Casey Ta
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Casey Ta has a total of 161 co-authors. Here are the top co-authors by number of shared publications:
- Chunhua Weng
Department of Biomedical Informatics, Columbia University, New York, USA
16 shared publications - Cong Liu
Division of Genetics and Genomics, Department of Pediatrics, Boston Children’s Hospital, Harvard Medical School, Boston, USA
8 shared publications - Karthik Natarajan
Data Sciences Unit Columbia University New York New York USA
5 shared publications - Fangyi Chen
Department of Biomedical Informatics Columbia University New York New York USA
4 shared publications - Jae Hyun Kim
School of Pharmacy and Institute of New Drug Development, Jeonbuk National University, Republic of Korea
4 shared publications - Junghwan Lee
H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology , Atlanta, GA 30332
4 shared publications
Latest Publications
- Open Access
- 11 October 2026
- Published by Wiley in Autism Research
- Vol. 19 (11)
Characterizing Autism Phenotype Onset in the Electronic Health Records to Enable Age‐Specific Screening
Autism exhibits substantial phenotypic heterogeneity, yet age‐dependent variation in phenotypic profiles and onset timing remains poorly understood. We aimed to characterize stage‐specific phenotypes and temporal onset patterns of autism using longitudinal electronic health records (EHRs). We analyzed EHR data from Columbia University Irving Medical Center. Autism cases were stratified by age at diagnosis. Phenotype enrichment analyses were performed using chi‐square and Fisher's exact tests to identify phenotypes overrepresented in autism cases compared with matched controls, and temporal patterns of phenotype onset were subsequently evaluated to identify stage‐specific clinical signatures and provide insight into differences between early‐ and late‐diagnosed individuals. Among 4818 cases and 1433 phenotypes, 195 and 155 were enriched in toddlerhood and adolescence groups. Early diagnosed individuals exhibited a higher prevalence of sensory‐related conditions, greater exposure to maternal and perinatal complications than their matched controls (all ORs > 1; p < 0.01), and earlier onset of neurologic features, including coordination abnormalities and convulsions, compared with later‐diagnosed individuals. The study provides a publicly accessible resource to enrich autism‐related phenotypic profiles, uncovering notable phenotypic differences and variations in timing of phenotype onset between early‐diagnosed and later‐diagnosed individuals. These findings underscore phenotypic heterogeneity and highlight the need for age‐sensitive screening strategies. To our knowledge, this is among the first studies to jointly characterize phenotypic profiles and their temporal onset patterns across developmental stages in autism using longitudinal EHR data. These findings offer novel insights into age‐dependent clinical signatures, informing earlier and more precise screening strategies.
- 11 October 2026
- Published by Wiley in Autism Research
- Vol. 19 (11)
Open AccessCharacterizing Autism Phenotype Onset in the Electronic Health Records to Enable Age‐Specific Screening
Autism exhibits substantial phenotypic heterogeneity, yet age‐dependent variation in phenotypic profiles and onset timing remains poorly understood. We aimed to characterize stage‐specific phenotypes and temporal onset patterns of autism using longitudinal electronic health records (EHRs). We analyzed EHR data from Columbia University Irving Medical Center. Autism cases were stratified by age at diagnosis. Phenotype enrichment analyses were performed using chi‐square and Fisher's exact tests to identify phenotypes overrepresented in autism cases compared with matched controls, and temporal patterns of phenotype onset were subsequently evaluated to identify stage‐specific clinical signatures and provide insight into differences between early‐ and late‐diagnosed individuals. Among 4818 cases and 1433 phenotypes, 195 and 155 were enriched in toddlerhood and adolescence groups. Early diagnosed individuals exhibited a higher prevalence of sensory‐related conditions, greater exposure to maternal and perinatal complications than their matched controls (all ORs > 1; p < 0.01), and earlier onset of neurologic features, including coordination abnormalities and convulsions, compared with later‐diagnosed individuals. The study provides a publicly accessible resource to enrich autism‐related phenotypic profiles, uncovering notable phenotypic differences and variations in timing of phenotype onset between early‐diagnosed and later‐diagnosed individuals. These findings underscore phenotypic heterogeneity and highlight the need for age‐sensitive screening strategies. To our knowledge, this is among the first studies to jointly characterize phenotypic profiles and their temporal onset patterns across developmental stages in autism using longitudinal EHR data. These findings offer novel insights into age‐dependent clinical signatures, informing earlier and more precise screening strategies.
- Open Access
- 6 October 2026
- Published by Oxford University Press (OUP) in Journal of the American Medical Informatics Association
- Vol. 33 (12) , 2312-2322
Lost in transformation: diagnostic aggregation and semantic drift in ICD-derived kidney phenotypes
Objectives: To evaluate how cross-terminology mapping of International Classification of Diseases (ICD)-coded diagnoses is associated with preservation of diagnostic distinctions and traceability in electronic health record (EHR)-derived phenotypes for genomic research. Materials and Methods: We conducted a retrospective, cross-sectional study of real-world ICD code utilization and semantic fidelity among kidney-related diagnoses. Kidney-relevant ICD-9-CM and ICD-10-CM codes were identified using clinician-curated OHDSI ATLAS searches and evaluated in a large academic health system. ICD codes were mapped to Phecodes, SNOMED-CT, the Human Phenotype Ontology (HPO), and Online Mendelian Inheritance in Man (OMIM) using standardized crosswalks. Mapping coverage, richness, and multiplicity were quantified, with multiplicity operationalizing diagnostic aggregation. Three nephrologists independently rated label-level semantic fidelity (0-2 scale). Ordinal logistic regression evaluated associations between mapping multiplicity and semantic fidelity. Results: Among 585 nephrology-relevant ICD codes (571 435 patients), mapping coverage and richness varied across terminologies. Phecodes showed high coverage and mapping multiplicity, whereas SNOMED-CT, HPO, and OMIM mappings were predominantly one-to-one. Among ICD-Phecode mappings, 39% were rated poor, whereas no poor matches were observed among mapped SNOMED-CT, HPO, or OMIM concepts. Greater mapping multiplicity was associated with lower semantic fidelity (OR, 0.33; 95% CI, 0.29-0.37; P < .001). Most ICD-9-CM codes (96%) were represented in at least 1 external cohort. Discussion: Diagnostic aggregation was associated with reduced semantic fidelity and traceability, with potential implications for phenotype construction and interpretation in GWAS, PheWAS, and biobank-scale studies. Conclusion: This study provides an empirical framework for evaluating semantic drift in EHR-derived phenotypes and highlights a central tradeoff: abstraction enables scalability, whereas traceability enables translation.
- 6 October 2026
- Published by Oxford University Press (OUP) in Journal of the American Medical Informatics Association
- Vol. 33 (12) , 2312-2322
Open AccessLost in transformation: diagnostic aggregation and semantic drift in ICD-derived kidney phenotypes
Objectives: To evaluate how cross-terminology mapping of International Classification of Diseases (ICD)-coded diagnoses is associated with preservation of diagnostic distinctions and traceability in electronic health record (EHR)-derived phenotypes for genomic research. Materials and Methods: We conducted a retrospective, cross-sectional study of real-world ICD code utilization and semantic fidelity among kidney-related diagnoses. Kidney-relevant ICD-9-CM and ICD-10-CM codes were identified using clinician-curated OHDSI ATLAS searches and evaluated in a large academic health system. ICD codes were mapped to Phecodes, SNOMED-CT, the Human Phenotype Ontology (HPO), and Online Mendelian Inheritance in Man (OMIM) using standardized crosswalks. Mapping coverage, richness, and multiplicity were quantified, with multiplicity operationalizing diagnostic aggregation. Three nephrologists independently rated label-level semantic fidelity (0-2 scale). Ordinal logistic regression evaluated associations between mapping multiplicity and semantic fidelity. Results: Among 585 nephrology-relevant ICD codes (571 435 patients), mapping coverage and richness varied across terminologies. Phecodes showed high coverage and mapping multiplicity, whereas SNOMED-CT, HPO, and OMIM mappings were predominantly one-to-one. Among ICD-Phecode mappings, 39% were rated poor, whereas no poor matches were observed among mapped SNOMED-CT, HPO, or OMIM concepts. Greater mapping multiplicity was associated with lower semantic fidelity (OR, 0.33; 95% CI, 0.29-0.37; P < .001). Most ICD-9-CM codes (96%) were represented in at least 1 external cohort. Discussion: Diagnostic aggregation was associated with reduced semantic fidelity and traceability, with potential implications for phenotype construction and interpretation in GWAS, PheWAS, and biobank-scale studies. Conclusion: This study provides an empirical framework for evaluating semantic drift in EHR-derived phenotypes and highlights a central tradeoff: abstraction enables scalability, whereas traceability enables translation.
- Open Access
- 1 October 2026
- Published by Oxford University Press (OUP) in Journal of the American Medical Informatics Association
Guideline-grounded large language models for extracting genome-informed clinical recommendations from electronic health records
Background: Precision medicine requires delivery and tracking of genome-informed risk assessments (GIRAs), often documented in unstructured electronic health record (EHR) notes, making large-scale evaluation reliant on labor-intensive manual chart review. Large language models (LLMs) offer a promising approach to automated extraction, but their performance in genome-informed clinical contexts remains incompletely characterized. Objectives: To evaluate guideline-grounded LLM approaches for extracting genome-informed clinical recommendations from EHR notes in the eMERGE study. Materials and Methods: We developed an LLM-based pipeline to identify clinical recommendations associated with GIRA reports. Three LLMs (GPT-4o, LLaMA-3.3-70B, and LLaMA-3.1-8B) were evaluated using baseline prompting, guideline-aware prompting, and retrieval-augmented generation (RAG), validated against manual chart review (N = 18 participants; N = 34 documents). Results: Baseline prompting showed limited performance (F1 = 0.50-0.54). Incorporating guideline knowledge improved F1 by 0.14-0.31 across models, with LLaMA-3.1-8B using RAG achieving the highest performance (F1 = 0.81). Conclusion: Domain-grounded LLM approaches can support scalable extraction of genome-informed clinical recommendations from EHR data. Larger studies are needed to assess generalizability in real-world workflows.
- 1 October 2026
- Published by Oxford University Press (OUP) in Journal of the American Medical Informatics Association
Open AccessGuideline-grounded large language models for extracting genome-informed clinical recommendations from electronic health records
Background: Precision medicine requires delivery and tracking of genome-informed risk assessments (GIRAs), often documented in unstructured electronic health record (EHR) notes, making large-scale evaluation reliant on labor-intensive manual chart review. Large language models (LLMs) offer a promising approach to automated extraction, but their performance in genome-informed clinical contexts remains incompletely characterized. Objectives: To evaluate guideline-grounded LLM approaches for extracting genome-informed clinical recommendations from EHR notes in the eMERGE study. Materials and Methods: We developed an LLM-based pipeline to identify clinical recommendations associated with GIRA reports. Three LLMs (GPT-4o, LLaMA-3.3-70B, and LLaMA-3.1-8B) were evaluated using baseline prompting, guideline-aware prompting, and retrieval-augmented generation (RAG), validated against manual chart review (N = 18 participants; N = 34 documents). Results: Baseline prompting showed limited performance (F1 = 0.50-0.54). Incorporating guideline knowledge improved F1 by 0.14-0.31 across models, with LLaMA-3.1-8B using RAG achieving the highest performance (F1 = 0.81). Conclusion: Domain-grounded LLM approaches can support scalable extraction of genome-informed clinical recommendations from EHR data. Larger studies are needed to assess generalizability in real-world workflows.
- Open Access
Interpretable Fine-tuned Large Language Models Facilitate Making Genetic Test Decisions for Rare Diseases
Clinical decision making often relies on expert judgment guided by established guidelines, which can be challenging to standardize and abstract to implement. For example, selecting between gene panels and whole exome/genome sequencing (WES/WGS) for rare disease diagnosis frequently requires interpretation of evidence-based recommendations from the American College of Medical Genetics and Genomics (ACMG) guideline. Traditional machine learning (ML) models predicting suitable genetic tests often face interpretability limitations. We hypothesize that large language models (LLMs) can be fine-tuned to “mimic” clinicians’ reasoning patterns by interpreting and applying clinical guidelines with chain-of-thought (CoT). We present RareDAI, an integrative approach that addresses this challenge by analyzing heterogeneous clinical data, including unstructured notes and structured Phecodes. Using seven domain-specific questions, we guide the Llama 3.1 and Qwen 3 models to generate structured CoT outputs. These outputs are refined via our proposed self-distillation fine-tuning (SDFT) approach, enabling the model to produce interpretable reasoning prior to recommendation. RareDAI outperforms traditional supervised fine-tuning and base LLMs (e.g., Llama 3.1, GPT-4) by up to 10-20% in all metrics (accuracy, precision, recall, and F1-score) on both in-house data and external data, effectively assisting clinicians in selecting between diagnostic modalities across healthcare systems.
Open AccessInterpretable Fine-tuned Large Language Models Facilitate Making Genetic Test Decisions for Rare Diseases
Clinical decision making often relies on expert judgment guided by established guidelines, which can be challenging to standardize and abstract to implement. For example, selecting between gene panels and whole exome/genome sequencing (WES/WGS) for rare disease diagnosis frequently requires interpretation of evidence-based recommendations from the American College of Medical Genetics and Genomics (ACMG) guideline. Traditional machine learning (ML) models predicting suitable genetic tests often face interpretability limitations. We hypothesize that large language models (LLMs) can be fine-tuned to “mimic” clinicians’ reasoning patterns by interpreting and applying clinical guidelines with chain-of-thought (CoT). We present RareDAI, an integrative approach that addresses this challenge by analyzing heterogeneous clinical data, including unstructured notes and structured Phecodes. Using seven domain-specific questions, we guide the Llama 3.1 and Qwen 3 models to generate structured CoT outputs. These outputs are refined via our proposed self-distillation fine-tuning (SDFT) approach, enabling the model to produce interpretable reasoning prior to recommendation. RareDAI outperforms traditional supervised fine-tuning and base LLMs (e.g., Llama 3.1, GPT-4) by up to 10-20% in all metrics (accuracy, precision, recall, and F1-score) on both in-house data and external data, effectively assisting clinicians in selecting between diagnostic modalities across healthcare systems.
- 14 November 2025
- Published by Elsevier in Journal of Biomedical Informatics
- Vol. 172, 104956
A method for characterizing disease progression from acute kidney injury to chronic kidney disease
- 14 November 2025
- Published by Elsevier in Journal of Biomedical Informatics
- Vol. 172, 104956
A method for characterizing disease progression from acute kidney injury to chronic kidney disease
- Open Access
- 2 September 2025
- Published by Oxford University Press (OUP) in JAMIA Open
- Vol. 8 (5) , ooaf113
Implementing integrated genomic risk assessments for breast cancer: lessons learned from the Electronic Medical Records and Genomics study
Objectives: To implementation an automated multi-institutional pipeline that delivers breast-cancer risk integrated with polygenic risk scores, monogenic variants, family history, and clinical factors, emphasizing operational challenges and their solutions. Materials and Methods: A five-stage process was executed at ten sites. Data streams from REDCap surveys, PRS and monogenic reports, and MeTree pedigrees were normalized and forwarded through a REDCap plug-in to the CanRisk API. Results: Integrated risk was returned to >10 000 women; 3.6% were ≥25 % lifetime risk and 0.9% carried pathogenic variants. Pipeline generated score aligns well with manual generated ones. Major barriers such as heterogeneous pedigree formats, missing data, edge-case handling, and evolving model versions were identified and resolved through mapping rules, imputations, and iterative testing. Discussion: Cross-platform data harmonization and stakeholder alignment were decisive for success. Borderline-risk communication and model-version drift remain open issues. Conclusion: Large-scale PRS-integrated breast-cancer risk reporting is feasible but requires robust interoperability standards and iterative governance.
- 2 September 2025
- Published by Oxford University Press (OUP) in JAMIA Open
- Vol. 8 (5) , ooaf113
Open AccessImplementing integrated genomic risk assessments for breast cancer: lessons learned from the Electronic Medical Records and Genomics study
Objectives: To implementation an automated multi-institutional pipeline that delivers breast-cancer risk integrated with polygenic risk scores, monogenic variants, family history, and clinical factors, emphasizing operational challenges and their solutions. Materials and Methods: A five-stage process was executed at ten sites. Data streams from REDCap surveys, PRS and monogenic reports, and MeTree pedigrees were normalized and forwarded through a REDCap plug-in to the CanRisk API. Results: Integrated risk was returned to >10 000 women; 3.6% were ≥25 % lifetime risk and 0.9% carried pathogenic variants. Pipeline generated score aligns well with manual generated ones. Major barriers such as heterogeneous pedigree formats, missing data, edge-case handling, and evolving model versions were identified and resolved through mapping rules, imputations, and iterative testing. Discussion: Cross-platform data harmonization and stakeholder alignment were decisive for success. Borderline-risk communication and model-version drift remain open issues. Conclusion: Large-scale PRS-integrated breast-cancer risk reporting is feasible but requires robust interoperability standards and iterative governance.
- Open Access
- 9 July 2025
- Published by Wiley in Clinical and Translational Science
- Vol. 18 (7) , e70284
Announcing the Biomedical Data Translator: Initial Public Release
The growing availability of biomedical data offers vast potential to improve human health, but the complexity and lack of integration of these datasets often limit their utility. To address this, the Biomedical Data Translator Consortium has developed an open‐source knowledge graph–based system—Translator—designed to integrate, harmonize, and make inferences over diverse biomedical data sources. We announce here Translator's initial public release and provide an overview of its architecture, standards, user interface, and core features. Translator employs a scalable, federated, knowledge graph framework for the integration of clinical, genomic, pharmacological, and other biomedical knowledge sources, enabling query retrieval, inference, and hypothesis generation. Translator's user interface is designed to support the exploration of knowledge relationships and the generation of insights, without requiring deep technical expertise and gradually revealing more detailed evidence, provenance, and confidence information, as needed by a given user. To demonstrate Translator's application and impact, we highlight features of the user interface in the context of three real‐world use cases: suggesting potential therapeutics for patients with rare disease; explaining the mechanism of action of a pipeline drug; and screening and validating drug candidates in a model organism. We discuss strengths and limitations of reasoning within a largely federated system and the need for rich concept modeling and deep provenance tracking. Finally, we outline future directions for enhancing Translator's functionality and expanding its data sources. Translator represents a significant step forward in making complex biomedical knowledge more accessible and actionable, aiming to accelerate translational research and improve patient care.
- 9 July 2025
- Published by Wiley in Clinical and Translational Science
- Vol. 18 (7) , e70284
Open AccessAnnouncing the Biomedical Data Translator: Initial Public Release
The growing availability of biomedical data offers vast potential to improve human health, but the complexity and lack of integration of these datasets often limit their utility. To address this, the Biomedical Data Translator Consortium has developed an open‐source knowledge graph–based system—Translator—designed to integrate, harmonize, and make inferences over diverse biomedical data sources. We announce here Translator's initial public release and provide an overview of its architecture, standards, user interface, and core features. Translator employs a scalable, federated, knowledge graph framework for the integration of clinical, genomic, pharmacological, and other biomedical knowledge sources, enabling query retrieval, inference, and hypothesis generation. Translator's user interface is designed to support the exploration of knowledge relationships and the generation of insights, without requiring deep technical expertise and gradually revealing more detailed evidence, provenance, and confidence information, as needed by a given user. To demonstrate Translator's application and impact, we highlight features of the user interface in the context of three real‐world use cases: suggesting potential therapeutics for patients with rare disease; explaining the mechanism of action of a pipeline drug; and screening and validating drug candidates in a model organism. We discuss strengths and limitations of reasoning within a largely federated system and the need for rich concept modeling and deep provenance tracking. Finally, we outline future directions for enhancing Translator's functionality and expanding its data sources. Translator represents a significant step forward in making complex biomedical knowledge more accessible and actionable, aiming to accelerate translational research and improve patient care.
- Open Access
- 15 May 2025
- Published by IOS Press
- Vol. 327, 123-127
Identifying Phenotypes for Earlier Diagnosis of Rare Diseases
Rare diseases, while individually rare, cumulatively affect a large population, and patients often undergo long and arduous diagnostic odysseys. Toward the goal of supporting earlier diagnosis of rare diseases, we developed generalizable methods of extracting rare diseases and phenotypes from structured electronic health records and clinical notes. We analyzed the distributions of the age of onset of phenotypes per disease to identify disease-phenotype associations, producing a dataset with over 500 thousand associations covering 2300 rare diseases. Disease-phenotype associations are characterized by disease prevalence and mean age of onset of the phenotype to aid phenotype selection according to the priorities of the clinical decision support task.
- 15 May 2025
- Published by IOS Press
- Vol. 327, 123-127
Open AccessIdentifying Phenotypes for Earlier Diagnosis of Rare Diseases
Rare diseases, while individually rare, cumulatively affect a large population, and patients often undergo long and arduous diagnostic odysseys. Toward the goal of supporting earlier diagnosis of rare diseases, we developed generalizable methods of extracting rare diseases and phenotypes from structured electronic health records and clinical notes. We analyzed the distributions of the age of onset of phenotypes per disease to identify disease-phenotype associations, producing a dataset with over 500 thousand associations covering 2300 rare diseases. Disease-phenotype associations are characterized by disease prevalence and mean age of onset of the phenotype to aid phenotype selection according to the priorities of the clinical decision support task.
- Open Access
- 21 November 2024
- Published by Springer Nature in npj Digital Medicine
- Vol. 7 (1) , 1-12
Phenotype driven molecular genetic test recommendation for diagnosing pediatric rare disorders
Patients with rare diseases often experience prolonged diagnostic delays. Ordering appropriate genetic tests is crucial yet challenging, especially for general pediatricians without genetic expertise. Recent American College of Medical Genetics (ACMG) guidelines embrace early use of exome sequencing (ES) or genome sequencing (GS) for conditions like congenital anomalies or developmental delays while still recommend gene panels for patients exhibiting strong manifestations of a specific disease. Recognizing the difficulty in navigating these options, we developed a machine learning model trained on 1005 patient records from Columbia University Irving Medical Center to recommend appropriate genetic tests based on the phenotype information. The model achieved a remarkable performance with an AUROC of 0.823 and AUPRC of 0.918, aligning closely with decisions made by genetic specialists, and demonstrated strong generalizability (AUROC:0.77, AUPRC: 0.816) in an external cohort, indicating its potential value for general pediatricians to expedite rare disease diagnosis by enhancing genetic test ordering.
- 21 November 2024
- Published by Springer Nature in npj Digital Medicine
- Vol. 7 (1) , 1-12
Open AccessPhenotype driven molecular genetic test recommendation for diagnosing pediatric rare disorders
Patients with rare diseases often experience prolonged diagnostic delays. Ordering appropriate genetic tests is crucial yet challenging, especially for general pediatricians without genetic expertise. Recent American College of Medical Genetics (ACMG) guidelines embrace early use of exome sequencing (ES) or genome sequencing (GS) for conditions like congenital anomalies or developmental delays while still recommend gene panels for patients exhibiting strong manifestations of a specific disease. Recognizing the difficulty in navigating these options, we developed a machine learning model trained on 1005 patient records from Columbia University Irving Medical Center to recommend appropriate genetic tests based on the phenotype information. The model achieved a remarkable performance with an AUROC of 0.823 and AUPRC of 0.918, aligning closely with decisions made by genetic specialists, and demonstrated strong generalizability (AUROC:0.77, AUPRC: 0.816) in an external cohort, indicating its potential value for general pediatricians to expedite rare disease diagnosis by enhancing genetic test ordering.
- 9 November 2024
- Published by Oxford University Press (OUP) in Journal of the American Medical Informatics Association
- Vol. 32 (1) , 119-128
Mini-mental status examination phenotyping for Alzheimer’s disease patients using both structured and narrative electronic health record features
Objective: This study aims to automate the prediction of Mini-Mental State Examination (MMSE) scores, a widely adopted standard for cognitive assessment in patients with Alzheimer’s disease, using natural language processing (NLP) and machine learning (ML) on structured and unstructured EHR data. Materials and Methods: We extracted demographic data, diagnoses, medications, and unstructured clinical visit notes from the EHRs. We used Latent Dirichlet Allocation (LDA) for topic modeling and Term-Frequency Inverse Document Frequency (TF-IDF) for n-grams. In addition, we extracted meta-features such as age, ethnicity, and race. Model training and evaluation employed eXtreme Gradient Boosting (XGBoost), Stochastic Gradient Descent Regressor (SGDRegressor), and Multi-Layer Perceptron (MLP). Results: We analyzed 1654 clinical visit notes collected between September 2019 and June 2023 for 1000 Alzheimer’s disease patients. The average MMSE score was 20, with patients averaging 76.4 years old, 54.7% female, and 54.7% identifying as White. The best-performing model (ie, lowest root mean squared error (RMSE)) is MLP, which achieved an RMSE of 5.53 on the validation set using n-grams, indicating superior prediction performance over other models and feature sets. The RMSE on the test set was 5.85. Discussion: This study developed a ML method to predict MMSE scores from unstructured clinical notes, demonstrating the feasibility of utilizing NLP to support cognitive assessment. Future work should focus on refining the model and evaluating its clinical relevance across diverse settings. Conclusion: We contributed a model for automating MMSE estimation using EHR features, potentially transforming cognitive assessment for Alzheimer’s patients and paving the way for more informed clinical decisions and cohort identification.
- 9 November 2024
- Published by Oxford University Press (OUP) in Journal of the American Medical Informatics Association
- Vol. 32 (1) , 119-128
Mini-mental status examination phenotyping for Alzheimer’s disease patients using both structured and narrative electronic health record features
Objective: This study aims to automate the prediction of Mini-Mental State Examination (MMSE) scores, a widely adopted standard for cognitive assessment in patients with Alzheimer’s disease, using natural language processing (NLP) and machine learning (ML) on structured and unstructured EHR data. Materials and Methods: We extracted demographic data, diagnoses, medications, and unstructured clinical visit notes from the EHRs. We used Latent Dirichlet Allocation (LDA) for topic modeling and Term-Frequency Inverse Document Frequency (TF-IDF) for n-grams. In addition, we extracted meta-features such as age, ethnicity, and race. Model training and evaluation employed eXtreme Gradient Boosting (XGBoost), Stochastic Gradient Descent Regressor (SGDRegressor), and Multi-Layer Perceptron (MLP). Results: We analyzed 1654 clinical visit notes collected between September 2019 and June 2023 for 1000 Alzheimer’s disease patients. The average MMSE score was 20, with patients averaging 76.4 years old, 54.7% female, and 54.7% identifying as White. The best-performing model (ie, lowest root mean squared error (RMSE)) is MLP, which achieved an RMSE of 5.53 on the validation set using n-grams, indicating superior prediction performance over other models and feature sets. The RMSE on the test set was 5.85. Discussion: This study developed a ML method to predict MMSE scores from unstructured clinical notes, demonstrating the feasibility of utilizing NLP to support cognitive assessment. Future work should focus on refining the model and evaluating its clinical relevance across diverse settings. Conclusion: We contributed a model for automating MMSE estimation using EHR features, potentially transforming cognitive assessment for Alzheimer’s patients and paving the way for more informed clinical decisions and cohort identification.