Harnessing the Potential of LLMs in Hospital Management: Insights Into Medical Data Inquiry
4 affiliations
- 1Department of Clinical Laboratory Medicine Jinniu Maternity and Child Health Hospital of Chengdu Chengdu China
- 2School of Medicine University of Electronic Science and Technology of China Chengdu China
- 3Information Center West China Hospital Sichuan University Chengdu China
- 4State Key Laboratory of Oral Diseases & National Center for Stomatology & National Clinical Research Center for Oral Diseases & Department of Information Management West China Hospital of Stomatology Sichuan University Chengdu Sichuan China
Abstract
Background: Considering the rapid expansion of hospital operations and the increasing digitisation of medical data, there is a pressing need for efficient and intelligent methods to process and analyse large‐scale medical data.
Methods: We integrated the QLoRA algorithm with ChatGLM2‐6b, Llama2‐7b, and Llama2‐13b models, fine‐tuning them on a local SQL dataset to optimise query performance. Prompt‐Engineering with ChatGPT was further applied for effective SQL execution.
Results: Original open‐source models showed almost no SQL generation capability (overall EX ≈ 0 for ChatGLM2‐6B and Llama2‐7B; 0.04 for Llama2‐13B). QLoRA fine‐tuning substantially improved performance, with QLoRA‐Llama2‐13B achieving the best results among open‐source models (overall EX 0.41 ± 0.030). Proprietary models demonstrated significantly stronger performance. Zero‐shot ChatGPT‐3.5 achieved moderate accuracy (EX 0.44 ± 0.007), which improved to 0.94 ± 0.017 with few‐shot prompting. GPT‐4.1 further improved performance, reaching 0.78 ± 0.038 in the zero‐shot setting and 0.96 ± 0.011 with few‐shot prompting. The few‐shot GPT‐4.1 results were comparable to those of database engineers (EX 0.97 ± 0.017), with no significant difference (p = 0.42).
Conclusion: Fine‐tuned LLMs and few‐shot GPT‐4.1 demonstrate substantial improvements in SQL query execution, providing a robust framework for efficient medical data analysis and informed hospital decision‐making.
Related
- Harnessing the Immune System in Hepatocellular Carcinoma Targeting: Mechanistic Insights and Therapeutic Potential of Natural Products and Traditional Chinese Medicine
- In-Hospital Guideline-Directed Medical Therapy Intensification and Long-Term Outcomes in Patients Hospitalized for Heart Failure in the Current Clinical Paradigm ― Insights From the QHEART-R Study ―
- A digitalized nosocomial risk management framework during the COVID-19 pandemic: practice-based insights from a tertiary cancer hospital in Southwest China
- Harnessing Neurogenic Potential of Glial Cell Types as an Emerging Strategy for Treating Epilepsy
- Harnessing Methylselenol (CH₃SeH): A Highly Effective Platform for Stimuli-Responsive Delivery With Therapeutic Potential.
BibTeX
@article{li2026harnessing,
title = {{Harnessing the Potential of LLMs in Hospital Management: Insights Into Medical Data Inquiry}},
author = {Jiaxi Li and Yu Zhang and Jing Zhao and Shengyang He and Danni Li},
journal = {The International Journal of Health Planning and Management},
year = {2026},
month = jul,
doi = {10.1002/hpm.70106},
url = {https://doi.org/10.1002/hpm.70106},
note = {CC BY-NC 4.0},
}
RIS
TY - JOUR
TI - Harnessing the Potential of LLMs in Hospital Management: Insights Into Medical Data Inquiry
AU - Jiaxi Li
AU - Yu Zhang
AU - Jing Zhao
AU - Shengyang He
AU - Danni Li
PY - 2026
DA - 2026/07/31
JO - The International Journal of Health Planning and Management
DO - 10.1002/hpm.70106
UR - https://doi.org/10.1002/hpm.70106
AB - Background: Considering the rapid expansion of hospital operations and the increasing digitisation of medical data, there is a pressing need for efficient and intelligent methods to process and analyse large‐scale medical data. Methods: We integrated the QLoRA algorithm with ChatGLM2‐6b, Llama2‐7b, and Llama2‐13b models, fine‐tuning them on a local SQL dataset to optimise query performance. Prompt‐Engineering with ChatGPT was further applied for effective SQL execution. Results: Original open‐source models showed almost no SQL generation capability (overall EX ≈ 0 for ChatGLM2‐6B and Llama2‐7B; 0.04 for Llama2‐13B). QLoRA fine‐tuning substantially improved performance, with QLoRA‐Llama2‐13B achieving the best results among open‐source models (overall EX 0.41 ± 0.030). Proprietary models demonstrated significantly stronger performance. Zero‐shot ChatGPT‐3.5 achieved moderate accuracy (EX 0.44 ± 0.007), which improved to 0.94 ± 0.017 with few‐shot prompting. GPT‐4.1 further improved performance, reaching 0.78 ± 0.038 in the zero‐shot setting and 0.96 ± 0.011 with few‐shot prompting. The few‐shot GPT‐4.1 results were comparable to those of database engineers (EX 0.97 ± 0.017), with no significant difference (p = 0.42). Conclusion: Fine‐tuned LLMs and few‐shot GPT‐4.1 demonstrate substantial improvements in SQL query execution, providing a robust framework for efficient medical data analysis and informed hospital decision‐making.
SN - 0749-6753
SN - 1099-1751
ER -