Selected work

Case notes.

What the problem was, what was built, and what changed. Client names are withheld.

Health system
Clinical AI · Prior role

−5%

patient mortality, and 10% fewer code blue events, against matched control sites

Catching patient deterioration earlier

Patients who are about to deteriorate often show it in vital signs and lab results before it becomes an emergency. An early-warning system surfaces those patients to clinical teams sooner.

  • Evaluated against matched control sites, not a before-and-after guess
  • Mortality fell 5% and code blue events fell 10% against those sites

Health system
Review automation · Prior role

4 h → 30 min

per medical-necessity review

Clinical review in minutes, not hours

Manual chart review is slow, expensive, and hard to staff. Two review processes that consumed hours of clinical time per case now take a fraction of it.

  • Medical-necessity review: from a four-hour manual process to 30 minutes per case
  • Surgical site infection review: from roughly two weeks per reviewer each month to a single day

Health system
AI strategy and governance · Prior role

4,000+

clinical and administrative staff reached by AI education

An enterprise AI program with governance built in

A large health system needed one place to set AI strategy for clinical, operational, and customer functions, and a way to govern it that its board and policymakers could follow. An enterprise AI Center of Excellence became that place.

  • An executive steering committee and an AI governance and ethics committee
  • An applied AI and machine learning team built out from a few people in separate departments
  • AI education that reached more than 4,000 clinical and administrative staff

Health plan
Predictive outreach · Prior role

20%+

fewer preventable emergency room visits among members reached

Outreach that moved health outcomes

A health plan's marketing analytics moved off manual Oracle-based processes onto an automated cloud platform, and predictive models began matching each member to the programs most relevant to them.

  • A marketing data warehouse and MLOps framework in Azure Databricks, driving more than 50 next-best-action campaigns
  • Preventable ER visits down more than 20%; flu and COVID-19 vaccination rates up more than 30%
  • A new maternal health program where gestational diabetes and preeclampsia fell more than 15% among participants
  • Delivery time cut by more than half after the move to the cloud platform

Addiction treatment network
Outcomes data · Client engagement

Network-wide

outcome and recidivism metrics

Measuring what happens after discharge

A network of addiction treatment centers needed a consistent way to measure patient outcomes, including what happened after discharge. We built the outcomes-data program and defined the operational metrics the network used to track patient outcomes and recidivism.

Case notes marked Prior role come from programs our founder led in previous leadership roles at Corewell Health and Priority Health. They show the kind of work we do; they aren't presented as Freshwater Coast Analytics client engagements.

Contact

Start with a conversation.

Tell us what you're trying to decide. You'll get a plain answer on whether data or AI can help, and what it would take.

What happens next