基于半监督学习的骨科植入物推荐系统应用研究

Application research of bone implant recommendation system based on semi-supervised learning

硅/银羟基磷灰石涂层分别能提高金属植入物生物相容性与抗菌性。利用"Wide&Deep"和密集网络结构推荐系统,结合半监督学习训练推荐骨科植入材料比例,为每位患者提供针对性骨病治疗与康复方案。系统能解决病例特征多类型、标注数据量少的复杂问题,获得优化多特征融合网络结构的半监督学习优化算法,在测试数据集上获得高准确度的骨科植入方案推荐,为医生提供临床智能决策支持。

Silicon/silver hydroxyapatite bioactive coatings can effectively improve the biocompatibility and antibacterial properties of metal implants,respectively. The method utilizes‘Wide & Deep ’and Dense Net architecture and take advantage of semi-supervised learning train recommend bone implant ratios to provide an individual treatment and rehabilitation program for each patient. The recommendation system can solve the complex problem of less labelled data and multiple features with each patient,and obtain an optimized semi-supervised learning algorithm with the multi-feature fusion network. It can achieve high accuracy in realizing recommendation orthopaedic implant treatment scheme on test dataset,and provide intelligent decision support for clinical treatments.