Digital Prosthetics: Machine Learning and Human Challenges in Designing and Testing Next-Generation Personalised Limbs

尽管数字CAD/CAM设计工具和它们产生的数据提供了机会，但这项学生计划解决了假肢设计中效率低下和缺乏证据基础的主要挑战。这是对假肢插座设计和结果数据(舒适度和功能)的第一次机器学习分析，通过用户需求分析来最大限度地发挥临床翻译的潜力。由于残肢的稳定和适应，假体插座的提供可能永远是一个迭代的过程，特别是在新截肢的情况下(1，2)。然而，如果专家临床医生能够更好地利用精心选择和提供的数据，通过减少迭代可以减少相当大的成本、不适感和不便。这些数据可能建立在来自过去成功的临床实践的强有力的证据的组合上，无论是来自他们自己的实践还是来自经验丰富的同事，以及预测生物力学和优化方法(3)。这将旨在帮助修复师从事劳动密集型、低附加值的工作，使他们能够腾出更多时间，专注于最终的、高附加值的详细设计。

This studentship proposal addresses the major challenge of inefficiency and lack of evidence base in design of prosthetic limbs, despite the opportunities offered by digital CAD/CAM design tools and the data they generate. It is a first-of-kind machine learning analysis of prosthetic socket design and outcome data (comfort and function), informed by user-needs analysis to maximise potential for clinical translation.Prosthetic socket provision will probably always be an iterative process, due to stabilisation and adaptation of the residual limb, especially in newly amputated cases (1,2). However, considerable cost, discomfort and inconvenience could be saved by reducing iteration if expert clinicians were able to make better use of carefully selected and presented data. These data might be built upon a combination of a strong base of evidence from successful past clinical practice, either from their own practice or from experienced colleagues, alongside predictive biomechanics and optimisation approaches (3). This would be intended to assist the prosthetist in the labour-intensive, low value added aspects of their work, freeing them to spend more time and focus on the final, high value added, detailed design.