Estimating cardiac contraction through high resolution data assimilation of a personalized mechanical model

Cardiac computational models, individually personalized, can provide clinicians with useful diagnostic information and aid in treatment planning. A major bottleneck in this process can be determining model parameters to fit created models to individual patient data. However, adjoint-based data assimilation techniques can now rapidly estimate high dimensional parameter sets. This method is used on a cohort of heart failure patients, capturing cardiac mechanical information and comparing it with a healthy control group. Excellent fit (R2 ≥ 0.95) to systolic strains is obtained, and analysis shows a significant difference in estimated contractility between the two groups.