Development of a convolutional neural network (CNN) based assessment exercise recommendation system for individuals with chronic stroke: a feasibility study

BACKGROUND

The use of artificial intelligence (AI) is revolutionizing nearly every aspect of healthcare, but the application of AI in rehabilitation is lagging behind. Clinically, gait parameters and patterns are used to evaluate stroke-specific impairment. We hypothesized that gait kinematics of individuals with stroke provide rich information for the deep-learning to predict the clinical decisions made by physiotherapist.

OBJECTIVE

To investigate whether the results of clinical assessments and exercise recommendations by physiotherapists can be accurately predicted using a deep-learning algorithm with gait kinematics data.

METHOD

In this cross-sectional study, 40 individuals with stroke were assessed by a physiotherapist using the lower-extremity subscale of the Fugl-Meyer Assessment (FMA-LE) and Berg Balance Scale (BBS). The physiotherapist also decided whether or not the single-leg-stance was an appropriate balance training for each participant. The participants were classified as having good mobility and a low fall risk based on the cutoff scores of the two clinical scales. A convolutional neural network (CNN) was trained using gait kinematics to predict the assessment results and exercise recommendations.

RESULTS

The trained model accurately predicted the results of the clinical assessments and decisions with an average prediction accuracy of 0.84 for the FMA-LE, 0.66 for the BBS, and 0.78 for the recommendation of the single-leg-stance exercise.

CONCLUSIONS

This CNN deep-learning model provided time-effective and accurate prediction of clinical assessment results and exercise recommendations. This study provides preliminary evidence to support the use of biomechanical data and AI to assist treatment planning and shorten the decision-making process in rehabilitation.

Early stopping by correlating online indicators in neural networks

In order to minimize the generalization error in neural networks, a novel technique to identify overfitting phenomena when training the learner is formally introduced. This enables support of a reliable and trustworthy early stopping condition, thus improving the predictive power of that type of modeling. Our proposal exploits the correlation over time in a collection of online indicators, namely characteristic functions for indicating if a set of hypotheses are met, associated with a range of independent stopping conditions built from a canary judgment to evaluate the presence of overfitting. That way, we provide a formal basis for decision making in terms of interrupting the learning process. As opposed to previous approaches focused on a single criterion, we take advantage of subsidiarities between independent assessments, thus seeking both a wider operating range and greater diagnostic reliability. With a view to illustrating the effectiveness of the halting condition described, we choose to work in the sphere of natural language processing, an operational continuum increasingly based on machine learning. As a case study, we focus on parser generation, one of the most demanding and complex tasks in the domain. The selection of cross-validation as a canary function enables an actual comparison with the most representative early stopping conditions based on overfitting identification, pointing to a promising start toward an optimal bias and variance control.

A novel robust convolutional neural network for uniform resource locator classification from the view of cyber security

Uniform resource locator (URL)‐based cyber‐attacks form a major part of security threats in cyberspace. Even though the experience and awareness of the end‐users help them protect themselves from these attacks, a software‐based solution is necessary for comprehensive protection. To this end, a novel robust URL classification model based on convolutional neural network is proposed in this study. The proposed model classifies given URLs into five classes, namely, () , () , () , () , and () . The proposed model was trained and evaluated on a gold standard URL dataset comprising of samples. According to the experimental result, the proposed model obtained an accuracy as high as which outperformed the state‐of‐the‐art. Based on the same architecture, we proposed another classifier, a binary classifier that detects malicious URLs without dealing with their types. This binary classifier obtained an accuracy as high as which outperformed the state‐of‐the‐art as well. The experimental result demonstrates the feasibility of the proposed solution.

Farmland quality assessment using deep fully convolutional neural networks

Farmland is the cornerstone of agriculture and is important for food security and social production. Farmland assessment is essential but traditional methods are usually expensive and slow. Deep learning methods have been developed and widely applied recently in image recognition, semantic understanding, and many other application domains. In this research, we used fully convolutional networks (FCN) as the deep learning model to evaluate farmland grades. Normalized difference vegetation index (NDVI) derived from Landsat images was used as the input data, and the China National Cultivated Land Grade Database within Jiangsu Province was used to train the model on cloud computing. We also applied an image segmentation method to improve the original results from the FCN and compared the results with classical machine learning (ML) methods. Our research found that the FCN can predict farmland grades with an overall F1 score (the harmonic mean of precision and recall) of 0.719 and F1 score of 0.909, 0.590, 0.740, 0.642, and 0.023 for non-farmland, level I, II, III, and IV farmland, respectively. Combining the FCN and image segmentation method can further improve prediction accuracy with results of fewer noise pixels and more realistic edges. Compared with conventional ML, at least in farmland evaluation, FCN provides better results with higher precision, recall, and F1 score. Our research indicates that by using remote sensing NDVI data, the deep learning method can provide acceptable farmland assessment without fieldwork and can be used as a novel supplement to traditional methods. The method used in this research will save a lot of time and cost compared with traditional means.

Newborn Cry-Based Diagnostic System to Distinguish between Sepsis and Respiratory Distress Syndrome Using Combined Acoustic Features

Crying is the only means of communication for a newborn baby with its surrounding environment, but it also provides significant information about the newborn's health, emotions, and needs. The cries of newborn babies have long been known as a biomarker for the diagnosis of pathologies. However, to the best of our knowledge, exploring the discrimination of two pathology groups by means of cry signals is unprecedented. Therefore, this study aimed to identify septic newborns with Neonatal Respiratory Distress Syndrome (RDS) by employing the Machine Learning (ML) methods of Multilayer Perceptron (MLP) and Support Vector Machine (SVM). Furthermore, the cry signal was analyzed from the following two different perspectives: 1) the musical perspective by studying the spectral feature set of Harmonic Ratio (HR), and 2) the speech processing perspective using the short-term feature set of Gammatone Frequency Cepstral Coefficients (GFCCs). In order to assess the role of employing features from both short-term and spectral modalities in distinguishing the two pathology groups, they were fused in one feature set named the combined features. The hyperparameters (HPs) of the implemented ML approaches were fine-tuned to fit each experiment. Finally, by normalizing and fusing the features originating from the two modalities, the overall performance of the proposed design was improved across all evaluation measures, achieving accuracies of 92.49% and 95.3% by the MLP and SVM classifiers, respectively. The MLP classifier was outperformed in terms of all evaluation measures presented in this study, except for the Area Under Curve of Receiver Operator Characteristics (AUC-ROC), which signifies the ability of the proposed design in class separation. The achieved results highlighted the role of combining features from different levels and modalities for a more powerful analysis of the cry signals, as well as including a neural network (NN)-based classifier. Consequently, attaining a 95.3% accuracy for the separation of two entangled pathology groups of RDS and sepsis elucidated the promising potential for further studies with larger datasets and more pathology groups.

A novel COVID-19 sentiment analysis in Turkish based on the combination of convolutional neural network and bidirectional long-short term memory on Twitter

The whole world has been experiencing the COVID-19 pandemic since December 2019. During the pandemic, a new life has been started by necessity where people have extensively used social media to express their feelings, and find information. Twitter was used as the source of what people have shared regarding the COVID-19 pandemic. Sentiment analysis deals with the extraction of the sentiment of a given text. Most of the related works deal with sentiment analysis in English, while studies for Turkish sentiment analysis lack in the research field. To this end, a novel sentiment analysis model based on the combination of convolutional neural network and bidirectional long short-term memory was proposed in this study. The proposed deep neural network model was trained on the constructed Twitter dataset, which consists of 15 k Turkish tweets regarding the COVID-19 pandemic, to classify a given tweet into three sentiment classes, namely, (i) positive , (ii) negative , and (iii) neutral . A set of experiments were conducted for the evaluation of the proposed model. According to the experimental result, the proposed model obtained an accuracy as high as 97.895 % , which outperformed the state-of-the-art baseline models for sentiment analysis of tweets in Turkish.

JMASM 55: MATLAB Algorithms and Source Codes of 'cbnet' Function for Univariate Time Series Modeling with Neural Networks (MATLAB)

Artificial Neural Networks (ANN) can be designed as a nonparametric tool for time series modeling. MATLAB serves as a powerful environment for ANN modeling. Although Neural Network Time Series Tool (ntstool) is useful for modeling time series, more detailed functions could be more useful in order to get more detailed and comprehensive analysis results. For these purposes, cbnet function with properties such as input lag generator, step-ahead forecaster, trial-error based network selection strategy, alternative network selection with various performance measure and global repetition feature to obtain more alternative network has been developed, and MATLAB algorithms and source codes has been introduced. A detailed comparison with the ntstool is carried out, showing that the cbnet function covers the shortcomings of ntstool.

Feature Merged Network for Oil Spill Detection Using SAR Images

The frequency of marine oil spills has increased in recent years. The growing exploitation of marine oil and continuous increase in marine crude oil transportation has caused tremendous damage to the marine ecological environment. Using synthetic aperture radar (SAR) images to monitor marine oil spills can help control the spread of oil spill pollution over time and reduce the economic losses and environmental pollution caused by such spills. However, it is a significant challenge to distinguish between oil-spilled areas and oil-spill-like in SAR images. Semantic segmentation models based on deep learning have been used in this field to address this issue. In addition, this study is dedicated to improving the accuracy of the U-Shape Network (UNet) model in identifying oil spill areas and oil-spill-like areas and alleviating the overfitting problem of the model; a feature merge network (FMNet) is proposed for image segmentation. The global features of SAR image, which are high-frequency component in the frequency domain and represents the boundary between categories, are obtained by a threshold segmentation method. This can weaken the impact of spot noise in SAR image. Then high-dimensional features are extracted from the threshold segmentation results using convolution operation. These features are superimposed with to the down sampling and combined with the high-dimensional features of original image. The proposed model obtains more features, which allows the model to make more accurate decisions. The overall accuracy of the proposed method increased by 1.82% and reached 61.90% compared with the UNet. The recognition accuracy of oil spill areas and oil-spill-like areas increased by approximately 3% and reached 56.33%. The method proposed in this paper not only improves the recognition accuracy of the original model, but also alleviates the overfitting problem of the original model and provides a more effective monitoring method for marine oil spill monitoring. More importantly, the proposed method provides a design principle that opens up new development ideas for the optimization of other deep learning network models.

Prediction of Surface Roughness of an Abrasive Water Jet Cut Using an Artificial Neural Network

The study’s primary purpose was to explore the abrasive water jet (AWJ) cut machinability of stainless steel X5CrNi18-10 (1.4301). The study analyzed the effects of such process parameters as the traverse speed (TS), the depth of cut (DC), and the abrasive mass flow rate (AR) on the surface roughness (Ra) concerning the thickness of the workpiece. Three different thicknesses were cut under different conditions; the Ra was measured at the top, in the middle, and the bottom of the cut. Experimental results were used in the developed feed-forward artificial neural network (ANN) to predict the Ra. The ANN’s model was validated using k-fold cross-validation. A lowest test root mean squared error (RMSE) of 0.2084 was achieved. The results of the predicted Ra by the ANN model and the results of the experimental data were compared. Additionally, as TS and DC were recognized, analysis of variance at a 95% confidence level was used to determine the most significant factors. Consequently, the ANN input parameters were modified, resulting in improved prediction; results show that the proposed model could be a useful tool for optimizing AWJ cut process parameters for predicting Ra. Its main advantage is the reduced time needed for experimentation.

Estimations of giant dipole resonance parameters using artificial neural network

In this study; Giant Dipole Resonance (GDR) parameters of the spherical nucleus have been estimated by using artificial neural network (ANN) algorithms. The ANN training has been carried out with the Levenberg-Marquardt feed-forward algorithm in order to provide fast convergence and stability in ANN training and experimental data, taken from Reference Input Parameter Library (RIPL). R values of the system have been found as 0.99636, 0.94649, and 0.98318 for resonance energy, full width half maximum, and resonance cross-section, respectively. Obtained results have been compared with the GDR parameters which are taken from the literature. To validate our findings, newly acquired GDR parameters were then replaced with the existing GDR parameters in the TALYS 1.95 code and ^142-146Nd(γ,n)^141-145Nd reaction cross-sections have been calculated and compared with the experimental data taken from the literature. As a result of the study, it has been shown that ANN algorithms can be used to calculate the GDR parameters in the absence of the experimental data.

(Quasi-)Real-Time Inversion of Airborne Time-Domain Electromagnetic Data via Artificial Neural Network

The possibility to have results very quickly after, or even during, the collection of electromagnetic data would be important, not only for quality check purposes, but also for adjusting the location of the proposed flight lines during an airborne time-domain acquisition. This kind of readiness could have a large impact in terms of optimization of the Value of Information of the measurements to be acquired. In addition, the importance of having fast tools for retrieving resistivity models from airborne time-domain data is demonstrated by the fact that Conductivity-Depth Imaging methodologies are still the standard in mineral exploration. In fact, they are extremely computationally efficient, and, at the same time, they preserve a very high lateral resolution. For these reasons, they are often preferred to inversion strategies even if the latter approaches are generally more accurate in terms of proper reconstruction of the depth of the targets and of reliable retrieval of true resistivity values of the subsurface. In this research, we discuss a novel approach, based on neural network techniques, capable of retrieving resistivity models with a quality comparable with the inversion strategy, but in a fraction of the time. We demonstrate the advantages of the proposed novel approach on synthetic and field datasets.

Drone Image Segmentation Using Machine and Deep Learning for Mapping Raised Bog Vegetation Communities

The application of drones has recently revolutionised the mapping of wetlands due to their high spatial resolution and the flexibility in capturing images. In this study, the drone imagery was used to map key vegetation communities in an Irish wetland, Clara Bog, for the spring season. The mapping, carried out through image segmentation or semantic segmentation, was performed using machine learning (ML) and deep learning (DL) algorithms. With the aim of identifying the most appropriate, cost-efficient, and accurate segmentation method, multiple ML classifiers and DL models were compared. Random forest (RF) was identified as the best pixel-based ML classifier, which provided good accuracy (≈85%) when used in conjunction graph cut algorithm for image segmentation. Amongst the DL networks, a convolutional neural network (CNN) architecture in a transfer learning framework was utilised. A combination of ResNet50 and SegNet architecture gave the best semantic segmentation results (≈90%). The high accuracy of DL networks was accompanied with significantly larger labelled training dataset, computation time and hardware requirements compared to ML classifiers with slightly lower accuracy. For specific applications such as wetland mapping where networks are required to be trained for each different site, topography, season, and other atmospheric conditions, ML classifiers proved to be a more pragmatic choice.

Classifying Parkinson's Disease Based on Acoustic Measures Using Artificial Neural Networks

In recent years, neural networks have become very popular in all kinds of prediction problems. In this paper, multiple feed-forward artificial neural networks (ANNs) with various configurations are used in the prediction of Parkinson’s disease (PD) of tested individuals, based on extracted features from 26 different voice samples per individual. Results are validated via the leave-one-subject-out (LOSO) scheme. Few feature selection procedures based on Pearson’s correlation coefficient, Kendall’s correlation coefficient, principal component analysis, and self-organizing maps, have been used for boosting the performance of algorithms and for data reduction. The best test accuracy result has been achieved with Kendall’s correlation coefficient-based feature selection, and the most relevant voice samples are recognized. Multiple ANNs have proven to be the best classification technique for diagnosis of PD without usage of the feature selection procedure (on raw data). Finally, a neural network is fine-tuned, and a test accuracy of 86.47% was achieved.

Dynamic system classifier

Stochastic differential equations describe well many physical, biological, and sociological systems, despite the simplification often made in their derivation. Here the usage of simple stochastic differential equations to characterize and classify complex dynamical systems is proposed within a Bayesian framework. To this end, we develop a dynamic system classifier (DSC). The DSC first abstracts training data of a system in terms of time-dependent coefficients of the descriptive stochastic differential equation. Thereby the DSC identifies unique correlation structures within the training data. For definiteness we restrict the presentation of the DSC to oscillation processes with a time-dependent frequency ω(t) and damping factor γ(t). Although real systems might be more complex, this simple oscillator captures many characteristic features. The ω and γ time lines represent the abstract system characterization and permit the construction of efficient signal classifiers. Numerical experiments show that such classifiers perform well even in the low signal-to-noise regime.

MLPNN Training via a Multiobjective Optimization of Training Error and Stochastic Sensitivity

The training of a multilayer perceptron neural network (MLPNN) concerns the selection of its architecture and the connection weights via the minimization of both the training error and a penalty term. Different penalty terms have been proposed to control the smoothness of the MLPNN for better generalization capability. However, controlling its smoothness using, for instance, the norm of weights or the Vapnik-Chervonenkis dimension cannot distinguish individual MLPNNs with the same number of free parameters or the same norm. In this paper, to enhance generalization capabilities, we propose a stochastic sensitivity measure (ST-SM) to realize a new penalty term for MLPNN training. The ST-SM determines the expectation of the squared output differences between the training samples and the unseen samples located within their Q -neighborhoods for a given MLPNN. It provides a direct measurement of the MLPNNs output fluctuations, i.e., smoothness. We adopt a two-phase Pareto-based multiobjective training algorithm for minimizing both the training error and the ST-SM as biobjective functions. Experiments on 20 UCI data sets show that the MLPNNs trained by the proposed algorithm yield better accuracies on testing data than several recent and classical MLPNN training methods.

A novel single neuron perceptron with universal approximation and XOR computation properties

We propose a biologically motivated brain-inspired single neuron perceptron (SNP) with universal approximation and XOR computation properties. This computational model extends the input pattern and is based on the excitatory and inhibitory learning rules inspired from neural connections in the human brain's nervous system. The resulting architecture of SNP can be trained by supervised excitatory and inhibitory online learning rules. The main features of proposed single layer perceptron are universal approximation property and low computational complexity. The method is tested on 6 UCI (University of California, Irvine) pattern recognition and classification datasets. Various comparisons with multilayer perceptron (MLP) with gradient decent backpropagation (GDBP) learning algorithm indicate the superiority of the approach in terms of higher accuracy, lower time, and spatial complexity, as well as faster training. Hence, we believe the proposed approach can be generally applicable to various problems such as in pattern recognition and classification.

Real-time parallel processing of grammatical structure in the fronto-striatal system: a recurrent network simulation study using reservoir computing

Sentence processing takes place in real-time. Previous words in the sentence can influence the processing of the current word in the timescale of hundreds of milliseconds. Recent neurophysiological studies in humans suggest that the fronto-striatal system (frontal cortex, and striatum--the major input locus of the basal ganglia) plays a crucial role in this process. The current research provides a possible explanation of how certain aspects of this real-time processing can occur, based on the dynamics of recurrent cortical networks, and plasticity in the cortico-striatal system. We simulate prefrontal area BA47 as a recurrent network that receives on-line input about word categories during sentence processing, with plastic connections between cortex and striatum. We exploit the homology between the cortico-striatal system and reservoir computing, where recurrent frontal cortical networks are the reservoir, and plastic cortico-striatal synapses are the readout. The system is trained on sentence-meaning pairs, where meaning is coded as activation in the striatum corresponding to the roles that different nouns and verbs play in the sentences. The model learns an extended set of grammatical constructions, and demonstrates the ability to generalize to novel constructions. It demonstrates how early in the sentence, a parallel set of predictions are made concerning the meaning, which are then confirmed or updated as the processing of the input sentence proceeds. It demonstrates how on-line responses to words are influenced by previous words in the sentence, and by previous sentences in the discourse, providing new insight into the neurophysiology of the P600 ERP scalp response to grammatical complexity. This demonstrates that a recurrent neural network can decode grammatical structure from sentences in real-time in order to generate a predictive representation of the meaning of the sentences. This can provide insight into the underlying mechanisms of human cortico-striatal function in sentence processing.