Anna Romanyukha, Mahta Mazloumi, Thomas De Waelheyns, Nidhi Mishra, Jurgen Jacobs, Niki Fitousi
{"title":"Development of a context-aware integrated training module based on large language models for continuous education in radiation protection","authors":"Anna Romanyukha, Mahta Mazloumi, Thomas De Waelheyns, Nidhi Mishra, Jurgen Jacobs, Niki Fitousi","doi":"10.1016/j.ejmp.2025.105090","DOIUrl":null,"url":null,"abstract":"<div><h3>Purpose</h3><div>Reliability of LLM-based AI chatbots can be enhanced by using domain-specific models with a controlled knowledge base, especially relevant in the context of education and training of healthcare professionals and researchers. The aim was to produce an LLM data model with the goal of continuous education in radiation protection, that allows users to access reliable scientific information by querying it on specific topics, eliminating the need for manually perusing educational materials.</div></div><div><h3>Method</h3><div>A domain-specific LLM data model was developed and trained using custom knowledge from several domains, tested applying various test scenarios, and fine-tuned to ensure optimal selections of hyperparameters including top k, chunk size, temperature, max tokens etc.</div></div><div><h3>Results</h3><div>The final model produced reliable and accurate answers to a variety of users and queries based on controlled educational materials. Embedding model and similarity cutoff had the greatest impact on model performance.</div></div><div><h3>Conclusion</h3><div>The developed model was trained and validated on radiation protection training material, allowing users to access information on topics including radiobiology and radiation protection in a quick and reliable manner.</div></div>","PeriodicalId":56092,"journal":{"name":"Physica Medica-European Journal of Medical Physics","volume":"137 ","pages":"Article 105090"},"PeriodicalIF":2.7000,"publicationDate":"2025-08-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Physica Medica-European Journal of Medical Physics","FirstCategoryId":"3","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1120179725002005","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING","Score":null,"Total":0}
引用次数: 0
Abstract
Purpose
Reliability of LLM-based AI chatbots can be enhanced by using domain-specific models with a controlled knowledge base, especially relevant in the context of education and training of healthcare professionals and researchers. The aim was to produce an LLM data model with the goal of continuous education in radiation protection, that allows users to access reliable scientific information by querying it on specific topics, eliminating the need for manually perusing educational materials.
Method
A domain-specific LLM data model was developed and trained using custom knowledge from several domains, tested applying various test scenarios, and fine-tuned to ensure optimal selections of hyperparameters including top k, chunk size, temperature, max tokens etc.
Results
The final model produced reliable and accurate answers to a variety of users and queries based on controlled educational materials. Embedding model and similarity cutoff had the greatest impact on model performance.
Conclusion
The developed model was trained and validated on radiation protection training material, allowing users to access information on topics including radiobiology and radiation protection in a quick and reliable manner.
期刊介绍:
Physica Medica, European Journal of Medical Physics, publishing with Elsevier from 2007, provides an international forum for research and reviews on the following main topics:
Medical Imaging
Radiation Therapy
Radiation Protection
Measuring Systems and Signal Processing
Education and training in Medical Physics
Professional issues in Medical Physics.