Scaling Cardiac Biomechanics Digital Twins for Personalised Medicine

从飞机到汽车再到桥梁，建模和模拟在设计中扮演着重要的角色。然而，连接和计算方面的进步现在使模型能够直接链接到特定的对象或系统，创造了一个“数字孪生兄弟”。数字双胞胎代表特定对象的计算代理，随着更多信息的获得，数字双胞胎会随着时间的推移而更新。然而，数字双胞胎并不局限于人造物品。该项目旨在开发数字双胞胎患者，其中一个模型将跟踪患者的时间。我们专注于在临床试验期间使用详细的成像数据集制作患者心脏的数字双胞胎。这是迈向实时更新模型的第一步，这种模型可以在患者的整个生命过程中进行跟踪，并直接反馈给患者护理。这种数字双胞胎方法建立在特定于患者的心脏计算机模型的基础上，目前正在英国、伦敦国王学院和美国进行评估，以指导手术。这些模型旨在为特定患者的病理生理优化治疗，但只模拟少量的心跳。数字双胞胎通过时间跟踪患者，将预测疾病进展和对治疗的反应。这代表了模拟引导治疗的下一步，将预测最佳治疗以及重要的是何时提供治疗。该项目将解决在校准大量患者的计算机模型方面的技术挑战，如何随着更多数据的获得而随着时间的推移有效地更新这些模型，如何分析在临床试验期间记录的心脏图像，以及如何预测心脏形状和功能的复杂变化。这些方法将在三项研究中应用于三组患者。首先，我们将测试多尺度心脏生物力学模型是否可以确定心力衰竭患者泵功能障碍的常见原因。其次，我们将测试数字双胞胎是否可以预测哪些已经从心力衰竭中恢复的患者可以停止心力衰竭的调解。第三，我们将测试数字双胞胎预测是否可以用来预测新诊断的心力衰竭患者的康复情况，并预防对高级心力衰竭治疗的需要。这将提供心脏生物力学数字双胞胎的第一个演示，使用真实的临床数据来回答重要的临床问题。

Modelling and simulation play important roles in designing everything from planes to cars to bridges. However, advances in connectivity and computing now enable models to be linked directly to a specific object or system, creating a "digital twin". Digital twins represent a computational surrogate for a particular object and are updated through time as more information becomes available. However, digital twins are not limited to manufactured objects alone. This project aims to develop digital twins of patients, where a model will track a patient through time. We focus on making digital twins of patients' hearts using detailed imaging data sets over the period of a clinical trial. This is the first step towards models that are updated in real-time, track the patient throughout their life and directly feed back into informing patient care. The digital twin approach builds on patient-specific computer models of the heart that are currently being evaluated to guide procedures in the UK at King's College London and in the US. These models are designed to optimise treatments for a specific patient's pathophysiology but only simulate a small number of heartbeats. Digital twins, which track a patient through time, will forecast disease progression and response to therapy. This represents the next step in simulation guided therapy, where the optimal treatment and, importantly, when to deliver it, will be predicted. This project will address the technical challenges in calibrating computer models of large numbers of patients, how to efficiently update these models through time as more data becomes available, how to analyse images of the heart recorded over the duration of a clinical trial and how to predict complex changes in shape and function of the heart. The approaches will be applied to study three patient groups in three studies. First, we will test if multi-scale cardiac biomechanics models can identify common causes of pump dysfunction in heart failure patients. Second, we will test if digital twins can predict which patients who have recovered from heart failure can stop their heart failure mediation. Thirdly, we will test if digital twin forecasts can be used to predict recovery and pre-empt the need for advanced heart failure therapy in newly diagnosed heart failure patients. This will provide the first demonstration of cardiac biomechanics digital twins using real clinical data to answer important clinical questions.