Pattern recognition of abnormal physiology in the emergency department

- Carrina Lee

Pattern recognition of abnormal physiology in the emergency department

Patient physiology is an important indicator for patient assessment, diagnosis and treatment. Despite the wealth of information they embody, physiological values are inefficiently presented to clinical staff. Displayed as numerical values on bedside monitors, interpretation requires time to process, is cognitively challenging and prone to erroneous interpretation. There is therefore a need for a novel device that utilises skills known to be inherent in human information processing; using simple pathognomonic shapes to make physiological information more obvious. This could thereby reduce the amount of information potentially being missed.

Our investigations have identified a ‘south-west pointing diamond’ to be synonymous with sepsis physiology. Analysis of shape configuration over time revealed responses to pain, analgesia and fluid resuscitation. The application of various quantitative analysis techniques also confirmed the validity and applicability of radar chart shapes. Evaluating our concept of shape-based pattern recognition also showed promising results; after a brief introduction, clinicians were more readily able to identify abnormal physiology.

As a poorly researched area, especially in the context of undifferentiated patients with time critical diseases requiring emergency medical treatment, our study sheds light on early pathological traits of common emergency department presenting complaints. Further investigations are required to refine shape development and expand our collection of derived pathognomonic shapes. Collaboration with biomedical engineers could steer the development of a machine learning approach that could potentially be used at the patient bedside.