Page 98 - AIH-2-4

Artificial Intelligence in Health

ORIGINAL RESEARCH ARTICLE
Artificial intelligence versus humans: A comparative

analysis of time, cost, and performance on a clinical
code conversion task

Carly Hudson 1,2,3 * , Marcus Randall 2 , Candice Bowman 1,4 , Anu Joy 4,5 ,
and Adrian Goldsworthy 1,6,7
1 Faculty of Health Sciences and Medicine, Bond University, Gold Coast, Queensland, Australia
2 Bond Business School, Bond University, Gold Coast, Queensland, Australia
3 Faculty of Medicine and Health, University of New England, Armidale, New South Wales, Australia
4 Mental Health and Specialist Services, Gold Coast Hospital and Health Service, Gold Coast,
Queensland, Australia
5 School of Applied Psychology, Griffith University, Brisbane, Queensland, Australia
6 Wesley Research Institute, Brisbane, Queensland, Australia
7 Critical Care Research Group, The Prince Charles Hospital, Brisbane, Queensland, Australia

Abstract
Healthcare services generate and store large quantities of data, requiring significant
*Corresponding author: resources to manually analyze and gain meaningful insights. Recent advancements
Carly Hudson in automation tools—such as generative artificial intelligence (GenAI)—provide new
(chudson@bond.edu.au) opportunities to reduce human labor. This study explores the potential utilization of
Citation: Hudson C, Randall M, GenAI for a healthcare data analysis task—specifically, the conversion of clinical data
Bowman C, Joy A, Goldsworthy A. from one diagnostic classification system to another (i.e., the Australian extension
Artificial intelligence versus
humans: A comparative analysis of of the Systematized Nomenclature of Medicine Clinical Terms to the International
th
time, cost, and performance on a Classification of Diseases, 10 Revision, Clinical Modification)—and examines the
clinical code conversion task. Artif time and cost benefits of performing this using GenAI compared to a human rater.
Intell Health. 2025;2(4):92-102.
doi: 10.36922/AIH025200045 Conversions were completed using three methods: manual conversion using the
National Library of Medicine’s I-MAGIC tool, ChatGPT-4o, and Claude 3.5 Sonnet. The
Received: May 12, 2025 accuracy of the GenAI tools was mapped against the manually extracted codes and
Revised: June 9, 2025 examined in terms of a perfect, partial, or incorrect match. Task completion time was
Accepted: June 18, 2025 recorded and extrapolated to calculate and compare the cost associated with each
method. When compared to the manually extracted codes, Claude 3.5 Sonnet yielded
Published online: July 11, 2025 the highest level of agreement over ChatGPT-4o, whilst being the most time- and
Copyright: © 2025 Author(s). cost-effective. GenAI tools have greater utility than they have currently been given
This is an Open-Access article credit for. The automation of big data healthcare analytics, whilst still the domain of
distributed under the terms of the
Creative Commons Attribution humans, is increasingly capable of being undertaken using automation tools with
License, permitting distribution, low barriers to entry. The further development of GenAI’s capabilities, alongside the
and reproduction in any medium, capability of the healthcare system to use it appropriately, has the potential to result
provided the original work is
properly cited. in significant resource savings.
Publisher’s Note: AccScience
Publishing remains neutral with Keywords: Data analytics; Diagnostic coding; Generative artificial intelligence;
regard to jurisdictional claims in th
published maps and institutional International Classification of Diseases 10 revision; Systematized Nomenclature of
affiliations. Medicine Clinical Terms; SNOMED

Volume 2 Issue 4 (2025) 92 doi: 10.36922/AIH025200045