Machine learning Machine Learning for the Improvement of Weaning from Mechanical Ventilation on the Intensive Care Unit

Supervisors:  Dr Raul Santos-Rodriguez, Department of Engineering Mathematics Dr Chris Bordeaux, NHS, Centre for Medical Education Dr Chris McWilliams, Department for Engineering Mathematics Email: m.wac@bristol.ac.uk

Project Summary:

Mechanical ventilation constitutes a large part of critical care and focuses on sustaining ventilatory functions in patients with acute respiratory failure. Despite being a life-saving process, prolonged mechanical ventilation may have serious long-term implications, suggesting that the weaning process should occur as expeditiously as possible. Removing the ventilator too soon, however, could lead to reintubation and result in a prolonged stay on the intensive care unit. The data-intensive nature of the intensive care unit environment makes the machine learning approach a viable solution to clinical problems within the critical care, and has a great potential to improve the clinical outcomes by supporting the clinicians working in this environment. This project focuses on implementing a decision-support tool that would integrate with the existing clinical information systems on the intensive care units in order to provide novel, real-time insights to clinicians. As such, it aims to utilise the existing patient data to create personalised predictions of weaning-readiness and suggest optimal treatment paths that would result in improved patient outcomes.

General Profile:

I am a passionate scientist and a software engineer by trade. In 2019 I graduated top of the class and with first class honours in Software Engineering (BSc) from Swansea University. Among the subjects I studied, I have grown a particular interest in high performance computing (HPC) and written my dissertation in computer graphics. My work revolved around the implementation of novel and efficient techniques for image quality improvement in photorealistic rendering using photon mapping and photon map relaxation.

In addition to my academic endeavours, I have already gained over five years of professional experience in software engineering, ranging from web development to low-level programming to system architecture and DevOps. During that time, I have had a chance to work with several med-tech start-ups and obtain valuable insights and a wide range of both technical and soft skills. The biggest project I have worked on was the development of software and its underlying architecture for CardioCube – a solution for cardiac patients. The product provided users with a voice interface and was integrated into third party devices such as Amazon Alexa as well as hospital electronic health records (EHR). The software automated the initial assessment at the clinic and enabled doctors to use their time more efficiently by focusing the patient, not the paperwork. CardioCube was tested in the Cedars Sinai hospital in Los Angeles, US and has since then been subjected to a clinical study, which found it to be highly accurate (97.51% agreement between verbal data and corresponding information).

As a person well versed in software engineering my interests definitely lie on the practical side of digital solutions. I am however open to any research topics and areas within healthcare.