Extraction of Crohn’s Disease Clinical Phenotypes from Clinical Text Using Natural Language Processing DOI Creative Commons

Linea Schmidt,

Susanne Ibing, Florian Borchert

и другие.

medRxiv (Cold Spring Harbor Laboratory), Год журнала: 2023, Номер unknown

Опубликована: Окт. 16, 2023

Abstract Real-world studies based on electronic health records often require manual chart review to derive patients’ clinical phenotypes, a labor-intensive task with limited scalability. Here, we developed and compared computable phenotyping rules using the spaCy frame-work Large Language Model (LLM), GPT-4, for disease behavior age at diagnosis of Crohn’s patients. We are first describe algorithms texts these complex tasks previously described inter-annotator agreements between 0.54 0.98. The data comprised notes radiology reports from 584 Mount Sinai Health System Overall, observed similar or better performance GPT-4 rules. On note-level, F1 score was least 0.90 0.82 diagnosis. could not find statistical evidence difference human experts this task. Our findings underline potential LLMs phenotyping. Graphical

Язык: Английский

Leveraging Large Language Models for Enhancing Safety in Maritime Operations DOI Creative Commons
Tymoteusz Miller, Irmina Durlik, Ewelina Kostecka

и другие.

Applied Sciences, Год журнала: 2025, Номер 15(3), С. 1666 - 1666

Опубликована: Фев. 6, 2025

Maritime operations play a critical role in global trade but face persistent safety challenges due to human error, environmental factors, and operational complexities. This review explores the transformative potential of Large Language Models (LLMs) enhancing maritime through improved communication, decision-making, compliance. Specific applications include multilingual communication for international crews, automated reporting, interactive training, real-time risk assessment. While LLMs offer innovative solutions, such as data privacy, integration, ethical considerations must be addressed. concludes with actionable recommendations insights leveraging build safer more resilient systems.

Язык: Английский

Процитировано

1

ChatGPT-4o and 4o1 Preview as Dietary Support Tools in a Real-World Medicated Obesity Program: A Prospective Comparative Analysis DOI Open Access
Louis Talay,

Leif Lagesen,

A. W. C. Yip

и другие.

Healthcare, Год журнала: 2025, Номер 13(6), С. 647 - 647

Опубликована: Март 16, 2025

Background/Objectives: Clinicians are becoming increasingly interested in the use of large language models (LLMs) obesity services. While most experts agree that LLM integration would increase access to care and its efficiency, many remain skeptical their scientific accuracy capacity convey human empathy. Recent studies have shown ChatGPT-3 capable emulating dietitian responses a range basic dietary questions. Methods: This study compared two ChatGPT-4o those from dietitians across 10 complex questions (5 broad; 5 narrow) derived patient–clinician interactions within real-world medicated digital weight loss service. Results: Investigators found neither nor Chat GPT-4o1 preview were statistically outperformed (p < 0.05) by on any study’s The same finding was made when scores aggregated ten following four individual criteria: correctness, comprehensibility, empathy/relatability, actionability. Conclusions: These results provide preliminary evidence advanced LLMs may be able play significant supporting role Research other contexts is needed before stronger conclusions about lifestyle coaching whether such initiatives access.

Язык: Английский

Процитировано

0

Factors associated with abusive head trauma in young children presenting to emergency medical services using a large language model DOI

Allison Broad,

Xiao Luo, Fattah Muhammad Tahabi

и другие.

Prehospital Emergency Care, Год журнала: 2025, Номер unknown, С. 1 - 16

Опубликована: Янв. 13, 2025

Abusive head trauma (AHT) is a leading cause of death in young children. Analyses patient characteristics presenting to Emergency Medical Services (EMS) are often limited structured data fields. Artificial Intelligence (AI) and Large Language Models (LLM) may identify rare presentations like AHT through factors not found data. Our goal was apply AI LLM EMS narrative documentation children detect AHT. This retrospective cohort study transports <36 months age with diagnosis injury from the 2018-2019 ESO Research Data Collaborative. Non-abusive closed (NA-CHI) distinguished child maltreatment (AHT-CAN) 2 expert reviewers; kappa statistic (k) assessed inter-rater reliability. A Natural Processing (NLP) framework using an augmented derived n-grams developed AHT-CAN. We compared test (sensitivity, specificity, negative predictive value (NPV)) between this NLP Generative Pretrained Transformer (GPT) or only models Association specific word tokens AHT-CAN analyzed Pearson's chi-square. Area Under Receiver Operator Curve (AUROC) Precision-Recall (AUPRC) also reported. There were 1082 encounters our cohort; 1030 (95.2%) NA-CHI 52 (4.8%) Inter-rater agreement substantial (k= 0.71). The had specificity sensitivity 72.4% 92.3%, respectively NPV 99.5%. In comparison, GPT model 69.2%, 97.1% 98.4% alone 53.8%, 62.0%, 96.4%. AUROC 0.91 AUPRC 0.52. total 44 bi-grams positively associated including "domestic", "various", "bruise", "cheek", "multiple", "doa", "not respond", "see EMS". LLMs have high free-text narratives. Words physical signs strongly list help that aid detection

Язык: Английский

Процитировано

0

Extraction of Crohn’s Disease Clinical Phenotypes from Clinical Text Using Natural Language Processing DOI Creative Commons

Linea Schmidt,

Susanne Ibing, Florian Borchert

и другие.

medRxiv (Cold Spring Harbor Laboratory), Год журнала: 2023, Номер unknown

Опубликована: Окт. 16, 2023

Abstract Real-world studies based on electronic health records often require manual chart review to derive patients’ clinical phenotypes, a labor-intensive task with limited scalability. Here, we developed and compared computable phenotyping rules using the spaCy frame-work Large Language Model (LLM), GPT-4, for disease behavior age at diagnosis of Crohn’s patients. We are first describe algorithms texts these complex tasks previously described inter-annotator agreements between 0.54 0.98. The data comprised notes radiology reports from 584 Mount Sinai Health System Overall, observed similar or better performance GPT-4 rules. On note-level, F1 score was least 0.90 0.82 diagnosis. could not find statistical evidence difference human experts this task. Our findings underline potential LLMs phenotyping. Graphical

Язык: Английский

Процитировано

1