Page 18 - AIH-1-4

Artificial Intelligence in Health

PERSPECTIVE ARTICLE
Artificial intelligence scribe: A new era in medical

documentation

Khalid Nawab*
Department of Internal Medicine, Penn State Holy Spirit Medical Center, Camp Hill, Pennsylvania,
United States of America

Abstract
The high workloads involved in clinical documentation represent one of the major
factors contributing to the significant escalation of clinician burnout. The emergence
of artificial intelligence (AI) has provided new avenues for relieving this burden by
automating certain tasks like clinical documentation through the generation of
clinical notes from a transcript of a clinical encounter. The advances in large language
models (LLMs) have led to the emergence of such startups, but they come with their
own set of challenges, predominantly surrounding the concerns of documentation
accuracy, completeness, and data security. These can be addressed with a multi-
faceted approach which could include fine-tuning the currently available models;
using domain-specific models and in-house AI systems to ensure data security; and
involving smaller LLMs and clinicians in the development and implementation of
such systems. We can imagine a future where these systems are deeply incorporated
into electronic health records, providing not only automated clinical documentation
but also improving Clinical Decision Support systems, research, and patient
communication.
*Corresponding author:
Khalid Nawab
(knawab@pennstatehealth.psu.edu) Keywords: Artificial intelligence; Large language models; Clinical documentation;
Citation: Nawab K. Artificial Automation; Clinician burnout
intelligence scribe: A new era in
medical documentation. Artif Intell
Health. 2024;1(4):12-15.
doi: 10.36922/aih.3103
1. Introduction
Received: March 6, 2024
The American Medical Association reports that in the United States of America,
Accepted: June 19, 2024
physician burnout is an epidemic with about 63% of physicians reporting signs of
Published Online: September 27, burnout at least once per week. Clinical documentation using electronic health record
1
2024
(EHR) is perceived as a significant contributor to clinicians’ burnout mostly due to poor
Copyright: © 2024 Author(s). usability and excessive time spent on EHRs. 2
This is an Open-Access article
distributed under the terms of the Artificial intelligence (AI) has emerged as a potential solution to various tasks
Creative Commons Attribution including documentation in healthcare. The idea can be traced back to 2017, with
License, permitting distribution,
3
and reproduction in any medium, “DeepScribe” being one of the earliest companies offering such a service. However,
provided the original work is increased adoption likely happened after the attention was drawn to AI by ChatGPT,
properly cited. a publicly available online application that is optimized for human-like conversation.
4
Publisher’s Note: AccScience Access to such powerful models through an application programming interface
Publishing remains neutral with (API) opened new venues to easily incorporate natural language processing and AI in
regard to jurisdictional claims in
published maps and institutional healthcare. An AI-based scribe application, incorporating speech-to-text transcription,
affiliations. and then using that to generate a clinical summary or other forms of notes, sounds

Volume 1 Issue 4 (2024) 12 doi: 10.36922/aih.3103