{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"medRxiv"}],"indexed":{"date-parts":[[2026,1,16]],"date-time":"2026-01-16T13:06:30Z","timestamp":1768568790547,"version":"3.49.0"},"posted":{"date-parts":[[2024,11,2]]},"group-title":"Health Informatics","reference-count":45,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2024,11,2]],"date-time":"2024-11-02T00:00:00Z","timestamp":1730505600000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2025,5,7]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                <jats:sec>\n                  <jats:title>Background<\/jats:title>\n                  <jats:p>Early detection of cognitive decline during the preclinical stage of Alzheimer\u2019s disease and related dementias (AD\/ADRD) is crucial for timely intervention and treatment. Clinical notes in the electronic health record contain valuable information that can aid in the early identification of cognitive decline. In this study, we utilize advanced large clinical language models, fine-tuned on clinical notes, to improve the early detection of cognitive decline.<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Methods<\/jats:title>\n                  <jats:p>We collected clinical notes from 2,166 patients spanning the 4 years preceding their initial mild cognitive impairment (MCI) diagnosis from the Enterprise Data Warehouse of Mass General Brigham. To train the model, we developed CD-Tron, built upon a large clinical language model that was finetuned using 4,949 expert-labeled note sections. For evaluation, the trained model was applied to 1,996 independent note sections to assess its performance on real-world unstructured clinical data. Additionally, we used explainable AI techniques, specifically SHAP values (SHapley Additive exPlanations), to interpret the model\u2019s predictions and provide insight into the most influential features. Error analysis was also facilitated to further analyze the model\u2019s prediction.<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Results<\/jats:title>\n                  <jats:p>CD-Tron significantly outperforms baseline models, achieving notable improvements in precision, recall, and AUC metrics for detecting cognitive decline (CD). Tested on many real-world clinical notes, CD-Tron demonstrated high sensitivity with only one false negative, crucial for clinical applications prioritizing early and accurate CD detection. SHAP-based interpretability analysis highlighted key textual features contributing to model predictions, supporting transparency and clinician understanding.<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Conclusion<\/jats:title>\n                  <jats:p>CD-Tron offers a novel approach to early cognitive decline detection by applying large clinical language models to free-text EHR data. Pretrained on real-world clinical notes, it accurately identifies early cognitive decline and integrates SHAP for interpretability, enhancing transparency in predictions.<\/jats:p>\n                <\/jats:sec>","DOI":"10.1101\/2024.10.31.24316386","type":"posted-content","created":{"date-parts":[[2024,11,2]],"date-time":"2024-11-02T17:20:13Z","timestamp":1730568013000},"source":"Crossref","is-referenced-by-count":0,"title":["CD-Tron: Leveraging Large Clinical Language Model for Early Detection of Cognitive Decline from Electronic Health Records"],"prefix":"10.64898","author":[{"given":"Hao","family":"Guan","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"John","family":"Novoa-Laurentiev","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Li","family":"Zhou","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"54368","reference":[{"issue":"1","key":"2025051008450753000_2024.10.31.24316386v2.1","first-page":"1","article-title":"Alzheimer\u2019s disease","volume":"1","year":"2015","journal-title":"Nature Reviews Disease Primers"},{"key":"2025051008450753000_2024.10.31.24316386v2.2","doi-asserted-by":"publisher","DOI":"10.1001\/jama.2021.5469"},{"issue":"2","key":"2025051008450753000_2024.10.31.24316386v2.3","first-page":"1","article-title":"National vital statistics reports","volume":"68","year":"2019","journal-title":"National Vital Statistics Reports"},{"issue":"1","key":"2025051008450753000_2024.10.31.24316386v2.4","first-page":"1","article-title":"Cost of care for Alzheimers disease and related dementias in the United States: 2016 to 2060","volume":"10","year":"2024","journal-title":"NPJ Aging"},{"key":"2025051008450753000_2024.10.31.24316386v2.5","first-page":"36","article-title":"Preclinical, prodromal, and dementia stages of Alzheimer\u2019s disease","volume":"15","year":"2019","journal-title":"Practical Neurology"},{"issue":"1","key":"2025051008450753000_2024.10.31.24316386v2.6","doi-asserted-by":"crossref","first-page":"369","DOI":"10.1146\/annurev-clinpsy-032816-045136","article-title":"Subjective cognitive decline in preclinical Alzheimer\u2019s disease","volume":"13","year":"2017","journal-title":"Annual review of clinical psychology"},{"key":"2025051008450753000_2024.10.31.24316386v2.7","doi-asserted-by":"publisher","DOI":"10.1186\/s13195-016-0227-5"},{"key":"2025051008450753000_2024.10.31.24316386v2.8","doi-asserted-by":"crossref","unstructured":"L. Robinson , E. Tang , J.-P. Taylor , Dementia: timely diagnosis and early intervention, BMJ 350 (2015).","DOI":"10.1136\/bmj.h3029"},{"key":"2025051008450753000_2024.10.31.24316386v2.9","doi-asserted-by":"publisher","DOI":"10.1186\/s13195-021-00948-8"},{"key":"2025051008450753000_2024.10.31.24316386v2.10","doi-asserted-by":"publisher","DOI":"10.1136\/jnnp-2021-328623"},{"issue":"3","key":"2025051008450753000_2024.10.31.24316386v2.11","doi-asserted-by":"crossref","first-page":"939","DOI":"10.3233\/JAD-161252","article-title":"Cerebrospinal fluid biomarkers and clinical progression in patients with subjective cognitive decline and mild cognitive impairment","volume":"58","year":"2017","journal-title":"Journal of Alzheimer\u2019s Disease"},{"issue":"8","key":"2025051008450753000_2024.10.31.24316386v2.12","doi-asserted-by":"crossref","first-page":"1134","DOI":"10.1002\/alz.12106","article-title":"Genome-wide association study of rate of cognitive decline in Alzheimer\u2019s disease patients identifies novel genes and pathways","volume":"16","year":"2020","journal-title":"Alzheimer\u2019s & Dementia"},{"issue":"8","key":"2025051008450753000_2024.10.31.24316386v2.13","doi-asserted-by":"crossref","first-page":"3506","DOI":"10.1002\/alz.12967","article-title":"Early prediction of Alzheimer\u2019s disease and related dementias using real-world electronic health records","volume":"19","year":"2023","journal-title":"Alzheimer\u2019s & Dementia"},{"issue":"6","key":"2025051008450753000_2024.10.31.24316386v2.14","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1145\/3127881","article-title":"Mining electronic health records (EHRs): A survey","volume":"50","year":"2018","journal-title":"ACM Computing Surveys (CSUR)"},{"key":"2025051008450753000_2024.10.31.24316386v2.15","first-page":"1","article-title":"Natural language processing in electronic health records in relation to healthcare decision-making: A systematic review","volume":"155","year":"2023","journal-title":"Computers in Biology and Medicine"},{"issue":"11","key":"2025051008450753000_2024.10.31.24316386v2.16","first-page":"1","article-title":"Development and validation of a deep learning model for earlier detection of cognitive decline from clinical notes in electronic health records","volume":"4","year":"2021","journal-title":"JAMA Network Open"},{"issue":"3","key":"2025051008450753000_2024.10.31.24316386v2.17","doi-asserted-by":"crossref","first-page":"3713","DOI":"10.1007\/s11042-022-13428-4","article-title":"Natural language processing: state of the art, current trends and challenges","volume":"82","year":"2023","journal-title":"Multimedia Tools and Applications"},{"key":"2025051008450753000_2024.10.31.24316386v2.18","first-page":"1","article-title":"Neural natural language processing for unstructured data in electronic health records: a review","volume":"46","year":"2022","journal-title":"Computer Science Review"},{"key":"2025051008450753000_2024.10.31.24316386v2.19","unstructured":"J. Devlin , M.-W. Chang , K. Lee , K. Toutanova , BERT: Pre-training of deep bidirectional transformers for language understanding, in: Proceedings of the 2019 Conference of the North American Chapter of the As-sociation for Computational Linguistics: Human Language Technologies), Association for Computational Linguistics, 2019, pp. 4171\u20134186."},{"key":"2025051008450753000_2024.10.31.24316386v2.20","article-title":"ClinicalBERT: Modeling clinical notes and predicting hospital readmission","year":"2019","journal-title":"arXiv preprint"},{"key":"2025051008450753000_2024.10.31.24316386v2.21","doi-asserted-by":"publisher","DOI":"10.1093\/bioinformatics\/btz682"},{"issue":"1","key":"2025051008450753000_2024.10.31.24316386v2.22","first-page":"1","article-title":"Med-BERT: pretrained contextualized embeddings on large-scale structured electronic health records for disease prediction","volume":"4","year":"2021","journal-title":"NPJ Digital Medicine"},{"key":"2025051008450753000_2024.10.31.24316386v2.23","doi-asserted-by":"publisher","DOI":"10.1145\/3605943"},{"key":"2025051008450753000_2024.10.31.24316386v2.24","doi-asserted-by":"crossref","unstructured":"P. Hager , F. Jungmann , R. Holland , K. Bhagat , I. Hubrecht , M. Knauer , J. Vielhauer , M. Makowski , R. Braren , G. Kaissis , et al., Evaluation and mitigation of the limitations of large language models in clinical decision-making, Nature Medicine (2024) 1\u201310.","DOI":"10.1101\/2024.01.26.24301810"},{"key":"2025051008450753000_2024.10.31.24316386v2.25","first-page":"1","article-title":"Ammu: a survey of transformer-based biomedical pretrained language models","volume":"126","year":"2022","journal-title":"Journal of Biomedical Informatics"},{"key":"2025051008450753000_2024.10.31.24316386v2.26","first-page":"1","article-title":"AD-BERT: Using pre-trained language model to predict the progression from mild cognitive impairment to Alzheimer\u2019s disease","volume":"144","year":"2023","journal-title":"Journal of Biomedical Informatics"},{"issue":"1","key":"2025051008450753000_2024.10.31.24316386v2.27","first-page":"1","article-title":"Extraction of clinical phenotypes for Alzheimer\u2019s disease dementia from clinical notes using natural language processing","volume":"6","year":"2023","journal-title":"JAMIA Open"},{"issue":"1","key":"2025051008450753000_2024.10.31.24316386v2.28","doi-asserted-by":"crossref","first-page":"42","DOI":"10.1007\/s41666-023-00125-6","article-title":"Can patients with dementia be identified in primary care electronic medical records using natural language processing?","volume":"7","year":"2023","journal-title":"Journal of Healthcare Informatics Research"},{"key":"2025051008450753000_2024.10.31.24316386v2.29","year":"2023","journal-title":"A survey of large language models"},{"key":"2025051008450753000_2024.10.31.24316386v2.30","article-title":"Large language models: A survey","year":"2024","journal-title":"arXiv preprint"},{"key":"2025051008450753000_2024.10.31.24316386v2.31","article-title":"A comprehensive overview of large language models","year":"2023","journal-title":"arXiv preprint"},{"key":"2025051008450753000_2024.10.31.24316386v2.32","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-023-02448-8"},{"key":"2025051008450753000_2024.10.31.24316386v2.33","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-023-06291-2"},{"key":"2025051008450753000_2024.10.31.24316386v2.34","first-page":"1","article-title":"Large language models in healthcare and medical domain: A review","volume":"11","year":"2024","journal-title":"in: Informatics"},{"key":"2025051008450753000_2024.10.31.24316386v2.35","doi-asserted-by":"crossref","unstructured":"X. Du , J. Novoa-Laurentiev , J. M. Plasek , Y.-W. Chuang , L. Wang , G. A. Marshall , S. K. Mueller , F. Chang , S. Datta , H. Paek , et al., Enhancing early detection of cognitive decline in the elderly: a comparative study utilizing large language models in clinical notes, EBioMedicine 109 (2024).","DOI":"10.1016\/j.ebiom.2024.105401"},{"issue":"1","key":"2025051008450753000_2024.10.31.24316386v2.36","first-page":"1","article-title":"A large language model for electronic health records","volume":"5","year":"2022","journal-title":"NPJ digital medicine"},{"issue":"7969","key":"2025051008450753000_2024.10.31.24316386v2.37","doi-asserted-by":"crossref","first-page":"357","DOI":"10.1038\/s41586-023-06160-y","article-title":"Health system-scale language models are all-purpose prediction engines","volume":"619","year":"2023","journal-title":"Nature"},{"issue":"6","key":"2025051008450753000_2024.10.31.24316386v2.38","doi-asserted-by":"crossref","first-page":"844","DOI":"10.1016\/j.jalz.2014.01.001","article-title":"A conceptual framework for research on subjective cognitive decline in preclinical Alzheimer\u2019s disease","volume":"10","year":"2014","journal-title":"Alzheimer\u2019s & Dementia"},{"key":"2025051008450753000_2024.10.31.24316386v2.39","first-page":"1639","volume-title":"in: AMIA Annual Symposium Proceedings","volume":"2011","year":"2011"},{"key":"2025051008450753000_2024.10.31.24316386v2.40","unstructured":"A. Vaswani , et al., Attention is all you need, Advances in Neural Information Processing Systems (2017) 1\u201311."},{"key":"2025051008450753000_2024.10.31.24316386v2.41","doi-asserted-by":"crossref","unstructured":"S. M. Jain , Hugging face, in: Introduction to transformers for NLP: With the hugging face library and models to solve problems, Springer, 2022, pp. 51\u201367.","DOI":"10.1007\/978-1-4842-8844-3_4"},{"key":"2025051008450753000_2024.10.31.24316386v2.42","first-page":"4765","article-title":"A unified approach to interpreting model predictions","volume":"30","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"2025051008450753000_2024.10.31.24316386v2.43","doi-asserted-by":"publisher","DOI":"10.1145\/3561048"},{"key":"2025051008450753000_2024.10.31.24316386v2.44","first-page":"3543","article-title":"Attention is not explanation","volume":"1","year":"2019","journal-title":"in: Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies"},{"key":"2025051008450753000_2024.10.31.24316386v2.45","doi-asserted-by":"crossref","unstructured":"S. Serrano , N. A. Smith , Is attention interpretable?, in: Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, 2019, pp. 2931\u20132951.","DOI":"10.18653\/v1\/P19-1282"}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.1101\/2024.10.31.24316386","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,1,15]],"date-time":"2026-01-15T20:19:04Z","timestamp":1768508344000},"score":1,"resource":{"primary":{"URL":"http:\/\/medrxiv.org\/lookup\/doi\/10.1101\/2024.10.31.24316386"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,11,2]]},"references-count":45,"URL":"https:\/\/doi.org\/10.1101\/2024.10.31.24316386","relation":{"is-preprint-of":[{"id-type":"doi","id":"10.1016\/j.jbi.2025.104830","asserted-by":"subject"}]},"subject":[],"published":{"date-parts":[[2024,11,2]]},"subtype":"preprint"}}