Prediction of Outcome by Echocardiography in Left Bundle Branch Block

Prediction of Heart-failure and Mortality by Echocardiographic Parameters and Machine Learning in Individuals With Left Bundle Branch Block

Patients with left bundle branch block have an increased risk for the development of heart-failure and death. However, risk factors for unfavorable outcomes are still poorly defined. This study aims to identify echocardiographic parameters and ECG characteristics by machine learning in order to develop individual risk assessment

The project investigates patients with left bundle branch block (LBBB) which describes a specific block in the electrical conduction system, where the electrical impulses must follow a detour, with the result that different parts of the heart-muscle do not contract at the same time. This condition is called left ventricular dyssynchrony. LBBB can be found in people who are otherwise completely healthy and need not have any practical consequences. In others LBBB is present in patients with different heart diseases such as after myocardial infarctions or other diseases involving the heart-muscle. Patients with implanted pacemakers have a similar failure in the conduction system. Both conditions can increase the risk for development of heart-failure and cardiovascular death. Dyssynchrony can be treated with a special pacemaker (cardiac resynchronisation therapy, CRT) in addition to regular medical treatment. The therapy is well established and has shown to reduce morbidity and mortality and even reverse heart-failure in some patients completely. However, the patients in need and responding to CRT treatment is still not optimally defined. New echocardiographic parameters based on strain imaging such as regional myocardial work are able quantify the degree of dyssynchrony and give new insights into the interplay of activation delay through the LBBB and loading conditions and weakness of the myocardium due to other diseases. These new and complex measures can be integrated with clinical information by machine learning (ML) as a promising tools for accurate patient selection for CRT. The project aims to find markers on ultrasound improved by ML based selection to distinguish those patients who have problems associated with the branch block from those who remain stable. This will facilitate both, an optimized patient selection for CRT treatment and follow-up schedule for those who have a stable condition.

Inclusion Criteria:

• QRS complex >130 ms and R-wave duration in
• V6 >70 ms
• ventricular pacing>50%
• Previously implanted cardiac resynchronisation therapy (CRT)

Exclusion Criteria:

• Typical right bundle branch block.
• No ability to give informed consent,
• non-cardiovascular co-mobidities with reduced life-expectancy < 1 year
• patients with complex congenital heart disease.