Using Bootstrap Aggregated Neural Networks for Peripheral Nerve Injury Treatment

摘要

Accidents and trauma can cause severe peripheral nerve injuries and may require surgical intervention. While autografts are considered the current gold standard for complete regeneration of damaged nerves: their scarcity, potential loss of function at the donor site, and potential mismatch in axon diameter limits their use in practice and begs the need for optimal nerve guidance conduits (NGCs), which are the current viable alternative. The major challenges in current NGC research is the inability to account for variations in gap lengths, materials, and enhancement factors. Also, there is an inability to estimate the performance of NGCs, without in vitro and in vivo studies, so that it may be optimized to achieve maximum recovery for an injury. We propose a prediction model based on bootstrap aggregated neural networks in this paper that addresses these challenges and can alleviate the conventional burdens involved in the development of an NGC.