GatorTron: A Large Clinical Language Model to Unlock Patient Information from Unstructured Electronic Health Records

Yang, Xi; PourNejatian, Nima; Shin, Hoo Chang; Smith, Kaleb E; Parisien, Christopher; Compas, Colin B.; Martin, Cheryl; Flores, Mona G.; Zhang, Ying; Magoč, Tanja; Harle, Christopher A.; Lipori, Gloria; Mitchell, Duane A.; Hogan, William R.; Shenkman, Elizabeth; Bian, Jiang; Wu, Yangjie

doi:10.48550/arxiv.2203.03540

Cited by 12 publications

(12 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the recent advancements in LLMs, their integration within healthcare has been evaluated in a diverse set of tasks such as summarizing patients' health records, writing discharge summaries, getting assistance for medical research and evaluating clinical scenarios to formulate differential diagnoses [24], [46], [50], [51]. In our perspective, the eligibility of LLMs for deployment in clinical related tasks hinges on their proficiency in two critical dimensions: their capability to serve as a robust and reliable knowledge repository and their capacity to effectively function as intelligent processors of natural language.…”

Section: Discussionmentioning

confidence: 99%

Occurrence of methicillin-resistant staphylococci in Veterinary Teaching Hospital, University of Peradeniya

Manathunga

Dissanayake

2022

Sri Lankan J Infec Dis

View full text Add to dashboard Cite

show abstract

Section: Discussionmentioning

confidence: 99%

Occurrence of methicillin-resistant staphylococci in Veterinary Teaching Hospital, University of Peradeniya

Manathunga

Dissanayake

2022

Sri Lankan J Infec Dis

View full text Add to dashboard Cite

show abstract

“…However for certain specialized tasks these models perform abysmally or generate incorrect information. Researchers have suggested [6] that this limitation can be overcome by using datasets from healthcare domains e.g., LLMs trained using EHR data like GatorTron [1] and [2], or LLMs trained using medical datasets like Med-PaLM 2 [6] and Flan-PaLM [5]. Hospital systems and healthcare organizations are still hesitant to employ LLMs in production because of high cost and liability associated with getting the questions wrong.…”

Section: Llms In Healthcarementioning

confidence: 99%

Creating Trustworthy LLMs: Dealing with Hallucinations in Healthcare AI

Ahmad,

Yaramic,

Roy

2023

Preprint

View full text Add to dashboard Cite

Large language models have proliferated across multiple domains in as short period of time. There is however hesitation in the medical and healthcare domain towards their adoption because of issues like factuality, coherence, and hallucinations. Give the high stakes nature of healthcare, many researchers have even cautioned against its usage until these issues are resolved. The key to the implementation and deployment of LLMs in healthcare is to make these models trustworthy, transparent (as much possible) and explainable. In this paper we describe the key elements in creating reliable, trustworthy, and unbiased models as a necessary condition for their adoption in healthcare. Specifically we focus on the quantification, validation, and mitigation of hallucinations in the context in healthcare. Lastly, we discuss how the future of LLMs in healthcare may look like.

show abstract

“…We are currently testing the models with ventilator settings, current lines and drains, and patient-centered features, like goals of care, in addition to improving model performance with higher-risk surgeries and patient populations (Ruppert 2022). Finally, the University of Florida houses GatorTron, one of the largest clinical language learning models (Yang et al 2022). Incorporation of clinical text into models is a new frontier in risk prediction which we are actively exploring through natural language processing techniques.…”

Section: Future Directionsmentioning

confidence: 99%

Building an automated, machine learning-enabled platform for predicting post-operative complications

Balch

Ruppert²,

Shickel³

et al. 2023

Physiol. Meas.

View full text Add to dashboard Cite

In 2019, the University of Florida College of Medicine launched the MySurgeryRisk algorithm to predict eight major post-operative complications using automatically extracted data from the electronic health record. This project was developed in parallel with our Intelligent Critical Care Center and represents a culmination of efforts to build an efficient and accurate model for data processing and predictive analytics. This paper discusses how our model was constructed and improved upon. We highlight the consolidation of the database, processing of fixed and time-series physiologic measurements, development and training of predictive models, and expansion of those models into different aspects of patient assessment and treatment. We end by discussing future directions of the model.

show abstract

GatorTron: A Large Clinical Language Model to Unlock Patient Information from Unstructured Electronic Health Records

Cited by 12 publications

References 0 publications

Occurrence of methicillin-resistant staphylococci in Veterinary Teaching Hospital, University of Peradeniya

Occurrence of methicillin-resistant staphylococci in Veterinary Teaching Hospital, University of Peradeniya

Creating Trustworthy LLMs: Dealing with Hallucinations in Healthcare AI

Building an automated, machine learning-enabled platform for predicting post-operative complications

Contact Info

Product

Resources

About