Fine-Tuning a Personalized OpenBioLLM Using Offline Reinforcement Learning

In the past decade, personalized treatment has become increasingly significant; however, the individualized care provided by physicians incurs substantial labor costs. The advancement of artificial intelligence, particularly the emergence of large language models, offers promising prospects for pers...

Full description

Saved in:
Bibliographic Details
Main Authors: Jinsheng Shi, Yuyu Yuan, Ao Wang, Meng Nie
Format: Article
Language:English
Published: MDPI AG 2025-02-01
Series:Applied Sciences
Subjects:
Online Access:https://www.mdpi.com/2076-3417/15/5/2486
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850050824964669440
author Jinsheng Shi
Yuyu Yuan
Ao Wang
Meng Nie
author_facet Jinsheng Shi
Yuyu Yuan
Ao Wang
Meng Nie
author_sort Jinsheng Shi
collection DOAJ
description In the past decade, personalized treatment has become increasingly significant; however, the individualized care provided by physicians incurs substantial labor costs. The advancement of artificial intelligence, particularly the emergence of large language models, offers promising prospects for personalized treatment. These models have the potential to create intelligent agents capable of delivering tailored medical care. Nonetheless, current large language models face challenges in achieving precise treatment recommendations and optimizing treatment strategies. This study aims to enhance the accuracy of medical large language models by employing representation learning, inverse reinforcement learning, and offline reinforcement learning techniques for fine-tuning. The fine-tuned models are capable of generating personalized treatment plans. Through experiments on the mechanical ventilation therapy dataset constructed from the MIMIC-III database, we found that the similarity between the treatment actions suggested by the fine-tuned models and those of clinical physicians reaches 90.4% in the survival dataset, while it decreases to 72.3% in the mortality dataset. These results suggest that fine-tuned models are able to effectively capture individual patient information and provide treatment strategies that differ from routine clinical practice in complex situations, rather than merely mimicking physician behavior. Nevertheless, we emphasize that the recommendations of the model should be used as an auxiliary tool for physician decision making, and the final medical decision still needs to be comprehensively considered in combination with the patient’s clinical background, medical history, and autonomy.
format Article
id doaj-art-9faddc84b91e49cbb6d50b162d8248f5
institution DOAJ
issn 2076-3417
language English
publishDate 2025-02-01
publisher MDPI AG
record_format Article
series Applied Sciences
spelling doaj-art-9faddc84b91e49cbb6d50b162d8248f52025-08-20T02:53:19ZengMDPI AGApplied Sciences2076-34172025-02-01155248610.3390/app15052486Fine-Tuning a Personalized OpenBioLLM Using Offline Reinforcement LearningJinsheng Shi0Yuyu Yuan1Ao Wang2Meng Nie3School of Computer Science (National Pilot Software Engineering School), Beijing University of Posts and Telecommunications, Xitucheng Road 10, Haidian, Beijing 100876, ChinaSchool of Computer Science (National Pilot Software Engineering School), Beijing University of Posts and Telecommunications, Xitucheng Road 10, Haidian, Beijing 100876, ChinaSchool of Computer Science (National Pilot Software Engineering School), Beijing University of Posts and Telecommunications, Xitucheng Road 10, Haidian, Beijing 100876, ChinaSchool of Computer Science (National Pilot Software Engineering School), Beijing University of Posts and Telecommunications, Xitucheng Road 10, Haidian, Beijing 100876, ChinaIn the past decade, personalized treatment has become increasingly significant; however, the individualized care provided by physicians incurs substantial labor costs. The advancement of artificial intelligence, particularly the emergence of large language models, offers promising prospects for personalized treatment. These models have the potential to create intelligent agents capable of delivering tailored medical care. Nonetheless, current large language models face challenges in achieving precise treatment recommendations and optimizing treatment strategies. This study aims to enhance the accuracy of medical large language models by employing representation learning, inverse reinforcement learning, and offline reinforcement learning techniques for fine-tuning. The fine-tuned models are capable of generating personalized treatment plans. Through experiments on the mechanical ventilation therapy dataset constructed from the MIMIC-III database, we found that the similarity between the treatment actions suggested by the fine-tuned models and those of clinical physicians reaches 90.4% in the survival dataset, while it decreases to 72.3% in the mortality dataset. These results suggest that fine-tuned models are able to effectively capture individual patient information and provide treatment strategies that differ from routine clinical practice in complex situations, rather than merely mimicking physician behavior. Nevertheless, we emphasize that the recommendations of the model should be used as an auxiliary tool for physician decision making, and the final medical decision still needs to be comprehensively considered in combination with the patient’s clinical background, medical history, and autonomy.https://www.mdpi.com/2076-3417/15/5/2486representation learninginverse reinforcement learningoffline reinforcement learningmedical large language modelpersonalized treatment
spellingShingle Jinsheng Shi
Yuyu Yuan
Ao Wang
Meng Nie
Fine-Tuning a Personalized OpenBioLLM Using Offline Reinforcement Learning
Applied Sciences
representation learning
inverse reinforcement learning
offline reinforcement learning
medical large language model
personalized treatment
title Fine-Tuning a Personalized OpenBioLLM Using Offline Reinforcement Learning
title_full Fine-Tuning a Personalized OpenBioLLM Using Offline Reinforcement Learning
title_fullStr Fine-Tuning a Personalized OpenBioLLM Using Offline Reinforcement Learning
title_full_unstemmed Fine-Tuning a Personalized OpenBioLLM Using Offline Reinforcement Learning
title_short Fine-Tuning a Personalized OpenBioLLM Using Offline Reinforcement Learning
title_sort fine tuning a personalized openbiollm using offline reinforcement learning
topic representation learning
inverse reinforcement learning
offline reinforcement learning
medical large language model
personalized treatment
url https://www.mdpi.com/2076-3417/15/5/2486
work_keys_str_mv AT jinshengshi finetuningapersonalizedopenbiollmusingofflinereinforcementlearning
AT yuyuyuan finetuningapersonalizedopenbiollmusingofflinereinforcementlearning
AT aowang finetuningapersonalizedopenbiollmusingofflinereinforcementlearning
AT mengnie finetuningapersonalizedopenbiollmusingofflinereinforcementlearning