In this paper, a real-time knee extension monitoring and rehabilitation system for people, such as patients, the elderly, athletes, etc., is developed and tested. The proposed system has three major functions. The first function is two-channel surface electromyography (EMG) signal measurement and processing for the vastus lateralis (VL) and vastus medialis (VM) muscles using a developed EMG device set. The second function is the knee extension range of motion (ROM) measurement using an angle sensor device set (i.e., accelerometer sensor). Both functions are connected and parallelly processed by the NI-myRIO embedded device. Finally, the third function is the graphical user interface (GUI) using LabVIEW, where the knee rehabilitation program can be defined and flexibly set, as recommended by physical therapists and physicians. Experimental results obtained from six healthy subjects demonstrated that the proposed system can efficiently work with real-time response. It can support multiple rehabilitation users with data collection, where EMG signals with mean absolute value (MAV) and root mean square value (RMS) results and knee extension ROM data can be automatically measured and recorded based on the defined rehabilitation program. Furthermore, the proposed system is also employed in the hospital for validation and evaluation, where bio-feedback EMG and ROM data from six patients, including (a) knee osteoarthritis, (b) herniated disc, (c) knee ligament injury, (d) ischemic stroke, (e) hemorrhagic stroke, and (f) Parkinson are obtained. Such data are also collected for one month for tracking, evaluation, and treatment. With our proposed system, results indicate that the rehabilitation people can practice themselves and know their rehabilitation progress during the time of testing. The system can also evaluate (as a primary treatment) whether the therapy training is successful or not, while experts can simultaneously review the progress and set the optimal treatment program in response to the rehabilitation users. This technology can also be integrated as a part of the Internet of Things (IoT) and smart healthcare systems.

Deep learning algorithms are employed in many applications, especially in medical fields such as gait analysis and human pose detection for rehabilitation. However, creating the desired model with deep learning algorithms requires high memory and computing costs, which is problematic because deep learning technologies must be run on low-power devices such as edge computing equipment. To deal with these problems, feature reduction methods reduce the memory and energy costs. This paper presents an empirical analysis of deep learning with feature reduction. The method classifies foot images for knee rehabilitation using convolutional and dense autoencoders. The obtained results are compared with those of conventional methods (histograms of oriented gradients and local binary pattern algorithms). The features were classified and compared using support vector machine, k-nearest neighbor, and multilayer perceptron methods. The experimental results demonstrate that the conventional method uses fewer features than the deep learning method with higher accuracy because its algorithm projects pixels onto the histogram. In addition, using fewer features in deep learning layers maintains high accuracy, which is beneficial for edge computing implementations.

Background The goal of this study was to assess the reliability of electromyography and range of motion measurements obtained using a knee exercise monitoring system. This device was developed to collect data on knee exercise activities. Methods Twenty healthy individuals performed isotonic quadriceps exercises in this study. The vastus medialis surface electromyography (sEMG) and range of motion (ROM) of the knee were recorded during the exercise using the isotonic knee exercise monitoring device, the Mobi6-6b, and a video camera system. Each subject underwent a second measuring session at least 24 h after the first session. To determine reliability, the intraclass correlation coefficients (ICCs) and standard error of measurement (SEM) at the 95% confidence interval were calculated, and a Bland–Altman analysis was performed. Results For inter-rater reliability, the ICCs of the mean absolute value (MAV) and root mean square (RMS) of sEMG were 0.73 (0.49, 0.86) and 0.79 (0.61, 0.89), respectively. ROM had an ICC of 0.93 (0.02, 0.98). The intra-rater reliability of the MAV of the sEMG was 0.89 (0.71, 0.96) and the intra-rater reliability of RMS of the sEMG was 0.88 (0.70, 0.95). The ROM between days had an intra-rater reliability of 0.82 (0.54, 0.93). The Bland–Altman analysis demonstrated no systematic bias in the MAV and RMS of sEMG, but revealed a small, systematic bias in ROM (−0.8311 degrees). Conclusion For sEMG and range of motion measures, the isotonic knee exercise monitoring equipment revealed moderate to excellent inter- and intra-rater agreement. However, the confidence interval of ROM inter-rater reliability was quite large, indicating a small agreement bias; hence, the isotonic knee exercise monitor may not be suitable for measuring ROM. This isotonic knee exercise monitor could detect and collect information on a patient’s exercise activity for the benefit of healthcare providers.

Lower-body detection can be useful in many applications, such as the detection of falling and injuries during exercises. However, it can be challenging to detect the lower-body, especially under various lighting and occlusion conditions. This paper presents a novel lower-body detection framework using proposed anthropometric ratios and compares the performance of deep learning (convolutional neural networks and OpenPose) and traditional detection methods. According to the results, the proposed framework helps to successfully detect the accurate boundaries of the lower-body under various illumination and occlusion conditions for lower-limb monitoring. The proposed framework of anthropometric ratios combined with convolutional neural networks (A-CNNs) also achieves high accuracy ($90.14\%$), while the combination of anthropometric ratios and traditional techniques (A-Traditional) for lower-body detection shows satisfactory performance with an averaged accuracy ($74.81\%$). Although the accuracy of OpenPose ($95.82\%$) is higher than the A-CNNs for lower-body detection, the A-CNNs provides lower complexity than the OpenPose, which is advantageous for lower-body detection and implementation on monitoring systems.

Accurate differentiation of mononuclear cells is essential for diagnosing hematological conditions. Nonetheless, their similar morphology can lead to misclassification. The application of artificial intelligence enables the extraction of discriminative features from cell images through advanced mathematical modeling. While a compact model is necessary to enable dynamic-monitoring in clinical practice, achieving high accuracy typically requires large, well-balanced datasets, presenting significant challenges in balancing model complexity and classification performance. This study compares traditional features like Local Binary Patterns (LBP), Histogram of Oriented Gradients (HOG), and Gray-Level Co-occurrence Matrix (GLCM) with transfer learning using fine-tuned Vision transformer (ViT) for capturing features of images. Using a dataset of 4,315 images focusing on mononuclear cells, we integrated class balancing techniques and tested models with ten classifiers. Results showed HOG with XGBoost achieved 0.77 accuracy, while fine-tuned ViT with augmentation and XGBoost reached 0.82. The computational intensity of ViT is offset by its real-time monitoring potential.