AI @ Oscar
As a tech-driven health insurer, we have a unique view and approach to AI. In general, research on AI is moving unpredictably fast, and there are new applications developed every day. It is our belief that the best approach to this moment is through experimentation, learning, and sharing. We’re continually putting into development promising experiments and are committed to doing so in a way that’s responsible, in compliance with regulatory requirements, and ensures the best experience for our members. This is an approach we’re calling a continuous hackathon.
Continuous hackathon at Oscar
In the spirit of constantly learning, we are experimenting with several AI use cases; here are a few examples of ideas we’re working on*.Automated Claims System Configuration
One of the key inputs into the processing of a health insurance claim is the contractual language that was agreed on between payor and provider. These agreements are often non-standardized, and even small discrepancies in how they’re configured can delay payment and cause headaches for patients. With AI, we’re exploring ways to automate the translation of provider contracts into highly accurate claims configurations.
Campaign Builder Actions
GPT enables new types of automation through Campaign Builder. For example, we can monitor inbound member inquiries from a variety of data sources to look for subtle indications that the member might benefit from enrollment in a maternity program. This allows Oscar and +Oscar clients to deliver relevant interventions and intelligently monitor for signals to better serve members’ and patients’ clinical needs.
Claim Assistant
A bot is trained to understand system logs and PDFs, breaking down the cost of procedures and identifying how insurance logic was applied in each case. This allows our operations team to use the claims assistant to rapidly ask questions about how a claim was priced and troubleshoot issues for members and care providers.
Clinical Documentation
An AI-generated clinical documentation system that automatically summarizes and documents conversations between our medical team and patients. This decreases clinical burnout generated by documentation needs and increases valuable time between patient and clinician.
Our thoughts, our approach, and our entire documentation of notes are publicly available here and will be constantly updated as we continue to learn, experiment and implement.