Page 106 - AIH-2-4

Artificial Intelligence in Health AI vs humans in clinical code conversion

and greater processing speed than their free versions. Investigation: Carly Hudson, Marcus Randall, Adrian
This study could be replicated using the free versions of Goldsworthy
these tools to compare whether the paid versions yield any Methodology: Carly Hudson, Marcus Randall, Adrian
difference in terms of level of agreement and processing Goldsworthy
time. It is also yet to be determined whether the time and Project administration: Carly Hudson
cost savings observed in this task would translate to other Resources: Candice Bowman
data conversion tasks. Further studies using GenAI tools Supervision: Marcus Randall, Candice Bowman
are needed to determine whether time and cost differences Writing – original draft: Carly Hudson, Marcus Randall,
are consistent across different types of tasks. Additionally, Adrian Goldsworthy
as new GenAI tools—such as DeepSeek —are released Writing – review & editing: All authors
37
with improvements in speed and functionality, it is
recommended that this study be repeated to examine how Ethics approval and consent to participate
these improvements impact the speed and accuracy with This research was approved by the Human Research Ethics
which this task can be completed. Although ChatGPT-4o Committee of the Gold Coast Hospital and Health Service
and Claude 3.5 Sonnet are not specifically designed for (HREC/2023/QGC/95219).
healthcare applications, these tools were selected due to
their relatively low cost and wide accessibility. The task Consent for publication
presented in this study should also be repeated using Not applicable.
GenAI tools specifically designed for clinical or healthcare
contexts. Furthermore, the completion of similar tasks Availability of data
using GenAI tools should be considered to further explore
their capabilities in healthcare data processing. Data is available from the corresponding author upon
reasonable request.
5. Conclusion Further disclosure
This study presents a case study demonstrating the use of
GenAI tools to complete manual data processing tasks that This research has been preprinted on Research Square
are typically tedious, time-consuming, costly, and both (https://doi.org/10.21203/rs.3.rs-5143761/v1).
mentally and physically demanding. The findings highlight References
that manual processing is often prohibitive in terms of time
and cost and that alternative methods – such as the use 1. Murphy K. How data will improve healthcare without
of GenAI – warrant further exploration. GenAI provides adding staff or beds. Cornell university, institut européen
a promising avenue for leveraging the vast quantities of d’administration des affaires. In: Organisation WIP, editors.
currently unanalyzed healthcare data, with the potential Global Innovation Index 2019. Ch. 8. Geneva: World
Intellectual Property Organization; 2019.
to enhance outcomes for healthcare professionals,
researchers, systems, and, most importantly, patients. 2. Dash S, Shakyawar SK, Sharma M, Kaushik S. Big data in
healthcare: Management, analysis and future prospects. J Big
Acknowledgments Data. 2019;6(1):54.
None. doi: 10.1186/s40537-019-0217-0
3. Australian Medical Association. 2024 Public Hopsital Report
Funding Card; 2024.
This study was supported by an Australian Government 4. Feuerriegel S, Hartmann J, Janiesch C, Zschech P. Generative
Research Training Program Scholarship. AI. Bus Inform Syst Eng. 2024;66(1):111-126.
Conflict of interest doi: 10.1007/s12599-023-00834-7
5. Oluwagbenro MB. Generative AI: Definition, Concepts,
The authors declare that they have no competing interests.
Applications, and Future Prospects. Authorea Preprints; 2024.
Author contributions 6. Banh L, Strobel G. Generative artificial intelligence. Electron
Mark. 2023;33(1):63.
Conceptualization: Carly Hudson
Data curation: Carly Hudson, Anu Joy, Adrian Goldsworthy doi: 10.1007/s12525-023-00680-1
Formal analysis: Carly Hudson, Marcus Randall, Adrian 7. Fui-Hoon Nah F, Zheng R, Cai J, Siau K, Chen L.
Goldsworthy Generative AI and chatGPT: Applications, challenges, and

Volume 2 Issue 4 (2025) 100 doi: 10.36922/AIH025200045