To develop a method for cooperative human gait training, we investigated whether interactive rhythmic cues could improve the gait performance of Parkinson's disease patients. The interactive rhythmic cues ware generated based on the mutual entrainment between the patient's gait rhythms and the cue rhythms input to the patient while the patient walked. Previously, we found that the dynamic characteristics of stride interval fluctuation in Parkinson's disease patients were improved to a healthy 1/f fluctuation level using interactive rhythmic cues and that this effect was maintained in the short term. However, two problems remained in our previous study. First, it was not clear whether the key factor underpinning the effect was the mutual entrainment between the gait rhythms and the cue rhythms or the rhythmic cue fluctuation itself. Second, it was not clear whether or not the gait restoration was maintained longitudinally and was relearned after repeating the cue-based gait training. Thus, the present study clarified these issues using 32 patients who participated in a four-day experimental program. The patients were assigned randomly to one of four experimental groups with the following rhythmic cues: (a) interactive rhythmic cue, (b) fixed tempo cue, (c) 1/f fluctuating tempo cue, and (d) no cue. It has been reported that the 1/f fluctuation of stride interval in healthy gait is absent in Parkinson's disease patients. Therefore, we used this dynamic characteristic as an evaluation index to analyze gait relearning in the four different conditions. We observed a significant effect in condition (a) that the gait fluctuation of the patients gradually returned to a healthy 1/f fluctuation level, whereas this did not occur in the other conditions. This result suggests that the mutual entrainment can facilitate gait relearning effectively. It is expected that interactive rhythmic cues will be widely applicable in the fields of rehabilitation and assistive technology.

Walking is generated by the interaction between neural rhythmic and physical activities. In fact, Parkinson’s disease (PD), which is an example of disease, causes not only neural rhythm generation disorders but also physical disabilities. However, the relationship between neural rhythm generation disorders and physical disabilities has not been determined. The aim of this study was to identify the mechanism of gait rhythm generation. In former research, neural rhythm generation disorders in PD patients’ walking were characterized by stride intervals, which are more variable and fluctuate randomly. The variability and fluctuation property were quantified using the coefficient of variation (CV) and scaling exponent α. Conversely, because walking is a dynamic process, postural reflex disorder (PRD) is considered the best way to estimate physical disabilities in walking. Therefore, we classified the severity of PRD using CV and α. Specifically, PD patients and healthy elderly were classified into three groups: no-PRD, mild-PRD, and obvious-PRD. We compared the contributions of CV and α to the accuracy of this classification. In this study, 45 PD patients and 17 healthy elderly people walked 200 m. The severity of PRD was determined using the modified Hoehn–Yahr scale (mH-Y). People with mH-Y scores of 2.5 and 3 had mild-PRD and obvious-PRD, respectively. As a result, CV differentiated no-PRD from PRD, indicating the correlation between CV and PRD. Considering that PRD is independent of neural rhythm generation, this result suggests the existence of feedback process from physical activities to neural rhythmic activities. Moreover, α differentiated obvious-PRD from mild-PRD. Considering α reflects the intensity of interaction between factors, this result suggests the change of the interaction. Therefore, the interaction between neural rhythmic and physical activities is thought to plays an important role for gait rhythm generation. These characteristics have potential to evaluate the symptoms of PD.

Scoring sleep stages from biological signals is an essential but labor-intensive inspection for sleep diagnosis. The existing automated scoring methods have achieved high accuracy but are not widely applied in clinical practice. In our understanding, the existing methods have failed to establish the trust of sleep experts (e.g., physicians and clinical technologists) due to a lack of ability to explain the evidences/clues for scoring. In this study, we developed a deep-learning-based scoring model with a reasoning mechanism called class activation mapping (CAM) to solve this problem. This mechanism explicitly shows which portions of the signals support our model’s sleep stage decision, and we verified that these portions overlap with the “characteristic waves,” which are evidences/clues used in the manual scoring process. In exchange for the acquisition of explainability, employing CAM makes it difficult to follow some scoring rules. Although we concerned the negative effect of CAM on the scoring accuracy, we have found that the impact is limited. The evaluation experiment shows that the proposed model achieved a scoring accuracy of $$86.9\%$$ 86.9 % . It is superior to those of some existing methods and the inter-rater reliability among the sleep experts. These results suggest that Sleep-CAM achieved both explainability and required scoring accuracy for practical usage.

No abstract

No abstract