Left Atrial Strain in the Assessment of Diastolic Function in Heart Failure: A Machine Learning Approach

doi:10.1161/circimaging.122.014605

Left Atrial Strain in the Assessment of Diastolic Function in Heart Failure: A Machine Learning Approach

Circ Cardiovasc ImagingPublished 8 Feb 2023Retrospective cohort study with prospective validation cohort

Performance · Association with death/HF rehospitalization (multivariable models)ResultNet reclassification improvement 0.166 95% CI 0.035–0.276

Read the original article ↗PubMed

Cite

Authors' conclusion

Integrating PALS by machine learning algorithm in DD classification improves risk stratification over recommended current criteria, regardless of ejection fraction status.

Study context

Population: HF patients in sinus rhythm (HFrEF n=541; HF with mildly reduced EF n=129; HFpEF n=194)validation cohort n=189
Participants: 864 patients

Comparison: Machine learning diastolic dysfunction (DD)-cluster classification including peak atrial longitudinal strain (PALS) vs guideline-based DD classification
Outcome: Death/HF rehospitalization
Follow-up: 36.5±29.4 months

Methodological detailsmachine learning, classification and regression tree analysis

Machine learning
Classification and regression tree analysis

Performance

As reported in the abstract

Association with death/HF rehospitalization (multivariable models) · 36.5±29.4 months

Net reclassification improvement

0.166

95% CI 0.035–0.276

P = 0.013

By group

Guideline-based classification

0.720

C-index

DD-clusters

0.733

C-index

Events: 318

Further findings

Outcome	DD-clusters	Guideline-based classification	P value
Outcome prediction, validation cohort · median follow-up 18.5 months			P = 0.093
C-index	0.80	0.78

DD-clusters identified — Agreement with guidelines-recommended classification (kappa): 0.40

DD-clusters identified: cluster-1 212 events
DD-clusters identified: cluster-2 376 events
DD-clusters identified: cluster-3 DD 276 events
DD-clusters identified: P<0.001
DD-clusters identified: Clusters predicted by algorithm of E/A ratio, LA volume index, E/e' ratio and PALS
Outcome prediction, validation cohort: prospective cohort n=189

Original publication

Left Atrial Strain in the Assessment of Diastolic Function in Heart Failure: A Machine Learning Approach. Circ Cardiovasc Imaging. Published 8 February 2023. doi:10.1161/circimaging.122.014605

Read the original article ↗

Composed by AI from the published abstract and checked against it. How Briefs are made →