Search Articles

View query in Help articles search

Search Results (1 to 10 of 10 Results)

Download search results: CSV END BibTex RIS


Using Large Language Models to Automate Data Extraction From Surgical Pathology Reports: Retrospective Cohort Study

Using Large Language Models to Automate Data Extraction From Surgical Pathology Reports: Retrospective Cohort Study

For reports of lengths greater than what the model could accommodate, we split the report text into 1200-character-long segments, followed by converting each of these segments into machine-readable numerical representations called embeddings. As embeddings encode meaning, calculating similarity scores using the cosine similarity metric between segment embeddings and posed questions allowed us to retrieve the pieces of text most directly related to the content of the question.

Denise Lee, Akhil Vaid, Kartikeya M Menon, Robert Freeman, David S Matteson, Michael L Marin, Girish N Nadkarni

JMIR Form Res 2025;9:e64544

Large Language Models–Supported Thrombectomy Decision-Making in Acute Ischemic Stroke Based on Radiology Reports: Feasibility Qualitative Study

Large Language Models–Supported Thrombectomy Decision-Making in Acute Ischemic Stroke Based on Radiology Reports: Feasibility Qualitative Study

As nearly all information can be expressed in human language, recorded in text form in the radiology report, recent advancements in the development of large language models (LLMs) have opened new opportunities for medical information processing [21-24]. One of these autoregressive LLMs is GPT-3 used within the pretrained chatbot application version, named Chat GPT, developed by Open AI [25].

Jonathan Kottlors, Robert Hahnfeldt, Lukas Görtz, Andra-Iza Iuga, Philipp Fervers, Johannes Bremm, David Zopfs, Kai R Laukamp, Oezguer A Onur, Simon Lennartz, Michael Schönfeld, David Maintz, Christoph Kabbasch, Thorsten Persigehl, Marc Schlamann

J Med Internet Res 2025;27:e48328

Automated System to Capture Patient Symptoms From Multitype Japanese Clinical Texts: Retrospective Study

Automated System to Capture Patient Symptoms From Multitype Japanese Clinical Texts: Retrospective Study

We compared these results with those reported manually in a previous report [13]. Workflow of our natural language processing system, which is composed of named entity recognition, normalization, and aggregation. Text X and Text Y are examples of 2 types of documents respectively, for example, physician progress notes and pharmacist progress notes.

Tomohiro Nishiyama, Ayane Yamaguchi, Peitao Han, Lis Weiji Kanashiro Pereira, Yuka Otsuki, Gabriel Herman Bernardim Andrade, Noriko Kudo, Shuntaro Yada, Shoko Wakamiya, Eiji Aramaki, Masahiro Takada, Masakazu Toi

JMIR Med Inform 2024;12:e58977

Descriptions of Scientific Evidence and Uncertainty of Unproven COVID-19 Therapies in US News: Content Analysis Study

Descriptions of Scientific Evidence and Uncertainty of Unproven COVID-19 Therapies in US News: Content Analysis Study

In instances of substantial similarity, only the most up-to-date report was included. Reports that were duplicative, unobtainable, or void of search terms were removed resulting in 1103 reports. If multiple therapeutics were discussed in a single report, the codebook was applied to each therapeutic. We randomly sampled a total of 550 news reports from our population to manage coder effort while allowing the capture of codes and themes among news reports.

Sara Watson, Tyler J Benning, Alessandro R Marcon, Xuan Zhu, Timothy Caulfield, Richard R Sharp, Zubin Master

JMIR Infodemiology 2024;4:e51328

A Machine Learning Approach with Human-AI Collaboration for Automated Classification of Patient Safety Event Reports: Algorithm Development and Validation Study

A Machine Learning Approach with Human-AI Collaboration for Automated Classification of Patient Safety Event Reports: Algorithm Development and Validation Study

PSE reporting systems are tools that allow frontline health care personnel to voluntarily report adverse events, near-misses, and unsafe conditions [11]. Each PSE report includes structured data, such as event types, patient harm level, date, and location of the event, as well as unstructured data, including a free-text section that contains the factual description of the event and the patient’s outcome [12].

Hongbo Chen, Eldan Cohen, Dulaney Wilson, Myrtede Alfred

JMIR Hum Factors 2024;11:e53378

Extracting Clinical Information From Japanese Radiology Reports Using a 2-Stage Deep Learning Approach: Algorithm Development and Validation

Extracting Clinical Information From Japanese Radiology Reports Using a 2-Stage Deep Learning Approach: Algorithm Development and Validation

They reported that their schema had a higher report coverage in their corpus. In this study, we developed a 2-stage deep learning system for extracting clinical information from CT reports. For secondary use of the radiology reports, we believe that our system has some advantages compared with recent related works [18,20,22,25,26]. First, our 2-stage NLP system can represent clinical information in a structured format, which can be challenging when only using an entity extraction approach.

Kento Sugimoto, Shoya Wada, Shozo Konishi, Katsuki Okada, Shirou Manabe, Yasushi Matsumura, Toshihiro Takeda

JMIR Med Inform 2023;11:e49041

Developing Reporting Guidelines for Studies of HIV Drug Resistance Prevalence: Protocol for a Mixed Methods Study

Developing Reporting Guidelines for Studies of HIV Drug Resistance Prevalence: Protocol for a Mixed Methods Study

Transcripts from the focus group discussions will provide key agreement-based insights on why these items are important to report. Mixed methods suit research objectives that cannot met by either qualitative or quantitative methodologies alone [20,21]. The sequential explanatory design is well suited for this research as the quantitative phase provides the recommended reporting items and the qualitative phase provides the rationale for reporting these items.

Cristian Garcia, Nadia Rehman, Daeria O Lawson, Pascal Djiadeu, Lawrence Mbuagbaw

JMIR Res Protoc 2022;11(5):e35969