A computational framework for personalising atrial fibrillation treatment to minimise heart failure risk

Atrial fibrillation (AF) is projected to affect one in three adults by 2060 . Patients with AF are at increased risk of heart failure (HF) and death. Indeed, more than half (57%) of subjects with new-onset HF already have AF and are associated with higher hospitalisation and mortality rates. Remarkably, AF alone represents 2% of the global healthcare expenditure. Catheter ablation is the most effective therapy for AF patients. However, 50% of patients experience recurrence after the first ablation, requiring additional and more extensive ablations. Paradoxically, while an extensive ablation lesion might be more effective for preventing AF, it could also result in atrial mechanical dysfunction, increasing the risk for HF. Thus, the ability to accurately identify patients in which ablation could lead to HF, and stratify them to optimal treatments to reduce the risk of both diseases would be of tremendous clinical value.In this fellowship, I propose to address the unmet need of personalising AF therapy to minimise the risk of HF (AFHF-TAILOR) by integrating whole-organ computer simulations powered by artificial intelligence (AI) and clinical data. The project takes in-silico studies to the next level, by integrating electro-mechanical simulations in thousands of virtual patients. Moreover, to ensure that the computer framework meets the needs for clinical translation, a dedicated AI framework will be trained on the simulations to enable rapid and accurate estimations of personalised cardiac mechanics. Two distinct clinical datasets, with important information such as cardiac imaging, catheter ablation lesions and mechanical biomarkers, will be used to externally validate the AI algorithm and the computer simulations. Accordingly, the project will provide a robust digital framework that balances ablation efficacy with the preservation of atrial function, ultimately reducing the risk of both diseases, AF and HF.