Darrell A. Hall MD, MSCIS

Clinical informatics · Data engineering · Toledo, Ohio

Physician who builds the software clinicians use.

Residency-trained family physician and certified data scientist. I design programs and the systems that measure them: outcome scoring, de-identified clinical records, geospatial risk models, and the infrastructure to run them.

See the work View the PMSG code
Clinical practice
14 years
Networks & systems
Since 1984
Data science
Certified 2022
Stack
Python · R · SQL · GCP

Featured case study

github.com/dah512/pmsg-platform →

An 8-week, non-pharmacologic program pairing pain neuroscience education, CBT, anti-inflammatory nutrition, and Qi Gong and Tai Chi. I designed the curriculum and the platform that tracks whether it works, using two validated instruments: the Brief Pain Inventory and the Oswestry Disability Index.

  • 8weeks
  • 32sessions
  • 80contact hours

What's in the repo

  • TESTEDFastAPI outcomes service: BPI-SF and ODI scoring with published missing-item rules, MCID change scores, Postgres history
  • WORKINGDocker Compose stack: Postgres, API, nginx serving the program site and curriculum, optional R Shiny analytics
  • WORKINGReact program site and Shiny dashboard: trajectories, paired t-test, effect size
  • IN PROGRESSR Bookdown technical manual; auth and scheduling on the roadmap

All participant data in the repository is synthetic. The dashboard demonstrates analysis on simulated trajectories, not clinical outcomes.

POST/score/bpi
{
  "worst": 8, "least": 2,
  "average": 5, "now": 5,
  "mood": 4, "sleep": 5,
  "walking": 7, "enjoyment": 5
}
200 response
{
  "severity": 5.0,
  "interference": 5.25,
  "interference_items_answered": 4
}
Interference is scored once 4 of 7 items are answered, per the BPI user guide. Severity requires all four.
FastAPI PostgreSQL 16 SQLAlchemy 2 pytest · GitHub Actions Docker Compose React · Tailwind R Shiny · Bookdown

More work

Systems across care settings

KATA

DEPLOYED

Behavioral health record and safety workflow for youth movement programs, built around Ohio's mandated-reporting statutes.

  • Six-role RBAC; de-identified analyst view with pgcrypto
  • Alembic migrations, 51 passing tests, CI
  • Cloud Run, Cloud SQL, Firebase Hosting

FastAPI · Postgres · React/TS · RAG

NABS

PROTOTYPE

Non-Ambulatory Bathing System: full-body hygiene for patients who can't be moved, with thermal sensing for skin surveillance.

  • FLIR Lepton 3.5 on Raspberry Pi 5
  • ResNet-50 staging classifier, Kafka streaming
  • Interactive demo and engineering drawing register

Interactive demoTechnical reportnabs.edmhs.us

Computer vision · Edge · Streaming

GeoCode Analytics

ACTIVE

Census-tract mapping of substance use risk across Lucas County, designed to take wearable telemetry alongside location.

  • 7,165 geocoded Toledo addresses
  • Eight-sector directional analysis
  • tidygeocoder and Leaflet

R · Spatial · Public health

Overdose mortality study

IN DESIGN

National county-level analysis of the 2022–24 overdose reversal, planned as a book with peer-reviewed papers beneath it.

  • Bayesian BYM/CAR models in INLA
  • Handles CDC WONDER small-count suppression
  • County panel plus document RAG corpus

Bayesian spatiotemporal · Epidemiology

VOX

DEPLOYED

Speech service shared by PMSG and KATA. Dictated notes land as unsigned drafts a clinician reviews and signs.

  • Whisper speech-to-text, Piper text-to-speech
  • Standalone service on Cloud Run
  • Staff-only Google sign-in

Speech · Services · GCP

Everything is on GitHub

Code, tests and READMEs that say plainly what's built and what's next.

github.com/dah512 →

Background

Networks first, then medicine, then both

I started in Air Force communications during the move from analog message switching to packet-switched networks, went to medical school, ran a multi-site practice whose IT I built myself, and came back to data science to put the two together.

  1. 1984–1991 USAF communications-electronics officer, EUCOM. Defense Data Network planning as AUTODIN gave way to packet switching.
  2. 1989 MS, Computer Information Systems, Boston University. Database systems and early rule-based AI.
  3. MD Wright State University (Boonshoft School of Medicine), then family medicine residency at the Medical College of Ohio.
  4. 2003–2012 Founded Visiting Physician Service. Two clinics and a mobile unit; implemented the EHR, networking and data integration across sites.
  5. 2022 Certified Professional Data Scientist. Data engineering and machine learning scientist tracks in Python and R.
  6. Now Founder and CMIO, Empowered Disease Management Health Services. Building the systems on this page.

Clinical data work that has to hold up

Open to AI/ML engineering, data architecture and clinical data science roles, and to consulting engagements.

dah512@edmhs.us github.com/dah512