Biomedical Signal Analysis and System Identification

本研究旨在开发用于生物医学信号和系统的分析和识别的算法和工具。所解决的问题的动机和生物研究的要求定义，但研究的重点是开发和验证实用的工具和算法，而不是调查具体的生物问题。两个生物医学领域的问题将得到解决。(1)研究姿态和运动控制的识别方法。其目的是提供所需的工具来表征正常日常功能过程中关节的动态力学特性，并了解其潜在机制。这对于理解正常和病理性运动功能是必不可少的，并且对于运动异常的诊断、治疗的定量评估以及诸如假肢和矫形器之类的先进辅助装置的设计具有重要的实际结果。(2)小儿心肺功能监测的自动信号分析。众所周知，呼吸和心脏行为包含可以帮助管理呼吸问题的信息，例如早产儿呼吸暂停，并且可以安全地终止婴儿从机械通气中拔管。然而，由于需要进行人工数据分析，这一信息的应用受到了阻碍。这是昂贵的，耗时的和不准确的。我们将通过开发一个全自动的心肺信号分析系统来解决这个问题，并将这些系统与机器学习技术联合收割机相结合，以预测婴儿将遇到不良呼吸事件的概率。

This research seeks to develop algorithms and tools for the analysis and identification of biomedical signals and systems. The problems addressed are motivated and defined by the requirements of biological research but the focus of the research is on the development and validation of practical tools and algorithms, rather than investigating specific biological questions. Problems from two biomedical areas will be addressed.(1) Identification methods to study the control of posture and movement. The objective is to provide the tools needed to characterize the dynamic, mechanical properties of joints during normal daily function and understand the underlying mechanisms. This is essential to understand both normal and pathological motor function and will have important practical results for the diagnosis of motor abnormalities, the quantitative assessment of treatment, and the design of advanced assistive devices such as prostheses and orthoses.(2) Automated signal analysis for pediatric cardio-respiratory function monitoring. It is known that respiratory and cardiac behavior contains information that could aid in the management of respiratory problems, such as apnea of prematurity and the extubation of infants from mechanical ventilation can safely terminated. However, application of this information is hampered by the need for manual data analysis. This is expensive, time consuming and inaccurate. We will address this issue by developing a fully automated system for the analysis of cardio-respiratory signals and combine these with machine learning techniques to predict the probability that an infant will encounter an adverse respiratory event.