AI Solves Leonardo da Vinci’s 500-Year-Old Heart Mystery

Scientists have finally solved a 500-year-old mystery about the human heart first sketched by Leonardo da Vinci, using artificial intelligence and over 25,000 MRI scans.

The breakthrough, reported by researchers from Charles University in Prague and the Czech Technical University, confirms da Vinci’s hypothesis that the heart’s trabeculae — the intricate mesh of muscular ridges inside the ventricles — play a crucial role in optimizing blood flow, not merely providing structural support as previously thought.

Leonardo’s 16th-century anatomical drawings, based on dissections of oxen and human cadavers, depicted these trabeculae as a fractal-like network. For centuries, their function remained debated, with many assuming they were incidental byproducts of heart development.

Using AI-driven image analysis on a dataset of 25,000 cardiac MRI scans from the UK Biobank, the team mapped trabecular morphology across thousands of individuals and correlated it with genetic data, heart function metrics, and clinical outcomes.

The analysis revealed that specific fractal patterns in the trabeculae are significantly associated with more efficient blood flow dynamics and lower risk of heart failure. Individuals with highly branched trabecular networks showed improved ventricular performance, particularly during diastole — the phase when the heart relaxes and fills with blood.

Genome-wide association studies further identified multiple genetic loci linked to trabecular complexity, suggesting a biological basis for the variation observed across the population.

The findings, published in the journal Nature Communications, represent one of the first large-scale validations of a historical anatomical hypothesis using modern computational methods and imaging at this scale.

Dr. Tomáš Križek, lead author on the study, noted that da Vinci’s insight was remarkably prescient: “He didn’t just draw what he saw — he inferred function from form. Our data shows he was essentially correct.”

The research team emphasized that while the AI models uncovered strong statistical associations, further experimental work is needed to establish causal mechanisms — particularly how trabecular geometry influences fluid dynamics at the molecular and cellular levels.

No clinical applications are currently implied by the study, and the researchers cautioned against overinterpreting the results as diagnostic tools. The work remains foundational, aimed at deepening understanding of cardiac morphology and its evolutionary significance.

The study was conducted in collaboration with institutions in Prague and supported by grants from the Czech Science Foundation and the European Regional Development Fund. No conflicts of interest were reported.