Towards Multimodal Equipment to Help in the Diagnosis of COVID-19 Using Machine Learning Algorithms

COVID-19 occurs due to infection through respiratory droplets containing the SARS-CoV-2 virus, which are released when someone sneezes, coughs, or talks. The gold-standard exam to detect the virus is Real-Time Polymerase Chain Reaction (RT-PCR); however, this is an expensive test and may require up to 3 days after infection for a reliable result, and if there is high demand, the labs could be overwhelmed, which can cause significant delays in providing results. Biomedical data (oxygen saturation level—SpO2, body temperature, heart rate, and cough) are acquired from individuals and are used to help infer infection by COVID-19, using machine learning algorithms. The goal of this study is to introduce the Integrated Portable Medical Assistant (IPMA), which is a multimodal piece of equipment that can collect biomedical data, such as oxygen saturation level, body temperature, heart rate, and cough sound, and helps infer the diagnosis of COVID-19 through machine learning algorithms. The IPMA has the capacity to store the biomedical data for continuous studies and can be used to infer other respiratory diseases. Quadratic kernel-free non-linear Support Vector Machine (QSVM) and Decision Tree (DT) were applied on three datasets with data of cough, speech, body temperature, heart rate, and SpO2, obtaining an Accuracy rate (ACC) and Area Under the Curve (AUC) of approximately up to 88.0% and 0.85, respectively, as well as an ACC up to 99% and AUC = 0.94, respectively, for COVID-19 infection inference. When applied to the data acquired with the IMPA, these algorithms achieved 100% accuracy. Regarding the easiness of using the equipment, 36 volunteers reported that the IPMA has a high usability, according to results from two metrics used for evaluation: System Usability Scale (SUS) and Post Study System Usability Questionnaire (PSSUQ), with scores of 85.5 and 1.41, respectively. In light of the worldwide needs for smart equipment to help fight the COVID-19 pandemic, this new equipment may help with the screening of COVID-19 through data collected from biomedical signals and cough sounds, as well as the use of machine learning algorithms.

Assessment of an Assistive Control Approach Applied in an Active Knee Orthosis Plus Walker for Post-Stroke Gait Rehabilitation

The goal of this study is the assessment of an assistive control approach applied to an active knee orthosis plus a walker for gait rehabilitation. The study evaluates post-stroke patients and healthy subjects (control group) in terms of kinematics, kinetics, and muscle activity. Muscle and gait information of interest were acquired from their lower limbs and trunk, and a comparison was conducted between patients and control group. Signals from plantar pressure, gait phase, and knee angle and torque were acquired during gait, which allowed us to verify that the stance control strategy proposed here was efficient at improving the patients’ gaits (comparing their results to the control group), without the necessity of imposing a fixed knee trajectory. An innovative evaluation of trunk muscles related to the maintenance of dynamic postural equilibrium during gait assisted by our active knee orthosis plus walker was also conducted through inertial sensors. An increase in gait cycle (stance phase) was also observed when comparing the results of this study to our previous work. Regarding the kinematics, the maximum knee torque was lower for patients when compared to the control group, which implies that our orthosis did not demand from the patients a knee torque greater than that for healthy subjects. Through surface electromyography (sEMG) analysis, a significant reduction in trunk muscle activation and fatigability, before and during the use of our orthosis by patients, was also observed. This suggest that our orthosis, together with the assistive control approach proposed here, is promising and could be considered to complement post-stroke patient gait rehabilitation.

Knee Impedance Modulation to Control an Active Orthosis Using Insole Sensors

Robotic devices for rehabilitation and gait assistance have greatly advanced with the objective of improving both the mobility and quality of life of people with motion impairments. To encourage active participation of the user, the use of admittance control strategy is one of the most appropriate approaches, which requires methods for online adjustment of impedance components. Such approach is cited by the literature as a challenge to guaranteeing a suitable dynamic performance. This work proposes a method for online knee impedance modulation, which generates variable gains through the gait cycle according to the users' anthropometric data and gait sub-phases recognized with footswitch signals. This approach was evaluated in an active knee orthosis with three variable gain patterns to obtain a suitable condition to implement a stance controller: two different gain patterns to support the knee in stance phase, and a third pattern for gait without knee support. The knee angle and torque were measured during the experimental protocol to compare both temporospatial parameters and kinematics data with other studies of gait with knee exoskeletons. The users rated scores related to their satisfaction with both the device and controller through QUEST questionnaires. Experimental results showed that the admittance controller proposed here offered knee support in 50% of the gait cycle, and the walking speed was not significantly different between the three gain patterns (p = 0.067). A positive effect of the controller on users regarding safety during gait was found with a score of 4 in a scale of 5. Therefore, the approach demonstrates good performance to adjust impedance components providing knee support in stance phase.

Adaptive Spatial Filter Based on Similarity Indices to Preserve the Neural Information on EEG Signals during On-Line Processing

This work presents a new on-line adaptive filter, which is based on a similarity analysis between standard electrode locations, in order to reduce artifacts and common interferences throughout electroencephalography (EEG) signals, but preserving the useful information. Standard deviation and Concordance Correlation Coefficient (CCC) between target electrodes and its correspondent neighbor electrodes are analyzed on sliding windows to select those neighbors that are highly correlated. Afterwards, a model based on CCC is applied to provide higher values of weight to those correlated electrodes with lower similarity to the target electrode. The approach was applied to brain computer-interfaces (BCIs) based on Canonical Correlation Analysis (CCA) to recognize 40 targets of steady-state visual evoked potential (SSVEP), providing an accuracy (ACC) of 86.44 ± 2.81%. In addition, also using this approach, features of low frequency were selected in the pre-processing stage of another BCI to recognize gait planning. In this case, the recognition was significantly (p<0.01) improved for most of the subjects (ACC≥74.79%), when compared with other BCIs based on Common Spatial Pattern, Filter Bank-Common Spatial Pattern, and Riemannian Geometry.

Knee motion pattern classification from trunk muscle based on sEMG signals

A prominent change is being carried out in the fields of rehabilitation and assistive exoskeletons in order to actively aid or restore legged locomotion for individuals suffering from muscular impairments, muscle weakness, neurologic injury, or disabilities that affect the lower limbs. This paper presents a characterization of knee motion patterns from Surface Electromyography (sEMG) signals, measured in the Erector spinae (ES) muscle. Feature extraction (mean absolute value, waveform length and auto-regressive model) and pattern classification methods (Linear Discrimination Analysis, K-Nearest Neighborhood and Support Vector Machine) are applied for recognition of eight-movement classes. Additionally, several channels setup are analyzed to obtain a suitable electrodes array. The results were evaluated based on signals measured from lower limb using quantitative metric such as error rate, sensitivity, specificity and predictive positive value. A high accuracy (> 95%) was obtained, which suggest that it is possible to detect the knee motion intention from ES muscle, as well as to reduce the electrode number (from 2 to 3 channels) to obtain an optimal electrodes array. This implementation can be applied for myoelectric control of lower limb active exoskeletons.