Machine learning-guided risk stratification for long QT syndrome genetic variants with hiPSC-derived cardiomyocytes

Area of Science:

• Cardiovascular Genetics
• Stem Cell Biology
• Machine Learning in Medicine

Background:

• Long QT syndrome (LQTS) is a genetic disorder causing dangerous heart rhythms.
• Current risk stratification for LQTS patients with pathogenic variants is insufficient.
• High-risk variants require improved identification for better patient management.

Purpose of the Study:

• To develop a refined risk stratification model for LQTS.
• Utilize machine learning on electrophysiological data from patient-specific cells.
• Improve identification and management of patients with high-risk genetic variants.

Main Methods:

• Generated patient-specific human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) from LQTS patients.
• Recorded electrophysiological responses using multielectrode arrays.
• Trained a machine learning classifier on in vitro electrophysiological readouts.

Main Results:

• hiPSC-CM electrophysiological profiles correlated with variant risk levels.
• Machine learning model achieved 89% accuracy in classifying variant risk.
• Demonstrated predictive performance on an independent validation cohort.

Conclusions:

• Integrating hiPSC-CM electrophysiology and machine learning offers robust risk stratification for LQTS.
• This approach enables granular, variant-specific risk assessment.
• Enhances the management of patients with LQTS genetic variants.