Melicoccus bijugatus (Guinep) is traditionally consumed in the Caribbean and Latin America for its health benefits, yet its cardiovascular effects remain underexplored. This study investigated the therapeutic potential of Guinep by combining experimental and computational approaches. The biological evaluation of the Guinep extract was conducted by assessing the effects of modulating Angiotensin-Converting Enzyme (ACE), Angiotensin II Type 1 Receptor (AT1R), and Voltage-Gated Calcium Channels (VGCC) on vascular reactivity. Metabolites previously identified by high-resolution UHPLC-Q-Orbitrap mass spectrometry were further examined using in silico tools, including ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) prediction (pkCSM), biological activity prediction (PASS server), and molecular docking (AutoDock Vina) against cardiovascular targets (ACE: PDB 1O86, AT1R: PDB 4ZUD, VGCC: PDB 8WE8). Docking results revealed that phytochemicals such as isorhamnetin-3-O-glucoside and 3-O-caffeoylquinic acid displayed strong binding affinities with ACE (−9.3 and −8.5 kcal/mol), AT1R (−8.2 and −7.6 kcal/mol), and VGCC (−8.6 and −7.6 kcal/mol), in several cases matching or surpassing standard antihypertensive drugs. Key hydrogen bond interactions closely resembled those of reference ligands, suggesting pharmacophoric similarity. ADMET modeling confirmed favorable pharmacokinetic profiles and low predicted toxicity, supporting their drug-like potential. These findings are consistent with in vivo evidence of Guinep’s hypotensive, antioxidant, and vasodilatory properties. Vascular relaxation of Guinep extract was predominantly mediated by blockade of VGCC (53%) and AT1R (48%), while ACE inhibition accounted for 24%. Collectively, the results demonstrate that Guinep contains bioactive phytochemicals with multitarget cardiovascular activity, particularly as ACE, AT1R, and VGCC modulators. This study validates the traditional use of Guinep.

Machine Learning (ML) has demonstrated strong predictive capabilities in healthcare, often surpassing human performance in pattern recognition and decision-making. However, many high-performing models lack interpretability, which is critical in clinical settings where understanding and trusting predictions is essential. To address this, we introduce Multi-Perspective Machine Learning (MPML)—an ensemble approach that integrates multiple complementary techniques to form a high-performing, interpretable model. MPML organizes features into meaningful subsets, or perspectives, enabling both global and instance-level interpretability. Unlike traditional ensemble methods such as Bagging, Boosting, and Random Forest, MPML delivers significantly higher-quality predictions across all evaluation metrics while maintaining a transparent structure. Applied to a heart disease dataset, MPML not only improves predictive accuracy but also provides detailed, accessible explanations for individual patient outcomes, advancing the potential for practical and ethical deployment of ML in healthcare.