Search Articles

View query in Help articles search

Search Results (1 to 10 of 121 Results)

Download search results: CSV END BibTex RIS


Improving Large Language Models’ Summarization Accuracy by Adding Highlights to Discharge Notes: Comparative Evaluation

Improving Large Language Models’ Summarization Accuracy by Adding Highlights to Discharge Notes: Comparative Evaluation

Therefore, LLM-generated summaries may sometimes omit crucial information or lack proper structure. In this paper, we present the first step in the process of simplifying discharge notes by harnessing the summarization capabilities of LLMs. In this study, we aim to generate more accurate, structured summaries from discharge notes where headers provide clear orientation and make the content easier to understand [25].

Mahshad Koohi Habibi Dehkordi, Yehoshua Perl, Fadi P Deek, Zhe He, Vipina K Keloth, Hao Liu, Gai Elhanan, Andrew J Einstein

JMIR Med Inform 2025;13:e66476

Feasibility of a Randomized Controlled Trial of Large AI-Based Linguistic Models for Clinical Reasoning Training of Physical Therapy Students: Pilot Randomized Parallel-Group Study

Feasibility of a Randomized Controlled Trial of Large AI-Based Linguistic Models for Clinical Reasoning Training of Physical Therapy Students: Pilot Randomized Parallel-Group Study

After being randomized, for those students belonging to the experimental group (LLM Group), a personal LLM Chat GPT account in version 3.5 was generated for them for a period of 1 month. Using this account, the participants solved a total of 4 clinical cases for 4 weeks, one per week, in which the LLM will serve as a virtual patient, answering the questions that the student asked and based on a physical therapy diagnosis, participants proposed a treatment for the virtual patient.

Raúl Ferrer-Peña, Silvia Di-Bonaventura, Alberto Pérez-González, Alfredo Lerín-Calvo

JMIR Form Res 2025;9:e66126

Implementing Large Language Models in Health Care: Clinician-Focused Review With Interactive Guideline

Implementing Large Language Models in Health Care: Clinician-Focused Review With Interactive Guideline

To help clinicians and health practitioners select LLMs, we proposed an interactive guideline with a clinical LLM selector tool that relies on a large-scale decision tree containing hundreds of nodes (general description in Figure 1). Using LLM names as keys, we recorded the number of appearances of 330 identified LLMs and their frequency of performing best by clinical task and input and output modalities.

HongYi Li, Jun-Fen Fu, Andre Python

J Med Internet Res 2025;27:e71916

Evaluating a Large Language Model’s Ability to Synthesize a Health Science Master’s Thesis: Case Study

Evaluating a Large Language Model’s Ability to Synthesize a Health Science Master’s Thesis: Case Study

Conceptually our workflow advanced in the following logical sequence (with continuous prompt optimization based on the output from the LLM): (1) upload dataset, prompt the LLM to analyze and present results using specified method, (2) suggest methods chapter based on user predetermined context, (3) suggest introduction chapter, with emphasis on providing relevant scientific citations, (4) develop a discussion, with emphasis on integrating relevant scientific citations, (5) suggest abstract and title, and (6)

Pål Joranger, Sara Rivenes Lafontan, Asgeir Brevik

JMIR Form Res 2025;9:e73248

Enhancing Magnetic Resonance Imaging (MRI) Report Comprehension in Spinal Trauma: Readability Analysis of AI-Generated Explanations for Thoracolumbar Fractures

Enhancing Magnetic Resonance Imaging (MRI) Report Comprehension in Spinal Trauma: Readability Analysis of AI-Generated Explanations for Thoracolumbar Fractures

We hypothesized that Chat GPT-generated summaries would help provide clearer and more understandable MRI report findings that contain accurate explanations of imaging findings without any “hallucinated” or fabricated content—a flaw observed in earlier LLM versions where the AI program would often invent facts or cite nonexistent literature without clearly acknowledging the fabricated content.

David C Sing, Kishan S Shah, Michael Pompliano, Paul H Yi, Calogero Velluto, Ali Bagheri, Robert K Eastlack, Stephen R Stephan, Gregory M Mundis Jr

JMIR AI 2025;4:e69654

Supervised Natural Language Processing Classification of Violent Death Narratives: Development and Assessment of a Compact Large Language Model

Supervised Natural Language Processing Classification of Violent Death Narratives: Development and Assessment of a Compact Large Language Model

Target outcomes with class imbalance were selected, as this setting is likely of most use to NVDRS applications, and models were fit using a compact LLM to reflect settings where computing resources are limited.

Susan T Parker

JMIR AI 2025;4:e68212

Enhancing Pulmonary Disease Prediction Using Large Language Models With Feature Summarization and Hybrid Retrieval-Augmented Generation: Multicenter Methodological Study Based on Radiology Report

Enhancing Pulmonary Disease Prediction Using Large Language Models With Feature Summarization and Hybrid Retrieval-Augmented Generation: Multicenter Methodological Study Based on Radiology Report

These findings are further analyzed by the LLM to extract disease-specific features and rank their importance. Then the summarized features are used by the LLM to generates diagnostic questions to construct a logical reasoning pathway. During the prediction phase, the workflow retrieves similar imaging reports via a hybrid RAG framework to refine the LLM’s understanding of disease patterns, ultimately generating comprehensive and precise results for disease prediction.

Ronghao Li, Shuai Mao, Congmin Zhu, Yingliang Yang, Chunting Tan, Li Li, Xiangdong Mu, Honglei Liu, Yuqing Yang

J Med Internet Res 2025;27:e72638

Chatbot for the Return of Positive Genetic Screening Results for Hereditary Cancer Syndromes: Prompt Engineering Project

Chatbot for the Return of Positive Genetic Screening Results for Hereditary Cancer Syndromes: Prompt Engineering Project

Indeed, creating a hybrid chatbot with both rule-based and LLM components can offer a versatile and streamlined user experience by ensuring that key information is covered in the rule-based components of the chatbot and allowing for the LLM component to support complex, open-ended queries that are not covered in the scripted content.

Emma Coen, Guilherme Del Fiol, Kimberly A Kaphingst, Emerson Borsato, Jackilen Shannon, Hadley Smith, Aaron Masino, Caitlin G Allen

JMIR Cancer 2025;11:e65848