Allan Pang (Associate)

Research Topic: Inpatient Physiological Clinical Deterioration Models

Supervisors: Dr Alwyn Kotzé, Prof Geoff Hall, Mr Owen Johnson, Dr Marc de Kamps

About Allan: Having graduated from Leeds University Medical School (MBChB) in 2010 and gained my Fellowship of the Royal College of Anaesthesia (FRCA) in 2019, I am a Military Senior Anaesthetic Registrar on the Northern School of Anaesthesia and Intensive Care Medicine specialist training rotation. My sub-speciality interests include digital health and major trauma management. My proposed research programme will examine the use of machine learning techniques in producing a predictive adaptive model in perioperative outcomes and explore its utilisation by front-line healthcare workers.

Project Description: Predicting impending mortality in hospitals uses physiological data to produce a rules-based aggregated early warning score (EWS). Such systems widely used in clinical settings are NEWS2. These predictive frameworks only consider deterioration using vital signs at a single time point and do not consider previous clinical states or clinical trajectories.

Using a combination of time series machine learning techniques to encode trajectory information and data from inpatient electronic health records from Leeds NHS Hospitals Teaching Trust, I have shown the benefit of trajectory information for inpatient mortality prediction. Further work will consider the effects of ongoing treatments on physiological variables and how integration into future clinical decision support tools.