Neurological history both twinned and queried by generative artificial intelligence
Background and objectivesWe propose the use of GPT-4 to facilitate initial history-taking in neurology and other medical specialties. A large language model (LLM) could be utilized as a digital twin which could enhance queryable electronic medical record (EMR) systems and provide healthcare conversa...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2025-01-01
|
| Series: | Frontiers in Medicine |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fmed.2024.1496866/full |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841525860119609344 |
|---|---|
| author | Jung-Hyun Lee Jung-Hyun Lee Jung-Hyun Lee Eunhee Choi Sergio L. Angulo Sergio L. Angulo Robert A. McDougal Robert A. McDougal Robert A. McDougal Robert A. McDougal William W. Lytton William W. Lytton William W. Lytton |
| author_facet | Jung-Hyun Lee Jung-Hyun Lee Jung-Hyun Lee Eunhee Choi Sergio L. Angulo Sergio L. Angulo Robert A. McDougal Robert A. McDougal Robert A. McDougal Robert A. McDougal William W. Lytton William W. Lytton William W. Lytton |
| author_sort | Jung-Hyun Lee |
| collection | DOAJ |
| description | Background and objectivesWe propose the use of GPT-4 to facilitate initial history-taking in neurology and other medical specialties. A large language model (LLM) could be utilized as a digital twin which could enhance queryable electronic medical record (EMR) systems and provide healthcare conversational agents (HCAs) to replace waiting-room questionnaires.MethodsIn this observational pilot study, we presented verbatim history of present illness (HPI) narratives from published case reports of headache, stroke, and neurodegenerative diseases. Three standard GPT-4 models were designated Models P: patient digital twin; N: neurologist to query Model P; and S: supervisor to synthesize the N-P dialogue into a derived HPI and formulate the differential diagnosis. Given the random variability of GPT-4 output, each case was presented five separate times to check consistency and reliability.ResultsThe study achieved an overall HPI content retrieval accuracy of 81%, with accuracies of 84% for headache, 82% for stroke, and 77% for neurodegenerative diseases. Retrieval accuracies for individual HPI components were as follows: 93% for chief complaints, 47% for associated symptoms and review of systems, 76% for relevant symptom details, and 94% for histories of past medical, surgical, allergies, social, and family factors. The ranking of case diagnoses in the differential diagnosis list averaged in the 89th percentile.DiscussionOur tripartite LLM model demonstrated accuracy in extracting essential information from published case reports. Further validation with EMR HPIs, and then with direct patient care will be needed to move toward adaptation of enhanced diagnostic digital twins that incorporate real-time data from health-monitoring devices and self-monitoring assessments. |
| format | Article |
| id | doaj-art-84c73795ba8e410e9b5086c77547da8f |
| institution | Kabale University |
| issn | 2296-858X |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Medicine |
| spelling | doaj-art-84c73795ba8e410e9b5086c77547da8f2025-01-17T06:50:39ZengFrontiers Media S.A.Frontiers in Medicine2296-858X2025-01-011110.3389/fmed.2024.14968661496866Neurological history both twinned and queried by generative artificial intelligenceJung-Hyun Lee0Jung-Hyun Lee1Jung-Hyun Lee2Eunhee Choi3Sergio L. Angulo4Sergio L. Angulo5Robert A. McDougal6Robert A. McDougal7Robert A. McDougal8Robert A. McDougal9William W. Lytton10William W. Lytton11William W. Lytton12Department of Neurology, State University of New York Downstate Health Sciences University, Brooklyn, NY, United StatesDepartment of Neurology, Kings County Hospital, Brooklyn, NY, United StatesDepartment of Neurology, Maimonides Medical Center, Brooklyn, NY, United StatesDepartment of Internal Medicine, Lincoln Medical Center, Bronx, NY, United StatesDepartment of Neurology, State University of New York Downstate Health Sciences University, Brooklyn, NY, United StatesDepartment of Neurology, Kings County Hospital, Brooklyn, NY, United StatesDepartment of Biostatistics, Yale School of Public Health, Yale University, New Haven, CT, United StatesComputational Biology and Bioinformatics, Yale University, New Haven, CT, United StatesWu Tsai Institute, Yale University, New Haven, CT, United StatesBiomedical Informatics and Data Science, Yale School of Medicine, Yale University, New Haven, CT, United StatesDepartment of Neurology, State University of New York Downstate Health Sciences University, Brooklyn, NY, United StatesDepartment of Neurology, Kings County Hospital, Brooklyn, NY, United StatesDepartment of Physiology and Pharmacology, State University of New York Downstate Health Sciences University, Brooklyn, NY, United StatesBackground and objectivesWe propose the use of GPT-4 to facilitate initial history-taking in neurology and other medical specialties. A large language model (LLM) could be utilized as a digital twin which could enhance queryable electronic medical record (EMR) systems and provide healthcare conversational agents (HCAs) to replace waiting-room questionnaires.MethodsIn this observational pilot study, we presented verbatim history of present illness (HPI) narratives from published case reports of headache, stroke, and neurodegenerative diseases. Three standard GPT-4 models were designated Models P: patient digital twin; N: neurologist to query Model P; and S: supervisor to synthesize the N-P dialogue into a derived HPI and formulate the differential diagnosis. Given the random variability of GPT-4 output, each case was presented five separate times to check consistency and reliability.ResultsThe study achieved an overall HPI content retrieval accuracy of 81%, with accuracies of 84% for headache, 82% for stroke, and 77% for neurodegenerative diseases. Retrieval accuracies for individual HPI components were as follows: 93% for chief complaints, 47% for associated symptoms and review of systems, 76% for relevant symptom details, and 94% for histories of past medical, surgical, allergies, social, and family factors. The ranking of case diagnoses in the differential diagnosis list averaged in the 89th percentile.DiscussionOur tripartite LLM model demonstrated accuracy in extracting essential information from published case reports. Further validation with EMR HPIs, and then with direct patient care will be needed to move toward adaptation of enhanced diagnostic digital twins that incorporate real-time data from health-monitoring devices and self-monitoring assessments.https://www.frontiersin.org/articles/10.3389/fmed.2024.1496866/fullneurology–clinicalstrokeheadacheneurodegenerative diseaselarge language model (LLM)history taking |
| spellingShingle | Jung-Hyun Lee Jung-Hyun Lee Jung-Hyun Lee Eunhee Choi Sergio L. Angulo Sergio L. Angulo Robert A. McDougal Robert A. McDougal Robert A. McDougal Robert A. McDougal William W. Lytton William W. Lytton William W. Lytton Neurological history both twinned and queried by generative artificial intelligence Frontiers in Medicine neurology–clinical stroke headache neurodegenerative disease large language model (LLM) history taking |
| title | Neurological history both twinned and queried by generative artificial intelligence |
| title_full | Neurological history both twinned and queried by generative artificial intelligence |
| title_fullStr | Neurological history both twinned and queried by generative artificial intelligence |
| title_full_unstemmed | Neurological history both twinned and queried by generative artificial intelligence |
| title_short | Neurological history both twinned and queried by generative artificial intelligence |
| title_sort | neurological history both twinned and queried by generative artificial intelligence |
| topic | neurology–clinical stroke headache neurodegenerative disease large language model (LLM) history taking |
| url | https://www.frontiersin.org/articles/10.3389/fmed.2024.1496866/full |
| work_keys_str_mv | AT junghyunlee neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT junghyunlee neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT junghyunlee neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT eunheechoi neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT sergiolangulo neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT sergiolangulo neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT robertamcdougal neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT robertamcdougal neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT robertamcdougal neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT robertamcdougal neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT williamwlytton neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT williamwlytton neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence AT williamwlytton neurologicalhistorybothtwinnedandqueriedbygenerativeartificialintelligence |