Ambient Artificial Intelligence Scribes in Oncology: Adoption, Feasibility, Acceptability, and Appropriateness

Abstract

Hospitals are looking to AI and other innovative applications to help alleviate provider burden and dissatisfaction associated with clinical documentation in oncology. Ambient artificial intelligence (AI) scribes are a promising technology to address these issues. However, they generally have not been optimized for oncology. This study aimed to evaluate an ambient AI scribe application with oncology providers to determine opportunities and potential challenges.This prospective pilot study of a scribe application was conducted over 4 months at a high-volume cancer center in New York City. Qualitative (interviews) and quantitative (surveys and utilization) data were collected to assess adoption, feasibility, acceptability, and appropriateness. The analysis included descriptive statistics and thematic content analysis.Thirty-one providers were included across oncology specialties. Twenty-five providers used the application at least once; of these, 18 completed a survey and 21 completed an interview. Providers used the application in 620 (13.9%) out of 4,449 in-person outpatient visits. Out of 18 survey respondents, 17 (94%) indicated they used the AI-drafted content at least sometimes, demonstrating feasibility. For acceptability, 11 (61%) indicated a moderate, strong, or very strong desire for continued access to the technology. All providers interviewed advocated for continued investment in ambient technology. Metrics around appropriateness showed variability based on its accuracy in capturing complex clinical scenarios and in the types of patients the technology was used with. For example, providers used the technology for 21.1% of new visits but only 12.2% of follow-up visits.This study demonstrated the potential for ambient AI scribes to be useful in oncology. Future research should evaluate the use of this technology at scale as it may realize workflow efficiencies and improve the clinical documentation process.

Ähnliche Arbeiten