We record or obtain data of behaviour in real-world environments, to fully capture the variance and richness of natural human actions.
We take a data-driven approach to analyse the high-dimensional data we record, often across complex action spaces.
We build data-driven models from these datasets, to gain insight into human behaviour in both health and disease.
At present, it takes a long time to establish if new therapy for diseases are working, as clinicians currently gauge disease progression ‘by eye’ instead of using measurable and objective methods in a real-world setting. In this arm of our research, we monitor patients’ behaviour on a 24/7 basis using smart-sensors on the 4 extremities to obtain a continuous understanding of their motor capabilities in real-life. The behavioural data, motor or otherwise, is then analysed using machine learning techniques to derive new digital biomarkers, which can capture individual variations in disease progression. This novel approach will significantly reduce the time taken to detect disease progression, potentially reducing the duration of future clinical trials.