A new classification of hemiplegia gait patterns based on bicluster analysis of joint moments.

PURPOSE Hemiplegia is a paralysis on one side of the body resulting from disease or injury to the motor centers of the brain that may lead to difficulty in walking and problems in balance. A new methodology for hemiplegia gait patterns classification based on bicluster analysis, which aims to identify a group of patients with similar gait patterns, and verify if spatial-temporal gait parameters are correlated with the Barthel Index, has been proposed. METHODS Eighteen hemiplegia patients were recruited. Measurements included spatialtemporal gait parameters and joint moments. Gait data were measured using a motion tracking system and two force platforms. Bicluster analysis was used to classify the subjects' gait patterns. The relation between Barthel Index and spatial-temporal gait parameters was determined based on the Spearman correlation. RESULTS A high correlation between spatial-temporal gait parameters and Barthel Index (r>0.5, p <0.05) was observed. Well-separated biclusters presenting similarity among the lower limb joints during the gait cycles were obtained from the data. CONCLUSIONS Bicluster analysis can be useful for identifying patients with similar gait patterns. The relation between the gait patterns and the underlying impairments would allow clinicians to target rehabilitation strategies at the patient's individual needs.