Artificial intelligence for phenotyping tetralogy of Fallot patients from electrocardiographic recordings

Abstract Background Tetralogy of Fallot (ToF) remains associated with significant morbidity and mortality. Artificial intelligence (AI) is a viable tool for identifying markers: deep learning (DL) can be used to automate measurements on the ECG trace dimensionality reduction (DR) algorithms grouping patients based ECG, imaging genetic markers. Such an AI-based pipeline aid in phenotyping by characteristics that correlate outcome. Methods A cohort ToF were recruited study (n=388). All underwent echo- electrocardiographic exams, had recorded information outcome (death, heart failure [HF], history ventricular tachycardia [VT] or atrial fibrillation [AF]), lifestyle markers (left ejection fraction [LVEF] NYHA score, among others). The analysis population consisted three steps (Figure 1A). Firstly, was delineated using DL model obtain P, QRS T onsets/offsets all cardiac cycles [1]. Secondly, most stable heartbeat selected, their morphology, usage into DR algorithm [2]. This allowed combining different leads, automatically assess inter-patient similarities. Thirdly, clustered respect as extracted previous step, said clusters correlated Results AI identify subset higher risk blue, green red contained right bundle branch block (RBBB) but morphologies, whereas orange cluster relatively normal morphology (no RBBB) lowest width 1B). With events (Table 1), significantly increased all-type negative (36.94%, p-value <0.0001). occurrence death AF. green, blue density VT. Despite red, presenting RBBB wide complex, these show very event rates, hinting at key stratify risk: higher-risk showed more fractionation reduced amplitude. Finally, it important note good correspondence despite low correlation (TAPSE, RV diameter), complementariness both modalities. Conclusion help clinical interest ToF, given its ability agglomerate morphological leads. In this work, data, allowing identification subgroup VT, AF, HF). shows pattern are not sole factors might affect Funding Acknowledgement Type funding sources: Public grant(s) – National budget only. Main source(s): Fundaciό La Maratό de TV3