Grant: Project Grant

Palliative care improves quality of life for people with serious illness by expertly managing physical symptoms and establishing supportive relationships. Yet, in busy healthcare settings, synthesizing information from complex electronic health records (EHRs) gets in the way of this this process. Canadian clinicians often juggle multiple platforms to access vital information scattered across clinical notes, radiology and lab reports, and prescriptions. Additionally, patients with serious illnesses interact frequently with diverse healthcare providers, resulting in a substantial volume of difficult-to-synthesize health records. Synthesizing this information is time-consuming and inefficient. It can also hinder patient assessment and relationship building and threaten the sustainable practice of an increasingly vital workforce. High-quality, accurate EHR summaries may allow for more efficient and meaningful patient interactions and better care. Advances in computer science have led to the development of large language models, a form of artificial intelligence (AI) that can swiftly process and summarize large quantities of text data. Our initial findings indicate that these models can succinctly and seemingly accurately summarize dispersed medical record fragments. This may enable clinicians to quickly assimilate information, affording more or more efficient time with patients and enriching visit quality. We propose a two-phase study in which we will first validate AI-generated EHR summaries for implementation in the outpatient palliative care setting. We will then evaluate the implementation and effect of these summaries in a hospital-based palliative care clinic while assessing the feasibility of our trial procedures. We will use our pilot trial data to determine the relevance and design of a larger trial to examine the effects of AI-generated EHR summaries on important patient-centered outcomes, including therapeutic alliance and communication and care quality.