Inside our Latest Hackathon
We recently wrapped up our latest Hackathon –– an event that brought together Oscar employees from various teams to brainstorm new AI use cases and explore how our tech can enhance the healthcare experience for our members, providers, and brokers.
With over 60 participants working on 16 different projects, the energy was palpable. In total, more than 100 individuals contributed in various ways — whether by participating in projects, hosting Office Hours, emceeing, judging, designing swag, providing logistical support for live streaming, or editing communications. This collective effort highlights the vibrant and supportive culture at Oscar, where many leaned in to ensure the Hackathon's success.
We had winners in several categories, and we're excited to share some of the incredible projects below. We look forward to continuing to iterate on these ideas and hopefully bringing them into production soon.
Category Winner: Engineering
Doctor Riage
Many of our processes here at Oscar are dependent on the receipt of documents. After the documents are received, they are ingested into a document queue to be triaged and handed off to the appropriate team. Doctor Riage’s team goal was to see how the use of AI can improve our accuracy and efficiency of doc triage, ultimately enhancing our downstream processes and the Member/Provider experience.
Oscar receives or creates an average of ~345,000 documents per day. Complex, cross-functional workflows are triggered depending on the kind of document received. If a document is triaged incorrectly and routed to an unrelated process, we can experience delays in processing, resulting in frustrated patients and providers.
So, what happens if we use AI to identify the types of documents we are receiving and creating? The team used OpenAI to categorize 200 documents, and assessed the outcomes against system defined document tags. Results were validated by manual review. The results were:
100% identification of Medical Records
100% identification of Provider Disputes
94% identification of Itemized Bills
These results demonstrated the powerful role that AI can play in sorting the many documents Oscar works with on a daily basis.
We currently store OCR (Optical Character Recognition) data from all documents we ingest in BigQuery. We used OCR’ed text content to ask GPT to identify what type of document it is.
Category Winner: Design
Health Autopilot
Being diagnosed with a chronic condition brings on anxiety & uncertainty around understanding what the future holds—from lifestyle changes to adjusting to new medications or treatments.
We know that we can help our members from the moment of diagnosis to receiving care by personalizing their experience with targeted content, setting expectations, and providing support—ultimately becoming their trusted partner throughout this new care journey.
Meet Health Autopilot: a personalized digital experience for members managing chronic conditions. Leveraging insights from a small sample of members diagnosed with Type 2 diabetes, the team created a resource hub with information that’s relevant, helpful and supportive for these members. With Health Autopilot, members can:
Learn about their diagnosis
Read frequently asked questions
Keep track of symptoms
Manage prescriptions
Access to relevant video content from Oscar Medical Group providers.
Members can find relevant information about their condition and care with Health AutoPilot.
Health Autopilot also provides an AI powered chat function, where members can ask questions about their condition. Because Oscar has information about each member’s allergies, medication, and profile, the recommendations are tailored to the member’s unique needs.
Health Autopilot can support members who have been diagnosed with a chronic illness by providing them with the information they need to manage their condition.
Category Winner: DE & I
Oscar SuperAgent
Healthcare has a knowledge problem – what if we could use AI to create a superagent that could answer any member question?
The Oscar Superagent team created a system that can aggregate the necessary information - both the content and data - to answer questions from members and the Care Team. The work builds on Oscar’s existing customer service automation work, pulling information from a much wider set of Oscar’s internal systems, public PDFs and guides, as well as Oscar’s FAQs page.
Members can ask the Superagent about their plan benefits, cost-sharing, and in-network care options.
Category Winner: Leadership choice
PrescriptionGenie
The PrescriptionGenie is designed to leverage AI to inform members if there may be lower-cost alternative medications. The primary goal of PrescriptionGenie is to make healthcare more affordable and accessible by providing cost-effective drug options without compromising on efficacy.
The team designed an embeddable web component that allows members to look up their current prescription medications and determine if there may be less expensive options. Members can select the name of their medication from a list of drugs they take, and PrescriptionGenie uses AI to find less expensive alternatives. Members can then discuss prescription options with their provider at their next visit.
With PrescriptionGenie, members select their medication from a personalized list to explore cost saving options.
PrescriptionGenie uses AI to generate less expensive alternatives.
This feature can also be leveraged by Oscar’s Care Guides, who can use PrescriptionGenie to inform members if the drugs they are taking have lower-cost alternatives or generics available. PrescriptionGenie has the ability to empower our members to discuss lower cost alternatives with their provider, and to create meaningful cost savings for them.
Category Winner: Overall winner
Internationalize Oscar app
The Oscar mobile app can be used in English or Spanish, depending on the member’s preference. The goal of this project was to make the Oscar mobile app more accessible to non-English and non-Spanish speakers, who currently rely on translation softwares to use the mobile app.
This project used GPT to convert Oscar’s app into five languages.
This project leveraged AI to make the Oscar mobile app more accessible by translating it into five languages, including Japanese and Mandarin.
Translating our app into more languages can provide a more seamless and accessible experience for our members.
Thank you to everyone who participated in our 2024 Hackathon! Let's continue to push boundaries, innovate, and make the next Hackathon even more spectacular.