Data-driven discovery of probable Alzheimer's disease and related dementia subphenotypes using electronic health records

Topic: Amyloid, tau and blood-test diagnosis

Introduction We sought to assess longitudinal electronic health records (EHRs) using machine learning (ML) methods to computationally derive probable Alzheimer's Disease (AD) and related dementia subphenotypes. Methods A retrospective analysis of EHR data from a cohort of 7587 patients seen at a large, multi-specialty urban academic medical center in New York was conducted. Subphenotypes were derived using hierarchical clustering from 792 probable AD patients (cases) who had received at least one diagnosis of AD using their clinical data. The other 6795 patients, labeled as controls, were matched on age and gender with the cases and randomly selected in the ratio of 9:1. Prediction models with multiple ML algorithms were trained on this cohort using 5-fold cross-validation. XGBoost was used to rank the variable importance. Results Four subphenotypes were computationally derived. Subphenotype A (n = 273; 28.2%) had more patients with cardiovascular diseases; subphenotype B (n = 221; 27.9%) had more patients with mental health illnesses, such as depression and anxiety; patients in subphenotype C (n = 183; 23.1%) were overall older (mean (SD) age, 79.5 (5.4) years) and had the most comorbidities including diabetes, cardiovascular diseases, and mental health disorders; and subphenotype D (n = 115; 14.5%) included patients who took anti-dementia drugs and had sensory problems, such as deafness and hearing impairment.The 0-year prediction model for AD risk achieved an area under the receiver operating curve (AUC) of 0.764 (SD: 0.02); the 6-month model, 0.751 (SD: 0.02); the 1-year model, 0.752 (SD: 0.02); the 2-year model, 0.749 (SD: 0.03); and the 3-year model, 0.735 (SD: 0.03), respectively. Based on variable importance, the top-ranked comorbidities included depression, stroke/transient ischemic attack, hypertension, anxiety, mobility impairments, and atrial fibrillation. The top-ranked medications included anti-dementia drugs, antipsychotics, antiepileptics, and antidepressants. Conclusions Four subphenotypes were computationally derived that correlated with cardiovascular diseases and mental health illnesses. ML algorithms based on patient demographics, diagnosis, and treatment demonstrated promising results in predicting the risk of developing AD at different time points across an individual's lifespan.

@article{xu2020-33083543,
  author  = {Xu, J. and Wang, F. and Xu, Z. and Adekkanattu, P. and Brandt, P. and Jiang, G. and Kiefer, R. C. and Luo, Y. and others},
  title   = {{Data-driven discovery of probable Alzheimer's disease and related dementia subphenotypes using electronic health records}},
  journal = {Learning health systems},
  year    = {2020},
  doi     = {10.1002/lrh2.10246},
  pmid    = {33083543},
  url     = {https://doi.org/10.1002/lrh2.10246}
}

Key findings

• Results Four subphenotypes were computationally derived.
• Subphenotype A (n = 273; 28.2%) had more patients with cardiovascular diseases; subphenotype B (n = 221; 27.9%) had more patients with mental health illnesses, such as depression and anxiety; patients in subphenotype C (n = 183; 23.1%) were overall older (mean (SD) age, 79.5 (5.4) years) and had the most comorbidities including diabetes, cardiovascular diseases, and mental health disorders; and subphenotype D (n = 115; 14.5%) included patients who took anti-dementia drugs and had sensory problems, such as deafness and hearing impairment.The 0-year prediction model for AD risk achieved an area under the receiver operating curve (AUC) of 0.764 (SD: 0.02); the 6-month model, 0.751 (SD: 0.02); the 1-year model, 0.752 (SD: 0.02); the 2-year model, 0.749 (SD: 0.03); and the 3-year model, 0.735 (SD: 0.03), respectively.
• ML algorithms based on patient demographics, diagnosis, and treatment demonstrated promising results in predicting the risk of developing AD at different time points across an individual's lifespan.

Why it matters

Using electronic health records and machine learning, this study derived four probable Alzheimer's disease subphenotypes and found prediction models showed promising results for predicting risk. It is about characterizing and predicting disease and does not report any treatment, cure, or prevention.

New to this subject? Start with our plain-English guide: Digital health →

Study details

• Study designlongitudinal study
• Participants273
• Effect size28.2%; 27.9%; 23.1%
• Evidence strengthModerate

Read the original

• DOI: 10.1002/lrh2.10246
• PubMed: 33083543
• Original source →
• Free full text: Learning Health Systems →

Full text: free to read — OpenAlex lists this copy as free to read: the link above (checked 6 Feb 1020).

Related

Studies in the library that share something with this one. The line above each title says what.

• shares topic tags: machine, mri, classification FDA approval of the Lumipulse G pTau217/β-amyloid 1-42 plasma ratio test: a new era in accessible Alzheimer's diagnosis Annals of Medicine and Surgery · 2025-12-17 · usefulness score 42/100
• shares topic tags: machine, mri, classification Amyloid-Related Imaging Abnormalities (ARIA) in Anti-Amyloid Therapies for Alzheimer's Disease: An Update from the Alzheimer's Association ARIA Workgroup Alzheimer's & Dementia · 2026-04-21 · usefulness score 69/100
• shares topic tags: machine, mri, classification Tablet-based cognitive assessment in a multisite study of early-onset AD: association with pen-and-paper tests and PET measures of amyloid and tau Alzheimer's & Dementia (AAIC 2025 conference abstract) · 2025 · usefulness score 16/100
• shares topic tags: machine, mri, classification Amyloid PET Quantitation and Centiloid Thresholds in the Diagnosis of Alzheimer Disease: An Individual Participant Data Meta-Analysis JAMA · 2026-07-13 · usefulness score 79/100

Studies this one cites

Referenced by this study and also in the alzhub library, so you can read them here.

• Data-driven models of dominantly-inherited Alzheimer's disease progression Brain : a journal of neurology · 2018-05-01 · usefulness score 67/100 · Risk genes beyond APOE
• Machine learning prediction of incidence of Alzheimer's disease using large-scale administrative health data NPJ digital medicine · 2020-03-26 · usefulness score 59/100 · Digital health tools

Later studies that cite this one

2 studies in the alzhub library list this one in their references.

• Advancing dementia identification using machine learning and real‐world sequential health data Alzheimer s & Dementia Diagnosis Assessment & Disease Monitoring · 2026-04-01 · usefulness score 53/100 · Digital health tools
• Artificial intelligence and open science in discovery of disease-modifying medicines for Alzheimer's disease Cell reports. Medicine · 2024-02-01 · usefulness score 50/100 · Digital health tools

Related studies in Digital health tools

• Detection of probable dementia cases in undiagnosed patients using structured and unstructured electronic health records BMC medical informatics and decision making · 2019-07-09 · usefulness score 53/100
• Unsupervised machine learning for the discovery of latent disease clusters and patient subgroups using electronic health records Journal of biomedical informatics · 2019-12-28 · usefulness score 43/100
• Identification of Physician-Diagnosed Alzheimer's Disease and Related Dementias in Population-Based Administrative Data: A Validation Study Using Family Physicians' Electronic Medical Records Journal of Alzheimer's disease : JAD · 2016-08-01 · usefulness score 50/100
• Multiple morbidity across the lifespan in people with Down syndrome or intellectual disabilities: a population-based cohort study using electronic health records The Lancet. Public health · 2023-04-26 · usefulness score 55/100
• Antidepressant Use at the Threshold: using electronic health records to characterise people prescribed antidepressants around the time of dementia diagnosis medRxiv · 2025-12-17 · usefulness score 16/100
• Stratifying risk for dementia onset using large-scale electronic health record data: A retrospective cohort study Alzheimer's & dementia : the journal of the Alzheimer's Association · 2020-01-16 · usefulness score 62/100

Also tagged “machine”

834 studies in the library carry the topic tag “machine”; these are this study's neighbours in a fixed, alphabetical-by-hash tour of them, not the closest in content. All 834 tagged studies.

• Sex-driven modifiers of Alzheimer risk: A multimodality brain imaging study Neurology · 2020-06-24 · usefulness score 53/100 · Genetics
• Using voxel-based morphometry to map the structural changes associated with rapid conversion in MCI: a longitudinal MRI study NeuroImage · 2005-10-01 · usefulness score 63/100 · Mild cognitive impairment
• Physical and Cognitive Activities and Trajectories of AD Neuroimaging Biomarkers Neurology · 2025-12-02 · usefulness score 50/100 · Disease mechanisms

Studies sharing this one's topic tags

• Utility of the Amygdalar Atrophy Scale to Identify Patients With Limbic-Predominant Age-Related TDP-43 Encephalopathy in a Memory Clinic Neurology · 2025-12-22 · usefulness score 56/100 · shares: machine, mri, classification
• A/T/N: An unbiased descriptive classification scheme for Alzheimer disease biomarkers Neurology · 2016-07-01 · usefulness score 64/100 · shares: machine, mri, classification
• Prevalence of amyloid PET positivity in dementia syndromes: a meta-analysis JAMA · 2015-05-01 · usefulness score 82/100 · shares: machine, mri, classification
• Association of Amyloid Positron Emission Tomography With Subsequent Change in Clinical Management Among Medicare Beneficiaries With Mild Cognitive Impairment or Dementia JAMA · 2019-04-01 · usefulness score 76/100 · shares: machine, mri, classification
• Appropriate use criteria for amyloid PET: a report of the Amyloid Imaging Task Force, the Society of Nuclear Medicine and Molecular Imaging, and the Alzheimer's Association Alzheimer's & dementia : the journal of the Alzheimer's Association · 2013-01-01 · usefulness score 55/100 · shares: machine, mri, classification

Similar studies

• Detection and Classification of Alzheimer’s Disease Using Deep and Machine Learning 2025 · Digital health tools
• AI approaches for phenotyping Alzheimer's disease and related dementias using electronic health records 2025 · Getting a diagnosis
• Classification and deep-learning-based prediction of Alzheimer disease subtypes by using genomic data 2023 · Biomarkers
• Using machine learning and electronic health record (EHR) data for the early prediction of Alzheimer's Disease and Related Dementias 2025 · How common it is
• Early Detection of Alzheimers Disease Using Deep Learning 2025 · Brain scans and imaging
• Data-driven subtypes of Alzheimer's Disease: A multicohort study 2026 · Getting a diagnosis

More on Digital health tools

• A review that covers this ground The validity of dementia diagnoses in routinely collected electronic health records in the United Kingdom: A systematic review Pharmacoepidemiology and drug safety · 2019-01-22 · usefulness score 62/100
• Newest in this subject Speaker-Wise Dementia Screening fromNaturalistic Speech with Lexical Entropy, Semantic Drift, and Multimodal Fusion Research Square · 2026-10-21 · Preprint — not yet peer-reviewed · usefulness score 16/100
• Newest in this subject Integrating Multisource Target Evidence with Machine-Learning Potency Prediction for Alzheimer’s Disease Drug Discovery bioRxiv · 2026-10-20 · Preprint — not yet peer-reviewed · usefulness score 16/100

See all studies in Digital health tools →