The impact of ankle-foot orthosis on walking features of drop foot patients

Drop foot is a syndrome characterised by the difficulty in lifting the front part of the foot, caused by the weakness or disability of muscles responsible for foot dorsiflexion. Ankle Foot Orthosis (AFO) is an orthopaedic support used in drop foot patients to support and maintain the foot and ankle in the proper position during movement, improving the gait pattern. In this study we investigated the effect and clinical relevance of the use of an AFO on walking ability in patients suffering drop foot syndrome. Two Machine Learning (ML) algorithms were implemented on Knime Analytic Platform: Decision Tree (DT) and Random Forest (RF). They were used on a set of spatio-temporal gait variables assessed by means of the IMU-based wearable system for gait analysis, namely Opal Mobility Lab by APDM, on 19 patients with unilateral drop foot syndrome, in two separate conditions: wearing or not the AFO. The following evaluation metrics were used to evaluate the performances of the two tree-based ML models: Recall, Precision, Sensitivity, Specificity, F-measure, Accuracy and Area under the Receiver Operating Characteristic Curve (AucRoc). Moreover a Feature Importance analysis was carried out. Results demonstrate RF was perfect in distinguishing the two classes - walking with and without AFO - while DT achieved a slightly lower classification score but still of high discriminative power, confirming the effective incidence of the use of the AFO on walking pattern. The following parameters were identified as the most informative by the Feature Importance analysis: stance and swing phase duration, cadence, step duration and stride length.