AI-based cluster analysis enables outcomes prediction among patients with increased LVM

TL;DR

This study uses AI to identify four distinct groups of patients with increased left ventricular mass, each with different risks and characteristics, improving outcomes prediction.

Contribution

The paper introduces a novel AI-based clustering method to classify patients with increased LVM and correlate these clusters with clinical outcomes.

Findings

Four distinct clusters of patients with increased LVM were identified using unsupervised cluster analysis.

Each cluster showed unique clinical characteristics and significantly different risks for major cardiac events.

The method provides a more accurate way to predict outcomes compared to traditional LVH classification.

Abstract

The traditional classification of left ventricular hypertrophy (LVH), which relies on left ventricular geometry, fails to correlate with outcomes among patients with increased LV mass (LVM).

To identify unique clinical phenotypes of increased LVM patients using unsupervised cluster analysis, and to explore their association with clinical outcomes.

Among the UK Biobank participants, increased LVM was defined as LVM index ≥72 g/m2 for men, and LVM index ≥55 g/m2 for women. Baseline demographic, clinical, and laboratory data were collected from the database. Using Ward's minimum variance method, patients were clustered based on 27 variables. The primary outcome was a composite of all-cause mortality with heart failure (HF) admissions, ventricular arrhythmia, and atrial fibrillation (AF). Cox proportional hazard model and Kaplan-Meier survival analysis were applied.

Increased LVM was found in 4,255 individuals, with an average age of 64 ± 7 years. Of these patients, 2,447 (58%) were women. Through cluster analysis, four distinct subgroups were identified. Over a median follow-up period of 5 years (IQR: 4-6), 100 patients (2%) died, 118 (2.8%) were admissioned due to HF, 29 (0.7%) were admissioned due to VA, and 208 (5%) were admissioned due to AF. Univariate Cox analysis demonstrated significantly elevated risks of major events for patients in the 2nd (HR = 1.6; 95% CI 1.2–2.16; p < .001), 3rd (HR = 2.04; 95% CI 1.49–2.78; p < .001), and 4th (HR = 2.64; 95% CI 1.92–3.62; p < .001) clusters compared to the 1st cluster. Further exploration of each cluster revealed unique clinical phenotypes: Cluster 2 comprised mostly overweight women with a high prevalence of chronic lung disease; Cluster 3 consisted mostly of men with a heightened burden of comorbidities; and Cluster 4, mostly men, exhibited the most abnormal cardiac measures.

Unsupervised cluster analysis identified four outcomes-correlated clusters among patients with increased LVM. This phenotypic classification holds promise in offering valuable insights regarding clinical course and outcomes of patients with increased LVM.

Figures