Windfall serves susceptible and deprived communities by way of compassionate, high-quality care. As one of many largest nonprofit well being programs in america—with 51 hospitals, over 1,000 outpatient clinics, and greater than 130,000 caregivers throughout seven states—our capacity to ship well timed, coordinated care depends upon reworking not solely medical outcomes but in addition the workflows that assist them.
Probably the most urgent situations is automating the best way we deal with faxes. Regardless of advances in digital well being, faxes stay a dominant type of communication in healthcare, particularly for referrals between suppliers. Windfall receives greater than 40 million faxes yearly, totaling over 160 million pages. A good portion of that quantity have to be manually reviewed and transcribed into Epic, our digital well being document (EHR) system.
The method is gradual, error-prone and contributes to multi-month backlogs that in the end delay take care of sufferers. We knew there needed to be a greater manner.
Tackling messy workflows and unstructured knowledge at scale
The core problem wasn’t simply technical—it was human. In healthcare, workflows differ broadly between clinics, roles and even people. One workers member would possibly print and scan referrals earlier than manually coming into them into Epic, whereas one other would possibly work inside a wholly digital queue. The dearth of standardization makes it troublesome to outline a “common” automation pipeline or create take a look at situations that replicate real-world complexity.
On prime of that, the underlying knowledge is usually fragmented and inconsistently saved. From handwritten notes to typed PDFs, the range of incoming fax paperwork creates a variety of inputs to course of, classify and extract data from. And while you’re coping with a number of optical character recognition (OCR) instruments, immediate methods and language fashions, tuning all these hyperparameters turns into exponentially more durable.
This complexity made it clear that our success would hinge on constructing a low-friction testing ecosystem. One which lets us experiment quickly, examine outcomes throughout 1000’s of permutations and repeatedly refine our fashions and prompts.
Accelerating GenAI experimentation with MLflow on Databricks
To satisfy that problem, we turned to the Databricks Information Intelligence Platform, and particularly MLflow, to orchestrate and scale our machine studying mannequin experimentation pipeline. Whereas our manufacturing infrastructure is constructed on microservices, the experimentation and validation phases are powered by Databricks, which is the place a lot of the worth lies.
For our eFax venture, we used MLflow to:
- Outline and execute parameterized jobs that sweep throughout combos of OCR fashions, immediate templates and different hyperparameters. By permitting customers to offer dynamic inputs at runtime, parameterized jobs make duties extra versatile and reusable. We handle jobs by way of our CI/CD pipelines, producing YAML recordsdata to configure giant assessments effectively and repeatably.
- Observe and log experiment outcomes centrally for environment friendly comparability. This offers our workforce clear visibility into what’s working and what wants tuning, with out duplicating effort. The central logging additionally helps deeper analysis of mannequin habits throughout doc sorts and referral situations.
- Leverage historic knowledge to simulate downstream outcomes and refine our fashions earlier than pushing to manufacturing. Catching points early within the testing cycle reduces threat and accelerates deployment. That is notably essential given the range of referral types and the necessity for compliance inside closely regulated EHR environments like Epic.
This course of was impressed by our success working with Databricks on our deep studying frameworks. We’ve since tailored and expanded it for our eFax work and enormous language mannequin (LLM) experimentation.
Whereas we use Azure AI Doc Intelligence for OCR and OpenAI’s GPT-4.0 fashions for extraction, the actual engineering accelerant has been the power to run managed, repeated assessments by way of MLflow pipelines—automating what would in any other case be guide, fragmented growth. With the unifying nature of the Databricks Information Intelligence Platform, we’re capable of remodel uncooked faxes, experiment with completely different AI strategies and validate outputs with pace and confidence in a single place.
All extracted referral knowledge have to be built-in into Epic, which requires seamless knowledge formatting, validation and safe supply. Databricks performs a important function in pre-processing and normalizing this data earlier than handoff to our EHR system.
We additionally depend on Databricks for batch ETL, metadata storage and downstream evaluation. Our broader tech stack consists of Azure Kubernetes Service (AKS) for containerized deployment, Azure Search to assist retrieval-augmented era (RAG) workflows and Postgres for structured storage. For future phases, we’re actively exploring Mosaic AI for RAG and Mannequin Serving to boost the accuracy, scalability and responsiveness of our AI options. With Mannequin Serving, we will likely be in a greater place to successfully deploy and handle fashions in actual time, guaranteeing extra constant workflows throughout all our AI efforts.
From months of backlog to real-time triage
Finally, the beneficiaries of this eFax answer are our caregivers—clinicians, medical data directors, nurses, and different frontline workers whose time is presently consumed by repetitive doc processing. By eradicating low-value guide bottlenecks, we goal to return that point to affected person care.
In some areas, faxes have sat in queues for as much as two to 3 months with out being reviewed—delays that may severely affect affected person care. With AI-powered automation, we’re shifting towards real-time processing of over 40 million faxes yearly, eliminating bottlenecks and enabling sooner referral consumption. This shift has not solely improved productiveness and diminished operational overhead but in addition accelerated remedy timelines, enhanced affected person outcomes, and freed up medical workers to give attention to higher-value care supply. By modernizing a traditionally guide workflow, we’re unlocking system-wide efficiencies that scale throughout our 1,000+ outpatient clinics, supporting our mission to offer well timed, coordinated care at scale.
Due to MLflow, we’re not simply experimenting. We’re operationalizing AI in a manner that’s aligned with our mission, our workflows, and the real-time wants of our caregivers and sufferers.
