Continuous non-invasive blood pressure estimation.

血压是五大生命体征之一，也是衡量患者健康状况的主要指标。高血压是心脏血管疾病(CVD)的最大危险因素之一，CVD是全球主要的死亡原因。目前测量血压的方法是使用血压计，即在患者手臂上放置充气袖带，充气直到血流闭塞。这提供了患者的收缩压(心脏收缩时的最高血压)和舒张压(心脏放松时的最低血压)的单一读数。然而，血压是一个全天变化的信号，因此单一读数不能为准确诊断心血管疾病或其他与血压相关的疾病提供足够的信息。此外，白大褂效应的存在表明，许多患者在检查血压时会被误诊。这证明了连续无创血压测量的必要性。目前在连续无创血压测量领域的研究主要集中在脉搏传递时间(PTT)，即血液从一个动脉部位到另一个动脉部位所需的时间。PTT的变化被证明与适当校准后血压的变化相关。为了记录两个不同位置的血流开始，大多数使用PTT来得出血压估计值的设备都是通过使用放置在患者身上的两个或更多设备来实现的。这方面的例子包括连接到同一系统的心电(ECG)和光体积描记(PPG)传感器，或两个PPG传感器。由于需要两个传感器，因此该设备不太可能一直使用，因为患者会发现他们不舒服，并会干扰他们的日常生活；因此，很难将该设备称为真正的连续性。DPhil项目旨在探讨一种新的实现连续血压估计的方法。这个项目目前正在考虑两种方法(然而，由于该项目还很年轻，很可能会有更多的方法有待探索)。第一种是使用放置在患者手腕上的单个PPG传感器来获得类似PTT的信号，或者使用机器学习技术根据PPG来估计血压。第二种方法将研究基于相机的血压估算方法。已经证明，患者的摄像机流能够得出患者的病理信号，这项工作可以扩展到考虑患者身上许多位置的血液流动，从而可以使用PTT分析从这些位置之间的时间差得出血压估计。该项目的理想应用是在夜间连续监测血压。最近的研究表明，50岁以上的住院患者夜间血压会上升，而年轻人的血压通常会下降。这项研究还发现了显著的性别差异，以及一周中不同日期的差异。这些结果表明，跟踪血压随时间的变化不仅可以帮助评估患有高血压等临床疾病的患者，还可以提供有关心血管健康的新信息。使用血压计重复测量夜间血压变化是困难的，因为有可能唤醒患者，从而导致血压上升。因此，持续的无创血压评估(或遵循相同趋势的衍生血压替代品)将使血压不下降的患者很容易被识别出来，以便进一步治疗。

Blood pressure is one of the five vital signs and a major indicator of a patient's health. High blood pressure (hypertension) is one of the strongest risk factors for Cardio Vascular Disease (CVD), which is the leading cause of death worldwide. Current methods for measuring blood pressure use a sphygmomanometer, an inflatable cuff placed around the patients arm and inflated until the blood flow is occluded. This provides a single reading of a patient's systolic (highest blood pressure when the heart contracts) and diastolic (lowest blood pressure when the heart relaxes) blood pressure.However, blood pressure is a signal that changes throughout the day and so a single reading cannot provide sufficient information for accurate diagnosis of CVD or other blood pressure related diseases. In addition, the presence of a "white coat" effect, in which a patient's blood pressure increases due to the clinician taking the measurement, has shown that many patients will have been misdiagnosed when having their blood pressure examined. This has proven the need for a continuous non-invasive blood pressure estimation.Current research efforts in the field of continuous non-invasive blood pressure estimation are mostly focussed on Pulse Transit Time (PTT) which is defined as the time it takes for blood to travel from one arterial site to another. Changes in PTT are shown to correlate with changes in blood pressure after appropriate calibration. In order to record the onset of the blood flow at the two different locations most devices that use PTT to derive an estimate of blood pressure do so by using two or more devices placed on a patient. Examples of these include an electrogardiogram (ECG) and a photoplethysmography (PPG) sensor linked to the same system, or two PPG sensors. Due to the need for two sensors it is very unlikely that the device will be used at all times as patients will find them uncomfortable and will get in the way of their daily life; and therefore, it is difficult to call the devices truly continuous.This DPhil project aims to approach a novel implementation of continuous blood pressure estimation. There are two methods that are currently being considered for this project (however, as the project is still young it is very likely that there will be more methods to be explored). The first is to use a single PPG sensor placed on the wrist of a patient to derive either a PTT-like signal or use machine learning techniques to estimate blood pressure from the PPG. The second method will study camera-based methods of deriving a blood pressure estimation. It has already been shown that a video camera stream of a patient is able to derive pathological signals of the patient, this work can be extended to consider the flow of blood in many locations on the patient and thus PTT analysis can be used to derive a blood pressure estimation from the time differences between these locations.An ideal application for this project is in continuous monitoring of blood pressure during the night time. Recent research has shown a nocturnal rise in systolic BP for in-hospital patients above the age of 50 as opposed to the usual dipping patterns found in younger individuals. The research has also identified significant gender-specific differences, as well as variations according to the day of the week. These results suggest that tracking BP changes over time could not only help the assessment of patients with clinical conditions such as hypertension, but also provide new information about cardiovascular health. It is difficult to use a sphygmomanometer to repeatedly measure blood pressure changes during the night for risk of waking the patient which would result in a rise in blood pressure. Therefore, continuous non-invasive blood pressure estimation (or a derived surrogate of blood pressure that follows the same trends) will allow for a patient whose blood pressure does not dip to be easily identified for further treatment.