Automating Messaging

Project overview

Can current significant language model technologies, such as GPT, create high-quality responses across the diverse spectrum of communication typically found in patient-provider interactions? We believe these advanced AI models have the potential to enhance the quality and efficiency of message responses significantly.

Current methods

We've developed an anonymous survey targeting Penn primary care providers to validate our hypothesis. This survey is a critical tool in assessing the performance of our initial machine-learning model. Participants in this study will review a series of messages, some written by real patients and others generated by our Language Learning Model (LLM). The objective is to discern which messages are AI-generated and which are from actual patients. The insights gathered from this study will be instrumental in shaping a more extensive trial and further research in this domain.

The goal

Our ultimate goal is to leverage AI in a way that not only streamlines patient-provider communication but also reduces the strain on healthcare providers, thus improving the overall healthcare experience for both parties.