STUDY OF THE PATHOGENESIS AND TREATMENT OF HEART FAILURE THROUGH THE HARMONIZATION OF EVIDENCE-BASED CLINICAL DATA, ARTIFICIAL INTELLIGENCE AND ANIMAL MODELS

Background: Heart failure (HF) is a major cause of death and hospitalization, especially in diabetic patients. The cornerstone of medical treatment for HF is represented by Sodium-glucose cotransporter type 2 inhibitors (SGLT2i), gliflozins, especially in heart failure with reduced left ventricular ejection fraction (HFrEF). Nevertheless, the effects of SGLT2i on ventricular remodeling and function have not been completely understood yet. Explainable artificial intelligence represents an unprecedented explorative option to clinical research in this field. Based on echocardiographic evaluations, our first aim was to identify some key clinical responses to gliflozins by employing a machine learning approach. In the setting of HF, atrial fibrillation (AF) represents an extremely frequent arrhythmia. Oral anticoagulant therapy (OAT) for managing AF encompasses vitamin K antagonists (VKAs) and direct-acting oral anticoagulants. Due to the lower risk of major bleeding associated with DOACs, anticoagulant switching is a common practice in AF patients. Nevertheless, there are issues related to OAT switching that still need to be fully understood, especially for patients in whom AF and heart failure (HF) coexist. As second aim, we therefore sought to assess the real positive and pleiotropic effects mediated by DOACs in addition to their anticoagulant activity. In particular, we aimed at the effective impact of the therapeutic switching from warfarin to DOACs in HF patients with AF by a machine learning (ML) analysis of a clinical database. Methods: for the first aim, we assessed 78 consecutive diabetic outpatients followed for HFrEF, using a random forest classification. A single subject analysis was performed to define the profile of patients treated with gliflozins. Moreover, an explainability analysis using Shapley values was used to outline clinical parameters that mostly improved after gliflozin therapy and machine learning runs highlighted specific variables predictive of gliflozin response. For the second aim, 42 consecutive outpatients with HFrEF and AF in OAT for at least one year were enrolled. The k-means clustering method and the Random Forest learning algorithm were adopted in order to evaluate how switching from warfarin to DOACs may affect the clinical progression of the patients. Results: The five-fold cross-validation analyses showed that gliflozins patients can be identified with a 0.70 ± 0.03% accuracy. The most relevant parameters distinguishing gliflozins patients were Right Ventricular S'-Velocity, Left Ventricular End Systolic Diameter and E/e' ratio. In addition, low Tricuspid Annular Plane Systolic Excursion values along with high Left Ventricular End Systolic Diameter and End Diastolic Volume values were associated with lower gliflozin efficacy in terms of anti-remodeling effects. In patients with AF and HFrEF, at the baseline, 75% of patients were correctly separated. At the follow-up, after the switch to DOACs, this accuracy decreased to 64%. The baseline model is more accurate, achieving an average accuracy of 78%. At follow-up, the accuracy decreases to 58%. The accuracy loss of about 20% is statistically significant and suggests a fundamental loss of the features’ discriminative power. N-terminal pro-brain natriuretic peptide (NTproBNP)brought a fundamental contribution in discriminating the clinical cohorts. Discussion and Conclusions: A ML analysis on a population of diabetic patients with HFrEF showed that SGLT2i treatment improved left ventricular remodeling. This cardiovascular response may be predicted with an explainable artificial intelligence approach, suggesting a lower efficacy in case of advanced stages of cardiac remodeling. Our ML analysis in patients with AF and HFrEF demonstrated that, when warfarin-treated patients switched to DOACs, they were no longer differentiable. This means that DOACs somehow modify the considered features which have specific clinical significance