Hybrid learning control for improving suppression of hand tremor

Real-Time Voluntary Motion Prediction and Parkinson's Tremor Reduction Using Deep Neural Networks

Wearable tremor suppression devices (WTSD) have been considered as a viable solution to manage parkinsonian tremor. WTSDs showed their ability to improve the quality of life of individuals suffering from parkinsonian tremor, by helping them to perform activities of daily living (ADL). Since parkinsonian tremor has been shown to be nonstationary, nonlinear, and stochastic in nature, the performance of the tremor models used by WTSDs is affected by their inability to adapt to the nonlinear behaviour of tremor. Another drawback that the models have is their limitation to estimate or predict one step ahead, which introduces delay when used in real time with WTSDs, which compromises performance. To address these issues, this work proposes a deep neural network model that learns the correlations and nonlinearities of tremor and voluntary motion, and is capable of multi-step prediction with minimal delay. A generalized model that is task and user-independent is presented. The model achieved an average estimation percentage accuracy of 99.2%. The average future voluntary motion prediction percentage accuracy with 10, 20, 50, and 100 steps ahead was 97.0%, 94.0%, 91.6%, and 89.9%, respectively, with prediction time as low as 1.5 ms for 100 steps ahead. The proposed model also achieved an average of 93.8% ± 1.5% in tremor reduction when it was tested in an experimental setup in real time. The tremor reduction showed an improvement of 25% over the Weighted Fourier Linear Combiner (WFLC), an estimator commonly used with WTSDs.

The Design of a Parkinson's Tremor Predictor and Estimator Using a Hybrid Convolutional-Multilayer Perceptron Neural Network

Parkinson's Disease (PD) is considered to be the second most common age-related neuroegenerative disorder, and it is estimated that seven to ten million people worldwide have PD. One of the symptoms of PD is tremor, and studies have shown that wearable assistive devices have the potential to assist in suppressing it. However, despite the progress in the development of these devices, their performance is limited by the tremor estimators they use. Thus, a need for a tremor model that helps the wearable assistive devices to increase tremor suppression without impeding voluntary motion remains. In this work, a user-independent and task-independent tremor and voluntary motion detection method based on neural networks is proposed. Inertial measurement units (IMUs) were used to measure acceleration and angular velocity from participants with PD, these data were then used to train the neural network. The achieved estimation percentage accuracy of voluntary motion was 99.0%, and the future prediction percentage accuracy was 97.3%, 93.7%, 91.4% and 90.3% for 10 ms, 20 ms, 50 ms and 100 ms ahead, respectively. The root mean squared error (RMSE) achieved for tremor estimation was an average of 0.00087^°/s on new unseen data, and the future prediction average RMSE across the different tasks achieved was 0.001^°/s, 0.002^°/s, 0.020^°/s and 0.049^°/s for 1 ms, 2 ms, 5 ms, and 10 ms ahead, respectively. Therefore, the proposed method shows promise for use in wearable suppression devices.

Telemonitoring Parkinson's disease using machine learning by combining tremor and voice analysis

BACKGROUND

With the growing number of the aged population, the number of Parkinson's disease (PD) affected people is also mounting. Unfortunately, due to insufficient resources and awareness in underdeveloped countries, proper and timely PD detection is highly challenged. Besides, all PD patients' symptoms are neither the same nor they all become pronounced at the same stage of the illness. Therefore, this work aims to combine more than one symptom (rest tremor and voice degradation) by collecting data remotely using smartphones and detect PD with the help of a cloud-based machine learning system for telemonitoring the PD patients in the developing countries.

METHOD

This proposed system receives rest tremor and vowel phonation data acquired by smartphones with built-in accelerometer and voice recorder sensors. The data are primarily collected from diagnosed PD patients and healthy people for building and optimizing machine learning models that exhibit higher performance. After that, data from newly suspected PD patients are collected, and the trained algorithms are evaluated to detect PD. Based on the majority-vote from those algorithms, PD-detected patients are connected with a nearby neurologist for consultation. Upon receiving patients' feedback after being diagnosed by the neurologist, the system may update the model by retraining using the latest data. Also, the system requests the detected patients periodically to upload new data to track their disease progress.

RESULT

The highest accuracy in PD detection using offline data was [Formula: see text] from voice data and [Formula: see text] from tremor data when used separately. In both cases, k-nearest neighbors (kNN) gave the highest accuracy over support vector machine (SVM) and naive Bayes (NB). The application of maximum relevance minimum redundancy (MRMR) feature selection method showed that by selecting different feature sets based on the patient's gender, we could improve the detection accuracy. This study's novelty is the application of ensemble averaging on the combined decisions generated from the analysis of voice and tremor data. The average accuracy of PD detection becomes [Formula: see text] when ensemble averaging was performed on majority-vote from kNN, SVM, and NB.

CONCLUSION

The proposed system can detect PD using a cloud-based system for computation, data preserving, and regular monitoring of voice and tremor samples captured by smartphones. Thus, this system can be a solution for healthcare authorities to ensure the older population's accessibility to a better medical diagnosis system in the developing countries, especially in the pandemic situation like COVID-19, when in-person monitoring is minimal.

Characterization of Parkinsonian Hand Tremor and Validation of a High-Order Tremor Estimator

Recent progress in wearable technology has made wearable tremor suppression devices (WTSDs) for Parkinson's patients a potentially viable alternative solution for tremor management. So far, in contrast to wrist and elbow tremor, finger tremors have not been studied in depth despite the huge impact that they have on a patient's daily life. In addition, more evidence has been found showing that the performance of current tremor estimators may be limited by their model order due to the multiple harmonics present in tremor. The aim of this paper is to characterize finger and wrist tremor in both the time and frequency domains, and to propose a high-order tremor estimation algorithm. Tremor magnitudes are reported in the forms of linear acceleration, angular velocity, and angular displacement. The activation of forearm flexor and extensor muscles is also investigated. The frequency analysis shows that Parkinsonian tremors produce oscillations of the hand with pronounced harmonics. At last, a high-order weighted-frequency Fourier linear combiner (WFLC)-based Kalman filter is proposed. The percentage estimation accuracy achieved from the proposed estimator is 96.3 ± 1.7%, showing average improvements of 28.5% and 48.9% over its lower-order counterpart and the WFLC. The proposed estimator shows promise for use in a WTSD.

A hybrid active force control of a lower limb exoskeleton for gait rehabilitation

Owing to the increasing demand for rehabilitation services, robotics have been engaged in addressing the drawbacks of conventional rehabilitation therapy. This paper focuses on the modelling and control of a three-link lower limb exoskeleton for gait rehabilitation that is restricted to the sagittal plane. The exoskeleton that is modelled together with a human lower limb model is subjected to a number of excitations at its joints while performing a joint space trajectory tracking, to investigate the effectiveness of the proposed controller in compensating disturbances. A particle swarm optimised active force control strategy is proposed to facilitate disturbance rejection of a conventional proportional-derivative (PD) control algorithm. The simulation study provides considerable insight into the robustness of the proposed method in attenuating the disturbance effect as compared to the conventional PD counterpart without compromising its tracking performance. The findings from the study further suggest its potential employment on a lower limb exoskeleton.