Human-Guided Feature Selection for Accurate Cardiomyocyte Dysfunction Classification.

Area of Science:

• Cardiovascular Biology
• Computational Biology
• Biomedical Engineering

Background:

• Diastolic heart failure (DHF) diagnosis hinges on early identification of cardiomyocyte dysfunction, specifically impaired left ventricular relaxation (ILVR).
• Intracellular calcium handling is vital for myocardial relaxation; impaired calcium removal during diastole leads to ILVR.
• Analyzing sarcomere length (SL) and intracellular calcium kinetics (CK) is essential for cellular-level relaxation characterization, but data complexity poses challenges.

Purpose of the Study:

• To develop a robust feature selection pipeline for identifying informative features from SL and CK data.
• To create an effective classifier for early detection of cardiomyocyte dysfunction using reduced feature sets.
• To compare the performance of a selected feature set against full datasets and PCA-derived features.

Main Methods:

• Utilized statistical significance testing (p-values), hierarchical clustering, and random forest (RF) classification for feature selection.
• Obtained SL and CK transients from a transgenic mouse model with ILVR (AAA mice) and wild-type controls (NTG).
• Trained and evaluated RF classifiers using reduced feature sets, full feature sets, and principal component analysis (PCA)-based dimension reduction.

Main Results:

• A reduced feature set, selected through the pipeline, achieved classification performance comparable to the full feature set.
• The selected feature set outperformed PCA-based dimension reduction in classifying cardiomyocyte dysfunction.
• The reduced feature set enhanced interpretability by retaining biologically relevant features.

Conclusions:

• A carefully selected, small set of biological features can effectively detect early signs of cardiomyocyte dysfunction.
• The developed feature selection pipeline offers a robust method for analyzing complex SL and CK data.
• This approach aids in the early prognosis of diastolic heart failure by identifying cellular-level abnormalities.