Machine learning and data informatics approaches for Personalised Outcome Prediction in Paediatric Intensive Care

这个博士项目将开发机器学习和信息学算法，用于儿科重症监护环境中临床数据的数据驱动链接。我们假设，患有危及生命的脑创伤的儿科患者的临床、生理和放射(结构)数据将提供大脑自动调节能力受损的信息，并且结合和挖掘这些多模式数据将能够检测到以前未确定的患者的临床结果较差的风险。常规临床实践产生了大量未被充分利用的数据，用于研究和质量改进。在儿科重症监护病房(PICU)尤其如此。然而，一旦患者出院，来自这些生理大数据的重要信息就会被丢弃，而不是被用来提高我们对患者的生理表型可能如何影响结果的理解。缺乏与常规临床护理期间收集的其他数据源的联系(例如，放射图像、结果，如再次入院)，阻碍了这些生理学数据的有效使用，以提高患者的护理和安全性。我们迫切需要利用数据科学来整合常规患者护理过程中从不同来源产生的数据，并开发用于危重护理的精确医学方法，以提供不断改善的患者护理和结果。

This PhD project will develop machine learning and informatics algorithms for data-driven linkage of clinical data in paediatric critical care settings. We hypothesise that clinical, physiological and radiological (structural) data in paediatric patients with life-threatening brain trauma will inform about damage to the brain's ability to auto-regulate, and that combining and mining these multimodal data will enable the detection of patients - previously unidentified - at a higher risk of poor clinical outcomes.Routine clinical practice generates a large amount of data that is under-used for research and quality improvement. This is particularly true in paediatric intensive care units (PICU). Yet once the patient is discharged, vital information from this physiological big data is discarded rather than being used to advance our understanding of how a patient's physiological phenotype may affect outcome. Lack of linkage to other data sources collected during routine clinical care (e.g., radiological images, outcome such as re-admission) prevents meaningful use of this physiology data to advance patient care and safety. We urgently need to utilise data science to integrate the data generated from different sources during routine patient care and develop precision medicine approaches for critical care to deliver continuously improved patient care and outcome.