Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
July 9, 2025PLoS ONEOpen Access

Unsupervised machine learning identifies 5 distinct hypertension subphenotypes with varying risk profiles.

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Hypertension is heterogeneous in presentation and treatment response, and identifying distinct subphenotypes may improve understanding of underlying mechanisms and guide more precise treatment or public health initiatives.

Population

40,686 adult Floridians with newly diagnosed HTN

Design

Cohort study using hierarchical clustering unsupervised machine learning

Key result

Unsupervised machine learning identified 5 distinct hypertension subphenotypes varying in demographic, socioeconomic, and risk profiles among 40,686 newly diagnosed adult patients.

Authors

JHJaclyn HallJYJie YuMWMarta Walsh

Discussion

Loading...

Member takes

Overview

Does not support immediate changes in hypertension care; leaves open whether subphenotypes predict outcomes or guide therapy.

Study Design

Type

Observational (n=40,686)

Multicenter

Yes

Structured PICO

P
Population
40,686 adult Floridians with newly diagnosed HTN (first diagnosis following two outpatient blood pressures ≥140/90 mmHg & no prior anti-HTN treatment), mean age 60.9, 55% women.
I
Intervention
Hierarchical clustering (unsupervised machine learning) using EHR and Medicaid claims data
O
Outcome
Identification of distinct subphenotypes within the HTN population

Unsupervised machine learning identified five distinct subphenotypes of newly diagnosed hypertension, highlighting significant heterogeneity in demographic, clinical, and socioeconomic profiles.

Cite This Study

Hall et al. (2025) conducted an observational in Newly diagnosed hypertension (n=40,686). Hierarchical clustering (unsupervised machine learning) was evaluated on Identification of distinct subphenotypes within the hypertension population. Unsupervised machine learning identified 5 distinct hypertension subphenotypes varying in demographic, socioeconomic, and risk profiles among 40,686 newly diagnosed adult patients.

synapsesocial.com/papers/6a13cd83bc9c1e8ad339dd73https://doi.org/10.1371/journal.pone.0326776
View Full Paper
Ask AI
Bookmark
Share