Biomaterials Design through Machine Learning for Sex- and Disease-Specific Bone Regeneration

Osteoporosis, a widespread disease that causes bone fragility, currently afflicts over 32 million Europeans, with rapidly increasingprevalence expected due to population aging. The costs of therapies for osteoporotic fragility fractures total more than 50 billion peryear in the EU.

Postmenopausal females are at especially increased risk for osteoporosis, and the regenerative capacity of healthybone also differs between the biological sexes.Although biomaterials have been developed to aid bone regeneration after injury, the influence of biological sex and osteoporoticdisease have not been taken into account for such biomaterials design.

Excitingly, the advent of machine learning (ML) offers thepotential to unravel the effects of sex and osteoporotic disease on biomaterials bone regenerative capacity, and to apply theseinsights to design biomaterials optimized to these characteristics.Therefore, the SeDiBone project will for the first time exploit ML to elucidate the influence of sex and osteoporotic disease onbone regeneration, and to translate these insights towards design of an entirely new class of biomaterials with sex and osteoporoticspecificity.

By enabling automated localization of tissue defect sites and detailed tissue, cell, and interface identification, SeDiBone will deliver an Analysis Tool that efficiently and reproducibly quantifies bone regeneration outcomes in male or female and healthy or osteoporotic preclinical animal models (Aim 1).

These in-depth insights will then serve as inputs for a Design Tool to determine the biomaterials designs correlating to optimal regenerative success based on sex and osteoporosis, with validation through ex vivo demonstration (Aim 2).

SeDiBone will thus crucially synergize and advance efforts to design a new generation of patient-specific biomaterials, and accordingly inform therapeutic solutions for regeneration of bone tissue in the unexplored areas of sex- and disease-based specificity.