03: Making the Healthcare System Usable

How we assemble the components of a useful care delivery system.

4 AI use cases

Construct Effective Networks

Find and recruit the most valuable physicians and health systems for Oscar

Get the most cost-efficient and high-quality doctors to sign up for our network.

Align with the right health systems in focused networks at the right unit costs, for the right value.

Manage the regulatory aspects of having providers in our network

Make our network perform well, by describing provider quality so well that members consistently seek out the highest-quality providers.

Make provider profiles more useful and interesting by enriching them with operating hours, deeper care specialization and other meaningful insights.

Status: Idea / Exploration
After our members have visits with providers in our network, we follow up with them to ask them for feedback. We use this feedback to evaluate providers in our network and ensure we’re maintaining access to doctors who meet our standards for both cost and quality. With LLMs, it’s possible to synthesize written member reviews into key insights & structured provider attributes that don’t otherwise show up in the formal provider rosters we receive.

Compile and present provider data like contact information and specialty with highest accuracy.

Status: Pilot / Prototype
Health plans spend a lot of energy maintaining up-to-date information on the providers in their network. At Oscar, we do this in a variety of ways and get signals from a variety of sources. We’ve prototyped automating intake for parts of this pipeline to streamline the processing of externally reported or internally identified provider data issues.

Enable providers to get paid fairly for the value they deliver.

Support providers with their questions around delivering care to our members.

Status: Pilot / Prototype
When providers call us with questions, our first priority is finding answers quickly. But after each call, our agents spend time summarizing their interactions in a set of notes that can be reviewed later both for quality improvement purposes and as part of the provider’s profile. This last step is ripe for automation through a combination of automated transcription and LLM-powered summarization—our prototyping has confirmed that this is a viable and high value use case.

Make it easy to understand how we paid providers.

Status: Launched / Coming soon
One of the core functions of a health plan is to process claims—these are basically invoices from providers that tell us what services were performed and from this we determine how much to pay. It sounds simple, but the adjudication process is incredibly complex. Even simple claims can invoke thousands of contractual, regulatory, plan-level, and system-level requirements. We’re using LLMs to translate raw adjudication tracing details into a plain English play-by-play to dramatically improve the way we assist providers.
Learn more about our new Oscar Claim Assistant Built On GPT-4

Explain to providers how to engage with our systems, rules and reimbursements.

Get providers paid.