Agricultural vehicles operating in rough environments experience increased fatigue damage accumulation, which may decrease machine safety and reliability. Autonomous agricultural machines offer an opportunity to incorporate fatigue damage considerations into path planning. This work investigates whether machine learning can predict fatigue damage to a tractor chassis using light detection and ranging (LiDAR)-based terrain features, vehicle speed, and rotational vehicle state data (e.g., triaxial angle, angular velocity, and angular acceleration). Fatigue damage was estimated using the Rupp filter and the Durability Transfer Concept. Following poor predictive performance of the machine learning models, an exploratory analysis of damage histograms, dominant frequency, and acceleration magnitude was performed. Results indicated that most estimated fatigue damage occurred in the 0–2 Hz band, which coincides with the frequency range of terrain-induced acceleration. On-road driving led to the greatest fatigue damage, potentially due to the harder driving surface and increased vehicle speed. Differences between root mean square (RMS) acceleration magnitude and fatigue damage indicate that isolated high-magnitude events may have contributed to increased estimated fatigue damage. Several suggestions for future development were identified. Identification of the endurance limit of the tractor chassis will permit the removal of nondamaging events, improving label accuracy. Furthermore, the presence of a front-loader implement may have impacted chassis acceleration. Thus, a comprehensive dataset with multiple implement configurations is needed to determine the influence of implement configuration on dynamics and resultant damage.

Wrist kinematics can provide insight into the development of repetitive strain injuries, which is important particularly in workplace environments. The emergence of markerless motion capture is beginning to revolutionize kinematic assessment such that it can be conducted outside of the laboratory. The purpose of this work was to apply open-source software (OSS) and machine learning (ML) by using DeepLabCut (OSS) to determine anatomical landmark locations and a variety of regression algorithms and neural networks to predict wrist angles. Sixteen participants completed a series of flexion–extension (FE) and radial–ulnar (RUD) range-of-motion (ROM) trials that were captured using a 13-camera VICON optical motion capture system (i.e., the gold standard), as well as 4 GoPro video cameras. DeepLabCut (version 2.3.3) was used to generate a 2D dataset of anatomical landmark coordinates from video obtained from one obliquely oriented GoPro video camera. Anipose (version 1.0.1) was used to generate a 3D dataset from video obtained from four GoPro cameras. Anipose and various ML algorithms were used to determine RUD and FE wrist angles. The algorithms were trained and tested using a 75%:25% data split with four folds for the 2D and 3D datasets. Of the seven ML techniques applied, deep neural networks resulted in the highest prediction accuracy (5.5) for both the 2D and 3D datasets. This was substantially higher than the wrist angle prediction accuracy provided by Anipose (FE−99; RUD−25.2). We found that, excluding cubic regression, all other studied algorithms exhibited reasonable performance that was similar to that reported by previous authors, showing that it is indeed possible to predict wrist kinematics using a low-cost motion capture system. In agreement with past research, the increased MAE for FE is thought to be due to a larger ROM.

The uniqueness of behavioral biometrics (e.g., voice or keystroke patterns) has been challenged by recent works. Statistical attacks have been proposed that infer general population statistics and target behavioral biometrics against a particular victim. We show that despite their success, these approaches require several attempts for successful attacks against different biometrics due to the different nature of overlap in users' behavior for these biometrics. Furthermore, no mechanism has been proposed to date that detects statistical attacks. In this work, we propose a new hypervolumes-based statistical attack and show that unlike existing methods, it (1) is successful against a variety of biometrics, (2) is successful against more users, and (3) requires fewest attempts for successful attacks. More specifically, across five diverse biometrics, for the first attempt, on average our attack is 18 percentage points more successful than the second best (37% vs. 19%). Similarly, for the fifth attack attempt, on average our attack is 18 percentage points more successful than the second best (67% vs. 49%). We propose and evaluate a mechanism that can detect the more devastating statistical attacks. False rejects in biometric systems are common, and by distinguishing statistical attacks from false rejects, our defense improves usability and security. The evaluation of the proposed detection mechanism shows its ability to detect on average 94% of the tested statistical attacks with an average probability of 3% to detect false rejects as a statistical attack. Given the serious threat posed by statistical attacks to biometrics that are used today (e.g., voice), our work highlights the need for defending against these attacks.

Current methods for ergonomic assessment often use video-analysis to estimate wrist postures during occupational tasks. Wearable sensing and machine learning have the potential to automate this tedious task, and in doing so greatly extend the amount of data available to clinicians and researchers. A method of predicting wrist posture from inertial measurement units placed on the wrist and hand via a deep convolutional neural network has been developed. This study has quantified the accuracy and reliability of the postures predicted by this system relative to the gold standard of optoelectronic motion capture. Ten participants performed 3 different simulated occupational tasks on 2 occasions while wearing inertial measurement units on the hand and wrist. Data from the occupational task recordings were used to train a convolutional neural network classifier to estimate wrist posture in flexion/extension, and radial/ulnar deviation. The model was trained and tested in a leave-one-out cross validation format. Agreement between the proposed system and optoelectronic motion capture was 65% with κ = 0.41 in flexion/extension and 60% with κ = 0.48 in radial/ulnar deviation. The proposed system can predict wrist posture in flexion/extension and radial/ulnar deviation with accuracy and reliability congruent with published values for human estimators. This system can estimate wrist posture during occupational tasks in a small fraction of the time it takes a human to perform the same task. This offers opportunity to expand the capabilities of practitioners by eliminating the tedium of manual postural assessment.

Individual responses to fatigue have been observed in lifting kinematics, suggesting a subject-specific approach is necessary for fatigue identification. One-class support vector machines (OCSVM) may provide an objective method to classify fatigue-related kinematic changes during repetitive lifting. Participants completed a repetitive lifting protocol while motion capture recorded lifting motions. Subject-specific kinematics from participants' first 35% of lifts trained OCSVM decision boundaries. The remaining lifts were separated into test sets and classified against the decision boundary to identify the percentage of outlier lifts within each test set. Spearman's correlation assessed if the test sets' percentage of outlier lifts increased concurrently with participants' rating of perceived exertion (RPE). Significant positive associations were found for participants who demonstrated evidence of fatigue, while no significant associations were found for participants who did not demonstrate evidence of fatigue. These results demonstrate the prospective efficacy of an outlier detection tool for fatigue detection during repetitive lifting.Practitioner Summary: An objective subject-specific fatigue detection method is desired for workplace tasks, such as lifting. An outlier detection machine learning approach was identified when lifting movement patterns changed from baseline throughout a repetitive lifting protocol. Participants who demonstrated an increase in outlier movement patterns had a concurrent increase in self-reported fatigue.

Unsupervised learning algorithms are valuable for exploring a variety of data domains. In this paper we compare the efficacy of the k-means and DBSCAN algorithms in the context of discerning structure in electrodermal data obtained using two different collection modalities for simultaneously collected data: the "gold standard" Biopac data platform and the wearable Empatica E4. Insights into the structure of the data from each system are provided, as is an analysis of the performance of each clustering algorithm at identifying interesting structure within the data.

COVID-19 is a highly contagious infectious disease that has infected millions of people worldwide. Polymerase Chain Reaction (PCR) is the gold standard diagnostic test available for COVID-19 detection. Alternatively, medical imaging techniques, including chest X-ray (CXR), has been instrumental in diagnosis and prognosis of patients with COVID-19. Enabling the CXR with machine learning-based automated diagnosis will be important for rapid diagnosis of the disease by minimizing manual assessment of images by the radiologists. In this work, we developed a deep learning model that utilizes the transfer learning approach using a pre-trained Residual Network model. The Residual Network 50 (ResNet50) is trained from scratch by utilizing the initial architecture and pre-trained weights to provide the classification results. Two types of classification (two-class and three-class) is performed using the developed model. A cascaded approach is adopted for two-class classification where the classification is performed in two phases. The dataset used for training and evaluating the model comprises of 8,254 images in total out of which 1651 images were considered for testing the cascaded model (15 COVID-19) and three-class classification (51 COVID-19). The model was evaluated using accuracy, sensitivity, specificity, and F1-score metrics. Our cascaded model yielded an accuracy of 91.8% for classification of abnormal and normal cases and 97.9% for the classification of pneumonia and COVID-19 images. In the three-class classification, our model reported an accuracy of 92% in classifying normal, pneumonia (bacterial and viral) and COVID-19 cases.

Mouse behavior is a primary outcome in evaluations of therapeutic efficacy. Exhaustive, continuous, multiparametric behavioral phenotyping is a valuable tool for understanding the pathophysiological status of mouse brain diseases. Automated home cage behavior analysis produces highly granulated data both in terms of number of features and sampling frequency. Previously, we demonstrated several ways to reduce feature dimensionality. In this study, we propose novel approaches for analyzing 33-Hz data generated by CleverSys software. We hypothesized that behavioral patterns within short time windows are reflective of physiological state, and that computer modeling of mouse behavioral routines can serve as a predictive tool in classification tasks. To remove bias due to researcher decisions, our data flow is indifferent to the quality, value, and importance of any given feature in isolation. To classify day and night behavior, as an example application, we developed a data preprocessing flow and utilized logistic regression (LG), support vector machines (SVM), random forest (RF), and one-dimensional convolutional neural networks paired with long short-term memory deep neural networks (1DConvBiLSTM). We determined that a 5-min video clip is sufficient to classify mouse behavior with high accuracy. LG, SVM, and RF performed similarly, predicting mouse behavior with 85% accuracy, and combining the three algorithms in an ensemble procedure increased accuracy to 90%. The best performance was achieved by combining the 1DConv and BiLSTM algorithms yielding 96% accuracy. Our findings demonstrate that computer modeling of the home-cage ethome can clearly define mouse physiological state. Furthermore, we showed that continuous behavioral data can be analyzed using approaches similar to natural language processing. These data provide proof of concept for future research in diagnostics of complex pathophysiological changes that are accompanied by changes in behavioral profile.

Purpose: The objective of this study is to develop and evaluate a fully automated, deep learning-based method for detection of COVID-19 infection from chest x-ray images. Approach: The proposed model was developed by replacing the final classifier layer in DenseNet201 with a new network consisting of global averaging layer, batch normalization layer, a dense layer with ReLU activation, and a final classification layer. Then, we performed an end-to-end training using the initial pretrained weights on all the layers. Our model was trained using a total of 8644 images with 4000 images each in normal and pneumonia cases and 644 in COVID-19 cases representing a large real dataset. The proposed method was evaluated based on accuracy, sensitivity, specificity, ROC curve, and F1-score using a test dataset comprising 1729 images (129 COVID-19, 800 normal, and 800 pneumonia). As a benchmark, we also compared the results of our method with those of seven state-of-the-art pretrained models and with a lightweight CNN architecture designed from scratch. Results: The proposed model based on DenseNet201 was able to achieve an accuracy of 94% in detecting COVID-19 and an overall accuracy of 92.19%. The model was able to achieve an AUC of 0.99 for COVID-19, 0.97 for normal, and 0.97 for pneumonia. The model was able to outperform alternative models in terms of overall accuracy, sensitivity, and specificity. Conclusions: Our proposed automated diagnostic model yielded an accuracy of 94% in the initial screening of COVID-19 patients and an overall accuracy of 92.19% using chest x-ray images.

We present a novel visualization tool designed to provide support for the analysis of data sets focused around deviation from a baseline and including data from multiple series. The incorporation of a floating baseline makes the curtain graph distinct from waterfall plots and bar charts. Each data series therefore has a visual anchor that assists in interpretability, gives focus, and provides a means for easily broadening analysis across all presented series. The use of this tool in real-world examples based on relative and absolute comparisons is discussed.

This work presents machine learning-based techniques for detecting mind-wandering and predicting hazard response time in driving using only easily measurable driving performance data (speed, horizontal and frontal acceleration, lane gap, and brake pressure). Such classifiers are relevant as research tools in the driving simulation community. We present a simple method, and a feature extraction-based method, of representing time-series driving performance data that both support machine learning-based predictions. We use the two types of representations to compare the effectiveness of support vector machines, random forest, and multi-layer perceptrons on data from 117 drives performed by 39 participants during a previous study in the high-fidelity driving simulator at the University of Guelph. Classification of mind-wandering and prediction of hazard response time was successful when compared to baseline measures. Specifically, random forest methods were most effective in both types of prediction and feature extraction supported the strongest random forest prediction of hazard response time. A discussion of the reasoning for this is included. While this has previously been studied using a single screen experimental setup, to our knowledge this is the first driving pattern-based classification of mind-wandering in a fully immersive driving simulator.

NEFCLASS is a common example of a neuro-fuzzy system. The popular NEFCLASS classifier exhibits surprising behaviour when the feature values of the training and testing datasets exhibit significant skew. This paper presents a combined approach to improve the classification accuracy and interpretability of the NEFCLASS classifier, when data distribution exhibits positive skewness. The proposed model consists of two steps. Firstly, we used an alternative discretization method to initialize fuzzy sets. Secondly, we devised a statistical rule pruning algorithm based on adjusted residual to reduce the number of rules, thus improving interpretability. This method improves the interpretability of NEFCLASS without significant accuracy deterioration. Moreover, a hybrid approach combining the two approaches is developed to increase the accuracy-interpretability trade-off of NEFCLASS.

Office workers frequently report discomfort in the neck, back and shoulders. Extreme and prolonged static postures are known risk factors. Sit-stand workstations encourage regular postural changes and tend to reduce discomfort. Head and upper back deviations were recorded for ten adults entering data using a seated desk and two sit-stand desks following three twenty-minute cycles with 6 or 9 min standing. Participants reported their perceived back pain, neck pain and fatigue following each recording.

Nefclass is a common example of the construction of a neuro-fuzzy system. The popular Nefclass classifier exhibits surprising behaviour when the feature values of the training and testing data sets exhibit significant skew. As skewed feature values are commonly observed in biological data sets, this is a topic that is of interest in terms of the applicability of such a classifier to these types of problems. This paper presents an approach to improve the classification accuracy of the Nefclass classifier, when data distribution exhibits positive skewness. The Nefclass classifier is extended to provide improved classification accuracy over the original Nefclass classifier when trained on skewed data. The proposed model uses two alternative discretization methods, MME and CAIM, to initialize fuzzy sets. From this study it is found that using the MME and CAIM discretization methods results in greater improvements in the classification accuracy of Nefclass as compared to using the original Equal-Width technique Nefclass uses by default.

A motor unit potential (MUP) template, which represents the shapes of the MUPs within a MUP train, provides information related to the morphology and physiology of the sampled motor unit. This work presents an improved MUP template estimation technique that uses local time warping and kernel weighted ensemble averaging. An analysis of the algorithm, and a description of the improvements compared with spike triggered averaging is given. MUP template estimates were evaluated using simulated EMG signals with a known gold standard template for each motor unit potential train. Statistically significant reduction in template estimation error is shown, both within the baseline and duration portions of a MUP.

Muscle motor unit numbers decrease markedly in old age, while remaining motor units are enlarged and can have reduced neuromuscular junction transmission stability. However, it is possible that regular intense physical activity throughout life can attenuate this remodeling. The aim of this study was to compare the number, size, and neuromuscular junction transmission stability of tibialis anterior (TA) motor units in healthy young and older men with those of exceptionally active master runners. The distribution of motor unit potential (MUP) size was determined from intramuscular electromyographic signals recorded in healthy male Young (mean +/- SD, 26 +/- 5 years), Old (71 +/- 4 years) and Master Athletes (69 +/- 3 years). Relative differences between groups in numbers of motor units was assessed using two methods, one comparing MUP size and muscle cross-sectional area (CSA) determined with MRI, the other comparing surface recorded MUPs with maximal compound muscle action potentials and commonly known as a "motor unit number estimate (MUNE)". Near fiber (NF) jiggle was measured to assess neuromuscular junction transmission stability. TA CSA did not differ between groups. MUNE values for the Old and Master Athletes were 45% and 40%, respectively, of the Young. Intramuscular MUPs of Old and Master Athletes were 43% and 56% larger than Young. NF jiggle was slightly higher in the Master Athletes, with no difference between Young and Old. These results show substantial and similar motor unit loss and remodeling in Master Athletes and Old individuals compared with Young, which suggests that lifelong training does not attenuate the age-related loss of motor units.

This paper presents a model for the treatment of skewness effects on the accuracy of the NEFCLASS classifier by changing the embedded discretization method within the classifier. NEFCLASS is a common example of the construction of a neuro-fuzzy system. The popular NEFCLASS classifier exhibits surprising behaviour when the feature values of the training and testing data sets exhibit significant skew. As skewed feature values are commonly observed in biological data sets, this is a topic that is of interest in terms of the applicability of such a classifier to these types of problems. From this study it is clear that the effect of skewness on classification accuracy is significant, and that this must be considered in work dealing with skewed data distributions. We compared accuracy of NEFCLASS classifier with two modified versions of NEFCLASS embedded with MME and CAIM discretization methods. From this study it is found the CAIM and MME discretization methods results in greater improvements in the classification accuracy of NEFCLASS as compared to using the original EQUAL-WIDTH technique NEFCLASS uses by default.

[This corrects the article DOI: 10.1371/journal.pone.0104572.].