limited (as if older adults are homogenious)
   • Patient or Case-mix dependent inference (individual consent derived from single parameter)
   • Sort of Public Health style (clinical trial literature is no different)
2. WebApp
   • Demographics (age, etc)
   • Exam (frailty phenotype)
   • Labs (inflammatory markers)
3. Deployment
   • Not a onetime thing (i.e., not truth, frequentist)
   • Iterations (i.e., freq updates, bayesian)
   • Neural networks capable of RLHF (credibility intervals)
4. RDC/NCHS Access Tokens
   • Compute: e(b), e(V) sizes
   • Iterations
   • Updates
5. Vince
6. GitHub
   • Transparency
   • Free
   • .js & .html vs. strimlit