Detection and Classification of Stroke Gaits by Deep Neural Networks Employing Inertial Measurement Units

Abstract

This paper develops Deep Neural Network (DNN) models that can recognize stroke gaits. Stroke patients usually suffer from partial disability and develop abnormal gaits that can vary widely and need targeted treatments. Evaluation of gait patterns is crucial for clinical experts to make decisions about the medication and rehabilitation strategies for the stroke patients. However, the evaluation is often subjective, and different clinicians might have different diagnoses of stroke gait patterns. In addition, some patients may present with mixed neurological gaits. Therefore, we apply artificial intelligence techniques to detect stroke gaits and to classify abnormal gait patterns. First, we collect clinical gait data from eight stroke patients and seven healthy subjects. We then apply these data to develop DNN models that can detect stroke gaits. Finally, we classify four common gait abnormalities seen in stroke patients. The developed models achieve an average accuracy of 99.35% in detecting the stroke gaits and an average accuracy of 97.31% in classifying the gait abnormality. Based on the results, the developed DNN models could help therapists or physicians to diagnose different abnormal gaits and to apply suitable rehabilitation strategies for stroke patients.

Wearable sensors and machine learning in post-stroke rehabilitation assessment: A systematic review

“…The system objective is to classify well and poorly executed tasks [52][53][54][55][56][57][58][59][60][61]63,64], to do so many approaches were followed. Some researchers implemented systems to distinguish between normal and abnormal gaits for lower-limb rehabilitation [56,59,60], in which participants executed 10 m walks. Other researchers assessed the execution of ADLs [54,57] like different kitchen related activities or routine bedroom tasks.…”

“…Over the past few years, effort has been put into developing unobtrusive, effective and objective motion-modeling systems, taking advantage of the progress made in the sensor technology which became more compact and more power-efficient [83]. All the included works utilised IMUs for the data acquisition [42][43][44][45][46][47][48][49][50][52][53][54][55][56][57][58][59][60][61][65][66][67][68][69][70]63,[71][72][73][74]64]. IMUs are devices that combine linear acceleration from accelerometer and the angular turning rates from gyroscopes [84].…”

“…These placements were targeted to monitor activities that involved using hands. By contrast, systems that focused on lower limbs for activities that involved walking utilised more frequently the chest in eight studies [48,54,58,61,65,67,69,73] the shank in four works [56,59,60,70], thighs [55,59,63], and feet [55,59,74] in three studies, while two works targeted either the hip [44,45], or the waist [47,50], and finally a single study used the lower back [59]. Fig.…”

“…For the study population, many works recruited stroke survivors with different degrees of severity after getting ethical approvals [43][44][45][46][47][48][49]52,54,[56][57][58][59][60][66][67][68][69][70]63,[71][72][73][74]64]. Some of them undertook a cohort study by combining them with Able Bodied (AB) participants [43][44][45][46]52,57,58,60,67,64] elderly [52] and neurologically disordered patients [59]. Other works only used AB [41,42,50], while some studies did not specify or did not use participants [61,53].…”

Wearable Devices for Gait Analysis in Intelligent Healthcare

“…Frequently recording of gaits of a wide range of people by wearable devices will open new opportunities for intelligent analysis. The wearable accelerometers and inertial measurement units can be attached to people in need outside healthcare institutes (Wang et al, 2021). These wearable devices provide long-term motion tracking during walking and allow for daily gait monitoring under different conditions.…”