An automatic wearable multi-sensor based gait analysis system for older adults.

Gait abnormalities in older adults are very common in clinical practice. They lead to serious adverse consequences such as falls and injury resulting in increased care cost. There is therefore a national imperative to address this challenge. Currently gait assessment is done using standardized clinical tools dependent on subjective evaluation. More objective gold standard methods (motion capture systems such as Qualisys and Vicon) to analyse gait rely on access to expensive complex equipment based in gait laboratories. These are not widely available for several reasons including a scarcity of equipment, need for technical staff, need for patients to attend in person, complicated time consuming procedures and overall expense. To broaden the use of accurate quantitative gait monitoring and assessment, the major goal of this thesis is to develop an affordable automatic gait analysis system that will provide comprehensive gait information and allow use in clinic or at home. It will also be able to quantify and visualize gait parameters, identify gait variables and changes, monitor abnormal gait patterns of older people in order to reduce the potential for falling and support falls risk management. A research program based on conducting experiments on volunteers is developed in collaboration with other researchers in Bournemouth University, The Royal Bournemouth Hospital and care homes. This thesis consists of five different studies toward addressing our major goal. Firstly, a study on the effects on sensor output from an Inertial Measurement Unit (IMU) attached to different anatomical foot locations. Placing an IMU over the bony prominence of the first cuboid bone is the best place as it delivers the most accurate data. Secondly, an automatic gait feature extraction method for analysing spatiotemporal gait features which shows that it is possible to extract gait features automatically outside of a gait laboratory. Thirdly, user friendly and easy to interpret visualization approaches are proposed to demonstrate real time spatiotemporal gait information. Four proposed approaches have the potential of helping professionals detect and interpret gait asymmetry. Fourthly, a validation study of spatiotemporal IMU extracted features compared with gold standard Motion Capture System and Treadmill measurements in young and older adults is conducted. The results obtained from three experimental conditions demonstrate that our IMU gait extracted features are highly valid for spatiotemporal gait variables in young and older adults. In the last study, an evaluation system using Procrustes and Euclidean distance matrix analysis is proposed to provide a comprehensive interpretation of shape and form differences between individual gaits. The results show that older gaits are distinguishable from young gaits. A pictorial and numerical system is proposed which indicates whether the assessed gait is normal or abnormal depending on their total feature values. This offers several advantages: 1) it is user friendly and is easy to set up and implement; 2) it does not require complex equipment with segmentation of body parts; 3) it is relatively inexpensive and therefore increases its affordability decreasing health inequality; and 4) its versatility increases its usability at home supporting inclusivity of patients who are home bound. A digital transformation strategy framework is proposed where stakeholders such as patients, health care professionals and industry partners can collaborate through development of new technologies, value creation, structural change, affordability and sustainability to improve the diagnosis and treatment of gait abnormalities.