EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Detecting New Lesions Using a Large Language Model: Applications in Real‐World Multiple Sclerosis Datasets

Shane Poole, Nikki Sisodia iD, Kanishka Koshal, Kyra Henderson, Jaeleene Wijangco, Danelvis Paredes, Chelsea Chen, William Rowles, Amit Akula, Jens Wuerfel, Vishakha Sharma, UCSF Multiple Sclerosis and Neuroinflammation Center clinicians und 3 weitere

Annals of Neurology · 2025-08 · 98 (2) · 308-316

DOI: 10.1002/ana.27251 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Annals of Neurology, online erschienen am 25.04.2025. Verfasst von Shane Poole und 14 weiteren Autor:innen. Beteiligte Einrichtung: UCSF Weill Institute for Neurosciences, University of California San Francisco, San Francisco, CA, USA. Hauptthemen laut MEDLINE: Multiple Sclerosis, Magnetic Resonance Imaging, Electronic Health Records. Die Arbeit wurde bisher 13-mal zitiert (Crossref) und nennt 33 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: F. Hoffmann-La Roche.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Objective

Neuroimaging is routinely utilized to identify new inflammatory activity in multiple sclerosis (MS). A large language model to classify narrative magnetic resonance imaging reports in the electronic health record (EHR) as discrete data could provide significant benefits for MS research. The objectives of the current study were to develop such a prompt and to illustrate its research applications through a common clinical scenario: monitoring response to B‐cell depleting therapy (BCDT).

Methods

An institutional ecosystem that securely connects healthcare data with ChatGPT4 was applied to clinical MS magnetic resonance imaging reports in a single institutional EHR (2000–2022). A prompt (msLesionprompt) was developed and iteratively refined to classify the presence or absence of new T2‐weighted lesions (newT2w) and contrast‐enhancing lesions (CEL). The multistep validation included evaluating efficiency (time and cost), comparison with manually annotated reports using standard confusion matrix, and application to identifying predictors of newT2w/CEL after BCDT start.

Results

Accuracy of msLesionprompt was high for detection of newT2w (97%) and CEL (96.8%). All 14,888 available reports were categorized in 4.13 hours ($28); 79% showed no newT2w or CEL. Data extracted showed expected suppression of new activity by BCDT (>97% monitoring magnetic resonance images after an initial “rebaseline” scan). Neighborhood poverty (Area Deprivation Index) was identified as a predictor of inflammatory activity (newT2w: OR 1.69, 95% CI 1.10–2.59, p = 0.017; CEL: OR 1.54, 95% CI 1.01–2.34, p = 0.046).

Interpretation

Extracting discrete information from narrative imaging reports using an large language model is feasible and efficient. This approach could augment many real‐world analyses of MS disease evolution and treatment response. ANN NEUROL 2025;98:308–316

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Multiple SclerosisMagnetic Resonance ImagingElectronic Health RecordsHumansMaleFemaleAdultMiddle AgedNeuroimagingLarge Language Models

Autor:innen und Institutionen

  • UCSF Weill Institute for Neurosciences, University of California San Francisco, San Francisco, CA, USA.
  • F. Hoffmann-La Roche, Basel, Switzerland.
  • Roche Diagnostics, Santa Clara, USA.
  • UCSF Center for Intelligent Imaging (ci2), Department of Radiology & Biomedical imaging, University of California, San Francisco, CA, USA.

ORCID-Profile: Nikki Sisodia, Roland G Henry

Förderung

  • F. Hoffmann-La Roche

Zitieren

APA

Poole, S., Sisodia, N., Koshal, K., Henderson, K., Wijangco, J., Paredes, D., Chen, C., Rowles, W., Akula, A., Wuerfel, J., Sharma, V., UCSF Multiple Sclerosis and Neuroinflammation Center clinicians, Rauschecker, A. M., Henry, R. G., & Bove, R. (2025). Detecting New Lesions Using a Large Language Model: Applications in Real‐World Multiple Sclerosis Datasets. Annals of Neurology, 98(2), 308-316. https://doi.org/10.1002/ana.27251

Vancouver

Poole S, Sisodia N, Koshal K, Henderson K, Wijangco J, Paredes D, et al. Detecting New Lesions Using a Large Language Model: Applications in Real‐World Multiple Sclerosis Datasets. Annals of Neurology. 2025;98(2):308-316. doi:10.1002/ana.27251

BibTeX
@article{Poole2025,
  author = {Poole, Shane and Sisodia, Nikki and Koshal, Kanishka and Henderson, Kyra and Wijangco, Jaeleene and Paredes, Danelvis and Chen, Chelsea and Rowles, William and Akula, Amit and Wuerfel, Jens and Sharma, Vishakha and UCSF Multiple Sclerosis and Neuroinflammation Center clinicians, UCSF Multiple Sclerosis and Neuroinflammation Center clinicians and Rauschecker, Andreas M and Henry, Roland G and Bove, Riley},
  title = {Detecting New Lesions Using a Large Language Model: Applications in Real‐World Multiple Sclerosis Datasets},
  journal = {Annals of Neurology},
  year = {2025},
  volume = {98},
  number = {2},
  pages = {308--316},
  doi = {10.1002/ana.27251}
}
Kostenlos starten →