Research Theme: e-Health




Early diagnosis of Parkinson's
Monitoring of people with prolonged disorder of consciousness

Diagnosis of people with mild Parkinson’s symptoms is difficult. Nevertheless, variations in gait pattern can be utilised to this purpose, when measured via Inertial Measurement Units (IMUs). Human gait, however, possesses a high degree of variability across individuals, and is subject to numerous nuisance factors. Therefore, off-the-shelf Machine Learning techniques may fail to classify it with the accuracy required in clinical trials.
In a recent paper we proposed a novel framework in which IMU gait measurement sequences sampled during a 10 metre walk are first encoded as hidden Markov models (HMMs) to extract their dynamics and provide a fixed-length representation. Given sufficient training samples, the distance between HMMs which optimises classification performance is learned and employed in a classical Nearest Neighbour classifier. Our tests demonstrate how this technique achieves accuracy of 85.51% over a 156 people with Parkinson’s with a representative range of severity and 424 typically developed adults, which is the top performance achieved so far over a cohort of such size, based on single measurement outcomes. The method displays the potential for further improvement and a wider application to distinguish other conditions.

People with extensive brain damage sometimes lose consciousness. Someone who has a disorder of consciousness lasting over two weeks is considered to have a ‘prolonged disorder of consciousness. To assess the degree to which these people are conscious, clinical scales such as the Coma Recovery Scale which assess patient behaviour and reaction to stimuli have been devised. These, however, only provide snapshots in time, do not allow for easily monitoring change, require trained observers (not available in most hospitals), and are subject to bias and subjective interpretation. A change in the law is looming which will enforce decisions to be made by physicians rather than by courts, so that the NHS urgently needs to put a proper process in place. Existing techniques, which directly or indirectly record brain activity, lack any substantive evidence that changes in activity in correspondence of an assigned task actually indicate that the person is aware, and cannot be practically applied to each and every patient.
In response, we propose a continuous monitoring system based on automatically analysing both video and auditory information via deep neural networks to identify patient activities judged by a consensus of experts to be of interest, summarising patterns of behaviour of patients as statistics of activities in time, and correlating them with clinical scales to provide solid evidence to clinical teams, upon which better and earlier decision can be made, circumventing issues with dispute or doubt.