Artificial intelligence analysis of the sterilization effects on the mechanical properties of human cranial bones

Abstract

According to the literature, the patient’s own (autologous) bone remains the optimal graft material for cranioplasty. Autografts have an undoubted advantage over competing plastic materials of complete biological compatibility, as well as the absence of interstitial conflict and accessibility. Presently there is still not a single synthetic material that meets all the complex requirements for skull reconstruction in children due to the special complication of accommodating the skull structure that continues to grow. At the same time, the use of autograft is accompanied by the highest percentage of complications in the form of resorption and infection. The purpose of the present study was to evaluate the large volume of measurements (“big data”) obtained after nano-hardness testing of an autograft bone after various preliminary treatments. The latter included the currently optimal method of bone sterilization (boiling peroxide treatment). The measurements performed on different patient allografts were compared and the effects of bone treatment on the mechanical properties were evaluated and compared. Due to the complex hierarchical structural organisation of the bone, the correspondence between the composition, processing, structure, and properties of natural human bone materials were not obvious. Machine Learning, part of the Artificial Intelligence (AI) tool set, was applied to reveal the differences between scull bones of different types, treatment history and patients. It was found that after obtaining the principal mechanical properties of the chosen human skull bone sample it was possible to identify algorithmically the nature of prior bone sterilization and the patient’s sex.