A time series forest for classification and feature extraction

Machine learning for automated electrical penetration graph analysis of aphid feeding behavior: Accelerating research on insect-plant interactions

The electrical penetration graph (EPG) is a well-known technique that provides insights into the feeding behavior of insects with piercing-sucking mouthparts, mostly hemipterans. Since its inception in the 1960s, EPG has become indispensable in studying plant-insect interactions, revealing critical information about host plant selection, plant resistance, virus transmission, and responses to environmental factors. By integrating the plant and insect into an electrical circuit, EPG allows researchers to identify specific feeding behaviors based on their distinctive waveform patterns. However, the traditional manual analysis of EPG waveform data is time-consuming and labor-intensive, limiting research throughput. This study presents a novel Machine Learning (ML) approach to automate the annotation of EPG signals. We rigorously evaluated six diverse ML models, including neural networks, tree-based models, and logistic regression, using an extensive dataset from multiple aphid feeding experiments. Our results demonstrate that a Residual Network (ResNet) architecture achieved the highest overall waveform classification accuracy of 96.8% and highest segmentation overlap rate of 84.4%, highlighting the potential of ML for accurate and efficient EPG analysis. This automated approach promises to accelerate research in this field significantly and broaden insights into insect-plant interactions, showcasing the power of computational techniques for insect biological research. The source code for all experiments conducted within this study is publicly available at https://github.com/HySonLab/ML4Insects.

Separation of stroke from vestibular neuritis using the video head impulse test: machine learning models versus expert clinicians

BACKGROUND

Acute vestibular syndrome usually represents either vestibular neuritis (VN), an innocuous viral illness, or posterior circulation stroke (PCS), a potentially life-threatening event. The video head impulse test (VHIT) is a quantitative measure of the vestibulo-ocular reflex that can distinguish between these two diagnoses. It can be rapidly performed at the bedside by any trained healthcare professional but requires interpretation by an expert clinician. We developed machine learning models to differentiate between PCS and VN using only the VHIT.

METHODS

We trained machine learning classification models using unedited head- and eye-velocity data from acute VHIT performed in an Emergency Room on patients presenting with acute vestibular syndrome and whose final diagnosis was VN or PCS. The models were validated using an independent test dataset collected at a second institution. We compared the performance of the models against expert clinicians as well as a widely used VHIT metric: the gain cutoff value.

RESULTS

The training and test datasets comprised 252 and 49 patients, respectively. In the test dataset, the best machine learning model identified VN with 87.8% (95% CI 77.6%-95.9%) accuracy. Model performance was not significantly different (p = 0.56) from that of blinded expert clinicians who achieved 85.7% accuracy (75.5%-93.9%) and was superior (p = 0.01) to that of the optimal gain cutoff value (75.5% accuracy (63.8%-85.7%)).

CONCLUSION

Machine learning models can effectively differentiate PCS from VN using only VHIT data, with comparable accuracy to expert clinicians. They hold promise as a tool to assist Emergency Room clinicians evaluating patients with acute vestibular syndrome.

A novel approach to the cause of death identification-multi-strategy integration of multi-organ FTIR spectroscopy information using machine learning

Identifying the cause of death has always been a major focus and challenge in forensic practice and research. Traditional techniques for determining the causes of death are time-consuming, labor-intensive, have high professional barriers, and are vulnerable to significant subjective bias. Additionally, most current studies on causes of death are limited to specific organs and single causes. To overcome these challenges, this study utilized simple and rapid fourier transform infrared spectroscopy (FTIR) detection technology, integrating data from six organs-heart, liver, spleen, lung, kidney, and brain. The optimum model for identifying seven different causes of death was determined by evaluating the performance of models developed using the model efficiencies of single-organ (SO), single-organ model fusion (SOMF), multi-organ data fusion (MODF), and multi-organ data model fusion (MODMF) modeling methods. Considering factors such as operational costs, model performance, and model complexity, the MODF artificial neural network (ANN) model was found to be the most suitable choice for constructing a cause of death identification model, with a cross-validation mean accuracy of 0.960 and a test set accuracy of 0.952. The heart and kidney contributed more spectral features to the construction of the cause of death identification model compared to other organs. This study not only demonstrated that data fusion and model fusion are effective strategies for improving model performance but also provided a comprehensive data analysis framework and process for modeling with small sample multi-modal data (multiple organ data). In conclusion, by exploring various approaches, this study offers new solutions for identifying the cause of death.

Language task-based fMRI analysis using machine learning and deep learning

Introduction

Task-based language fMRI is a non-invasive method of identifying brain regions subserving language that is used to plan neurosurgical resections which potentially encroach on eloquent regions. The use of unstructured fMRI paradigms, such as naturalistic fMRI, to map language is of increasing interest. Their analysis necessitates the use of alternative methods such as machine learning (ML) and deep learning (DL) because task regressors may be difficult to define in these paradigms.

Methods

Using task-based language fMRI as a starting point, this study investigates the use of different categories of ML and DL algorithms to identify brain regions subserving language. Data comprising of seven task-based language fMRI paradigms were collected from 26 individuals, and ML and DL models were trained to classify voxel-wise fMRI time series.

Results

The general machine learning and the interval-based methods were the most promising in identifying language areas using fMRI time series classification. The geneal machine learning method achieved a mean whole-brain Area Under the Receiver Operating Characteristic Curve (AUC) of 0.97 ± 0.03 , mean Dice coefficient of 0.6 ± 0.34 and mean Euclidean distance of 2.7 ± 2.4  mm between activation peaks across the evaluated regions of interest. The interval-based method achieved a mean whole-brain AUC of 0.96 ± 0.03 , mean Dice coefficient of 0.61 ± 0.33 and mean Euclidean distance of 3.3 ± 2.7  mm between activation peaks across the evaluated regions of interest.

Discussion

This study demonstrates the utility of different ML and DL methods in classifying task-based language fMRI time series. A potential application of these methods is the identification of language activation from unstructured paradigms.

Cross-dataset Evaluation of Dementia Longitudinal Progression Prediction Models

Accurate Alzheimer's Disease (AD) progression prediction is essential for early intervention. The TADPOLE challenge, involving 92 algorithms, used multimodal biomarkers to predict future clinical diagnosis, cognition, and ventricular volume. The winning algorithm, FROG, utilized a Longitudinal-to-Cross-sectional (L2C) transformation to convert variable longitudinal histories into fixed-length feature vectors, which contrasted with most existing approaches that fitted models to entire longitudinal histories, e.g., AD Course Map (AD-Map) and minimal recurrent neural networks (MinimalRNN). The TADPOLE challenge only utilized the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. To evaluate FROG's generalizability, we trained it on the ADNI dataset and tested it on three external datasets covering 2,312 participants and 13,200 timepoints. We also introduced two FROG variants. One variant, L2C feedforward neural network (L2C-FNN), unified all XGBoost models used by the original FROG with an FNN. Across external datasets, L2C-FNN and AD-Map were the best for predicting cognition and ventricular volume. For clinical diagnosis prediction, L2C-FNN was the best, while AD-Map was the worst. L2C-FNN compared favorably with other approaches regardless of the number of observed timepoints, and when predicting from 0 to 6 years into the future, underscoring its potential for long-term dementia progression prediction. Pretrained ADNI models are publicly available: GITHUB_LINK.

Correlation Fuzzy measure of multivariate time series for signature recognition

Distinguishing different time series, which is determinant or stochastic, is an important task in signal processing. In this work, a correlation measure constructs Correlation Fuzzy Entropy (CFE) to discriminate Chaos and stochastic series. It can be employed to distinguish chaotic signals from ARIMA series with different noises. With specific embedding dimensions, we implemented the CFE features by analyzing two available online signature databases MCYT-100 and SVC2004. The accurate rates of the CFE-based models exceed 99.3%.

End-to-End Signal Classification in Signed Cumulative Distribution Transform Space

This paper presents a new end-to-end signal classification method using the signed cumulative distribution transform (SCDT). We adopt a transport generative model to define the classification problem. We then make use of mathematical properties of the SCDT to render the problem easier in transform domain, and solve for the class of an unknown sample using a nearest local subspace (NLS) search algorithm in SCDT domain. Experiments show that the proposed method provides high accuracy classification results while being computationally cheap, data efficient, and robust to out-of-distribution samples with respect to the existing end-to-end classification methods. The implementation of the proposed method in Python language is integrated as a part of the software package PyTransKit [1].

Eigen-entropy based time series signatures to support multivariate time series classification

Most current algorithms for multivariate time series classification tend to overlook the correlations between time series of different variables. In this research, we propose a framework that leverages Eigen-entropy along with a cumulative moving window to derive time series signatures to support the classification task. These signatures are enumerations of correlations among different time series considering the temporal nature of the dataset. To manage dataset's dynamic nature, we employ preprocessing with dense multi scale entropy. Consequently, the proposed framework, Eigen-entropy-based Time Series Signatures, captures correlations among multivariate time series without losing its temporal and dynamic aspects. The efficacy of our algorithm is assessed using six binary datasets sourced from the University of East Anglia, in addition to a publicly available gait dataset and an institutional sepsis dataset from the Mayo Clinic. We use recall as the evaluation metric to compare our approach against baseline algorithms, including dependent dynamic time warping with 1 nearest neighbor and multivariate multi-scale permutation entropy. Our method demonstrates superior performance in terms of recall for seven out of the eight datasets.

The association between PM2.5 components and blood pressure changes in late pregnancy: A combined analysis of traditional and machine learning models

BACKGROUND

PM2.5 is a harmful mixture of various chemical components that pose a challenge in determining their individual and combined health effects due to multicollinearity issues with traditional linear regression models. This study aimed to develop an analytical methodology combining traditional and novel machine learning models to evaluate PM2.5's combined effects on blood pressure (BP) and identify the most toxic components.

METHODS

We measured late-pregnancy BP of 1138 women from the Heshan cohort while simultaneously analyzing 31 PM2.5 components. We utilized multiple linear regression modeling to establish the relationship between PM2.5 components and late-pregnancy BP and applied Random Forest (RF) and generalized Weighted Quantile Sum (gWQS) regression to identify the most toxic components contributing to elevated BP and to quantitatively evaluate the cumulative effect of the PM2.5 component mixtures.

RESULTS

The results revealed that 16 PM2.5 components, such as EC, OC, Ti, Fe, Mn, Cu, Cd, Mg, K, Pb, Se, Na+, K+, Cl-, NO3-, and F-, contributed to elevated systolic blood pressure (SBP), while 26 components, including two carbon components (EC, OC), fourteen metallics (Ti, Fe, Mn, Cr, Mo, Co, Cu, Zn, Cd, Na, Mg, Al, K, Pb), one metalloid (Se), and nine water-soluble ions (Na+, K+, Mg2+, Ca2+, NH4+, Cl-, NO3-, SO42-, F-), contributed to elevated diastolic blood pressure (DBP). Mn and Cr were the most toxic components for elevated SBP and DBP, respectively, as analyzed by RF and gWQS models and verified against each other. Exposure to PM2.5 component mixtures increased SBP by 1.04 mmHg (95% CI: 0.33-1.76) and DBP by 1.13 mmHg (95% CI: 0.47-1.78).

CONCLUSIONS

Our study highlights the effectiveness of combining traditional and novel models as an analytical strategy to quantify the health effects of PM2.5 constituent mixtures.

Enhancing Autonomous Vehicle Decision-Making at Intersections in Mixed-Autonomy Traffic: A Comparative Study Using an Explainable Classifier

The transition to fully autonomous roadways will include a long period of mixed-autonomy traffic. Mixed-autonomy roadways pose a challenge for autonomous vehicles (AVs) which use conservative driving behaviours to safely negotiate complex scenarios. This can lead to congestion and collisions with human drivers who are accustomed to more confident driving styles. In this work, an explainable multi-variate time series classifier, Time Series Forest (TSF), is compared to two state-of-the-art models in a priority-taking classification task. Responses to left-turning hazards at signalized and stop-sign-controlled intersections were collected using a full-vehicle driving simulator. The dataset was comprised of a combination of AV sensor-collected and V2V (vehicle-to-vehicle) transmitted features. Each scenario forced participants to either take ("go") or yield ("no go") priority at the intersection. TSF performed comparably for both the signalized and sign-controlled datasets, although all classifiers performed better on the signalized dataset. The inclusion of V2V data led to a slight increase in accuracy for all models and a substantial increase in the true positive rate of the stop-sign-controlled models. Additionally, incorporating the V2V data resulted in fewer chosen features, thereby decreasing the model complexity while maintaining accuracy. Including the selected features in an AV planning model is hypothesized to reduce the need for conservative AV driving behaviour without increasing the risk of collision.

Research on Fault Detection by Flow Sequence for Industrial Internet of Things in Sewage Treatment Plant Case

Classifying the flow subsequences of sensor networks is an effective way for fault detection in the Industrial Internet of Things (IIoT). Traditional fault detection algorithms identify exceptions by a single abnormal dataset and do not pay attention to the factors such as electromagnetic interference, network delay, sensor sample delay, and so on. This paper focuses on fault detection by continuous abnormal points. We proposed a fault detection algorithm within the module of sequence state generated by unsupervised learning (SSGBUL) and the module of integrated encoding sequence classification (IESC). Firstly, we built a network module based on unsupervised learning to encode the flow sequence of the different network cards in the IIoT gateway, and then combined the multiple code sequences into one integrated sequence. Next, we classified the integrated sequence by comparing the integrated sequence with the encoding fault type. The results obtained from the three IIoT datasets of a sewage treatment plant show that the accuracy of the SSGBUL-IESC algorithm exceeds 90% with subsequence length 10, which is significantly higher than the accuracies of the dynamic time warping (DTW) algorithm and the time series forest (TSF) algorithm. The proposed algorithm reaches the classification requirements for fault detection for the IIoT.

Monitoring Flow-Forming Processes Using Design of Experiments and a Machine Learning Approach Based on Randomized-Supervised Time Series Forest and Recursive Feature Elimination

The machines of WF Maschinenbau process metal blanks into various workpieces using so-called flow-forming processes. The quality of these workpieces depends largely on the quality of the blanks and the condition of the machine. This creates an urgent need for automated monitoring of the forming processes and the condition of the machine. Since the complexity of the flow-forming processes makes physical modeling impossible, the present work deals with data-driven modeling using machine learning algorithms. The main contributions of this work lie in showcasing the feasibility of utilizing machine learning and sensor data to monitor flow-forming processes, along with developing a practical approach for this purpose. The approach includes an experimental design capable of providing the necessary data, as well as a procedure for preprocessing the data and extracting features that capture the information needed by the machine learning models to detect defects in the blank and the machine. To make efficient use of the small number of experiments available, the experimental design is generated using Design of Experiments methods. They consist of two parts. In the first part, a pre-selection of influencing variables relevant to the forming process is performed. In the second part of the design, the selected variables are investigated in more detail. The preprocessing procedure consists of feature engineering, feature extraction and feature selection. In the feature engineering step, the data set is augmented with time series variables that are meaningful in the domain. For feature extraction, an algorithm was developed based on the mechanisms of the r-STSF, a state-of-the-art algorithm for time series classification, extending them for multivariate time series and metric target variables. This feature extraction algorithm itself can be seen as an additional contribution of this work, because it is not tied to the application domain of monitoring flow-forming processes, but can be used as a feature extraction algorithm for multivariate time series classification in general. For feature selection, a Recursive Feature Elimination is employed. With the resulting features, random forests are trained to detect several quality features of the blank and defects of the machine. The trained models achieve good prediction accuracy for most of the target variables. This shows that the application of machine learning is a promising approach for the monitoring of flow-forming processes, which requires further investigation for confirmation.

Physics-informed neural networks for modeling physiological time series for cuffless blood pressure estimation

The bold vision of AI-driven pervasive physiological monitoring, through the proliferation of off-the-shelf wearables that began a decade ago, has created immense opportunities to extract actionable information for precision medicine. These AI algorithms model input-output relationships of a system that, in many cases, exhibits complex nature and personalization requirements. A particular example is cuffless blood pressure estimation using wearable bioimpedance. However, these algorithms need training over significant amount of ground truth data. In the context of biomedical applications, collecting ground truth data, particularly at the personalized level is challenging, burdensome, and in some cases infeasible. Our objective is to establish physics-informed neural network (PINN) models for physiological time series data that would use minimal ground truth information to extract complex cardiovascular information. We achieve this by building Taylor's approximation for gradually changing known cardiovascular relationships between input and output (e.g., sensor measurements to blood pressure) and incorporating this approximation into our proposed neural network training. The effectiveness of the framework is demonstrated through a case study: continuous cuffless BP estimation from time series bioimpedance data. We show that by using PINNs over the state-of-the-art time series models tested on the same datasets, we retain high correlations (systolic: 0.90, diastolic: 0.89) and low error (systolic: 1.3 ± 7.6 mmHg, diastolic: 0.6 ± 6.4 mmHg) while reducing the amount of ground truth training data on average by a factor of 15. This could be helpful in developing future AI algorithms to help interpret pervasive physiologic data using minimal amount of training data.

Learning task-agnostic and interpretable subsequence-based representation of time series and its applications in fMRI analysis

The recent success of sequential learning models, such as deep recurrent neural networks, is largely due to their superior representation-learning capability for learning the informative representation of a targeted time series. The learning of these representations is generally goal-directed, resulting in their task-specific nature, giving rise to excellent performance in completing a single downstream task but hindering between-task generalisation. Meanwhile, with increasingly intricate sequential learning models, learned representation becomes abstract to human knowledge and comprehension. Hence, we propose a unified local predictive model based on the multi-task learning paradigm to learn the task-agnostic and interpretable subsequence-based time series representation, allowing versatile use of learned representations in temporal prediction, smoothing, and classification tasks. The targeted interpretable representation could convey the spectral information of the modelled time series to the level of human comprehension. Through a proof-of-concept evaluation study, we demonstrate the empirical superiority of learned task-agnostic and interpretable representation over task-specific and conventional subsequence-based representation, such as symbolic and recurrent learning-based representation, in solving temporal prediction, smoothing, and classification tasks. These learned task-agnostic representations can also reveal the ground-truth periodicity of the modelled time series. We further propose two applications of our unified local predictive model in functional magnetic resonance imaging (fMRI) analysis to reveal the spectral characterisation of cortical areas at rest and reconstruct more smoothed temporal dynamics of cortical activations in both resting-state and task-evoked fMRI data, giving rise to robust decoding.

Cross-subject aesthetic preference recognition of Chinese dance posture using EEG

Due to the differences in knowledge, experience, background, and social influence, people have subjective characteristics in the process of dance aesthetic cognition. To explore the neural mechanism of the human brain in the process of dance aesthetic preference, and to find a more objective determining criterion for dance aesthetic preference, this paper constructs a cross-subject aesthetic preference recognition model of Chinese dance posture. Specifically, Dai nationality dance (a classic Chinese folk dance) was used to design dance posture materials, and an experimental paradigm for aesthetic preference of Chinese dance posture was built. Then, 91 subjects were recruited for the experiment, and their EEG signals were collected. Finally, the transfer learning method and convolutional neural networks were used to identify the aesthetic preference of the EEG signals. Experimental results have shown the feasibility of the proposed model, and the objective aesthetic measurement in dance appreciation has been implemented. Based on the classification model, the accuracy of aesthetic preference recognition is 79.74%. Moreover, the recognition accuracies of different brain regions, different hemispheres, and different model parameters were also verified by the ablation study. Additionally, the experimental results reflected the following two facts: (1) in the visual aesthetic processing of Chinese dance posture, the occipital and frontal lobes are more activated and participate in dance aesthetic preference; (2) the right brain is more involved in the visual aesthetic processing of Chinese dance posture, which is consistent with the common knowledge that the right brain is responsible for processing artistic activities.

Using evolution rule in complex time series comparison

Complex time series appear in numerous applications and are related to some essential physiological and natural systems. Their comparison faces big challenges: 1) with different complexity; 2) with significant phase shift in one series or shift∖on the time axis. Existing methods work well for periodic time-series data, but fail to produce satisfactory results in complex time-series. In this paper, we introduce a novel distance function based on the evolution rule for complex time series comparison. Here, the evolution rule, as the innate generative mechanism of time series, is creatively used to characterize complicated dynamics from complex time series. The comparison includes different level comparisons: the coarse level is to compare the difference in complexity, and the fine level is to compare the difference in actual evolution behavior. The proposed method is inspired by the observation that similar sequences come from the same source, e.g. a person’s heart, in the case of ECG, thus two similar series will have the same innate generative mechanism. The performance has been verified by the conducting experiments, and the experiment results show that the proposed method is superior to the previously existing methods in clustering and classification on some real data sets.

A Privacy-Preserving Desk Sensor for Monitoring Healthy Movement Breaks in Smart Office Environments with the Internet of Things

Smart workplace Internet of Things (IoT) solutions rely on several sensors deployed efficiently in the workplace environment to collect accurate data to meet system goals. A vital issue for these sensor-based IoT solutions is privacy. Ideally, the occupants must be monitored discreetly, and the strategies for maintaining privacy are dependent on the nature of the data required. This paper proposes a new sensor design approach for IoT solutions in the workplace that protects occupants' privacy. We focus on a novel sensor that autonomously detects and captures human movements in the office to monitor a person's sedentary behavior. The sensor guides an eHealth solution that uses continuous feedback about desk behaviors to prompt healthy movement breaks for seated workers. The proposed sensor and its privacy-preserving characteristics can enhance the eHealth solution system's performance. Compared to self-reporting, intrusive, and other data collection techniques, this sensor can collect the information reliably and timely. We also present the data analysis specific to this new sensor that measures two physical distance parameters in real-time and uses their difference to determine human actions. This architecture aims to collect precise data at the sensor design level rather than to protect privacy during the data analysis phase.

ClaSP: parameter-free time series segmentation

The study of natural and human-made processes often results in long sequences of temporally-ordered values, aka time series (TS). Such processes often consist of multiple states, e.g. operating modes of a machine, such that state changes in the observed processes result in changes in the distribution of shape of the measured values. Time series segmentation (TSS) tries to find such changes in TS post-hoc to deduce changes in the data-generating process. TSS is typically approached as an unsupervised learning problem aiming at the identification of segments distinguishable by some statistical property. Current algorithms for TSS require domain-dependent hyper-parameters to be set by the user, make assumptions about the TS value distribution or the types of detectable changes which limits their applicability. Common hyper-parameters are the measure of segment homogeneity and the number of change points, which are particularly hard to tune for each data set. We present ClaSP, a novel, highly accurate, hyper-parameter-free and domain-agnostic method for TSS. ClaSP hierarchically splits a TS into two parts. A change point is determined by training a binary TS classifier for each possible split point and selecting the one split that is best at identifying subsequences to be from either of the partitions. ClaSP learns its main two model-parameters from the data using two novel bespoke algorithms. In our experimental evaluation using a benchmark of 107 data sets, we show that ClaSP outperforms the state of the art in terms of accuracy and is fast and scalable. Furthermore, we highlight properties of ClaSP using several real-world case studies.

Elastic similarity and distance measures for multivariate time series

This paper contributes multivariate versions of seven commonly used elastic similarity and distance measures for time series data analytics. Elastic similarity and distance measures can compensate for misalignments in the time axis of time series data. We adapt two existing strategies used in a multivariate version of the well-known Dynamic Time Warping (DTW), namely, Independent and Dependent DTW, to these seven measures. While these measures can be applied to various time series analysis tasks, we demonstrate their utility on multivariate time series classification using the nearest neighbor classifier. On 23 well-known datasets, we demonstrate that each of the measures but one achieves the highest accuracy relative to others on at least one dataset, supporting the value of developing a suite of multivariate similarity and distance measures. We also demonstrate that there are datasets for which either the dependent versions of all measures are more accurate than their independent counterparts or vice versa. In addition, we also construct a nearest neighbor-based ensemble of the measures and show that it is competitive to other state-of-the-art single-strategy multivariate time series classifiers.

Fuzzy Cognitive Map-Driven Comprehensive Time-Series Classification

This article presents a comprehensive approach for time-series classification. The proposed model employs a fuzzy cognitive map (FCM) as a classification engine. Preprocessed input data feed the employed FCM. Map responses, after a postprocessing procedure, are used in the calculation of the final classification decision. The time-series data are staged using the moving-window technique to capture the time flow in the training procedure. We use a backward error propagation algorithm to compute the required model hyperparameters. Four model hyperparameters require tuning. Two are crucial for the model construction: 1) FCM size (number of concepts) and 2) window size (for the moving-window technique). Other two are important for training the model: 1) the number of epochs and 2) the learning rate (for training). Two distinguishing aspects of the proposed model are worth noting: 1) the separation of the classification engine from pre- and post-processing and 2) the time flow capture for data from concept space. The proposed classifier joins the key advantage of the FCM model, which is the interpretability of the model, with the superior classification performance attributed to the specially designed pre- and postprocessing stages. This article presents the experiments performed, demonstrating that the proposed model performs well against a wide range of state-of-the-art time-series classification algorithms.

An evolutionary ensemble learning for diagnosing COVID-19 via cough signals

Objective The spread of the COVID-19 disease has caused great concern around the world and detecting the positive cases is crucial in curbing the pandemic. One of the symptoms of the disease is the dry cough it causes. It has previously been shown that cough signals can be used to identify a variety of diseases including tuberculosis, asthma, etc. In this paper, we proposed an algorithm to diagnose via cough signals the COVID-19 disease. Methods The proposed algorithm is an ensemble scheme that consists of a number of base learners, where each base learner uses a different feature extractor method, including statistical approaches and convolutional neural networks (CNN) for automatic feature extraction. Features are extracted from the raw signal and some transforms performed it, including Fourier, wavelet, Hilbert-Huang, and short-term Fourier transforms. The outputs of these base-learners are aggregated via a weighted voting scheme, with the weights optimised via an evolutionary paradigm. This paper also proposes a memetic algorithm for training the CNNs in the base-learners, which combines the speed of gradient descent (GD) algorithms and global search space coverage of the evolutionary algorithms. Results Experiments were performed on the proposed algorithm and different rival algorithms which included a number of CNN architectures in the literature and generic machine learning algorithms. The results suggested that the proposed algorithm achieves better performance compared to the existing algorithms in diagnosing COVID-19 via cough signals. Conclusion This research showed that COVID-19 could be diagnosed via cough signals and CNNs could be employed to process these signals and it may be further improved by the optimization of CNN architecture.

Combining with lab-on-chip technology and multi-organ fusion strategy to estimate post-mortem interval of rat

Background

The estimation of post-mortem interval (PMI) is one of the most important problems in forensic pathology all the time. Although many classical methods can be used to estimate time since death, accurate and rapid estimation of PMI is still a difficult task in forensic practice, so the estimation of PMI requires a faster, more accurate, and more convenient method.

Materials and methods

In this study, an experimental method, lab-on-chip, is used to analyze the characterizations of polypeptide fragments of the lung, liver, kidney, and skeletal muscle of rats at defined time points after death (0, 1, 2, 3, 5, 7, 9, 12, 15, 18, 21, 24, 27, and 30 days). Then, machine learning algorithms (base model: LR, SVM, RF, GBDT, and MLPC; ensemble model: stacking, soft voting, and soft-weighted voting) are applied to predict PMI with single organ. Multi-organ fusion strategy is designed to predict PMI based on multiple organs. Then, the ensemble pruning algorithm determines the best combination of multi-organ.

Results

The kidney is the best single organ for predicting the time of death, and its internal and external accuracy is 0.808 and 0.714, respectively. Multi-organ fusion strategy dramatically improves the performance of PMI estimation, and its internal and external accuracy is 0.962 and 0.893, respectively. Finally, the best organ combination determined by the ensemble pruning algorithm is all organs, such as lung, liver, kidney, and skeletal muscle.

Conclusion

Lab-on-chip is feasible to detect polypeptide fragments and multi-organ fusion is more accurate than single organ for PMI estimation.

Passive Fingerprinting of Same-Model Electrical Devices by Current Consumption

One possible device authentication method is based on device fingerprints, such as software- or hardware-based unique characteristics. In this paper, we propose a fingerprinting technique based on passive externally measured information, i.e., current consumption from the electrical network. The key insight is that small hardware discrepancies naturally exist even between same-electrical-circuit devices, making it feasible to identify slight variations in the consumed current under steady-state conditions. An experimental database of current consumption signals of two similar groups containing 20 same-model computer displays was collected. The resulting signals were classified using various state-of-the-art time-series classification (TSC) methods. We successfully identified 40 similar (same-model) electrical devices with about 94% precision, while most errors were concentrated in confusion between a small number of devices. A simplified empirical wavelet transform (EWT) paired with a linear discriminant analysis (LDA) classifier was shown to be the recommended classification method.

Classification of Caenorhabditis Elegans Locomotion Behaviors With Eigenfeature-Enhanced Long Short-Term Memory Networks

The free-living nematode Caenorhabditis elegans is an ideal model for understanding behavior and networks of neurons. Experimental and quantitative analyses of neural circuits and behavior have led to system-level understanding of behavioral genetics and process of transformation from sensory integration in stimulus environments to behavioral outcomes. The ability to differentiate locomotion behavior between wild-type and mutant Caenorhabditis elegans strains allows precise inference on and gaining insights into genetic and environmental influences on behaviors. This paper presents an eigenfeature-enhanced deep-learning method for classifying the dynamics of locomotion behavior of wild-type and mutant Caenorhabditis elegans. Classification results obtained from public benchmark time-series data of eigenworms illustrate the superior performance of the new method over several existing classifiers. The proposed method has potential as a useful artificial-intelligence tool for automated identification of the nematode worm behavioral patterns aiming at elucidating molecular and genetic mechanisms that control the nervous system.

When less is more powerful: Shapley value attributed ablation with augmented learning for practical time series sensor data classification

Time series sensor data classification tasks often suffer from training data scarcity issue due to the expenses associated with the expert-intervened annotation efforts. For example, Electrocardiogram (ECG) data classification for cardio-vascular disease (CVD) detection requires expensive labeling procedures with the help of cardiologists. Current state-of-the-art algorithms like deep learning models have shown outstanding performance under the general requirement of availability of large set of training examples. In this paper, we propose Shapley Attributed Ablation with Augmented Learning: ShapAAL, which demonstrates that deep learning algorithm with suitably selected subset of the seen examples or ablating the unimportant ones from the given limited training dataset can ensure consistently better classification performance under augmented training. In ShapAAL, additive perturbed training augments the input space to compensate the scarcity in training examples using Residual Network (ResNet) architecture through perturbation-induced inputs, while Shapley attribution seeks the subset from the augmented training space for better learnability with the goal of better general predictive performance, thanks to the "efficiency" and "null player" axioms of transferable utility games upon which Shapley value game is formulated. In ShapAAL, the subset of training examples that contribute positively to a supervised learning setup is derived from the notion of coalition games using Shapley values associated with each of the given inputs' contribution into the model prediction. ShapAAL is a novel push-pull deep architecture where the subset selection through Shapley value attribution pushes the model to lower dimension while augmented training augments the learning capability of the model over unseen data. We perform ablation study to provide the empirical evidence of our claim and we show that proposed ShapAAL method consistently outperforms the current baselines and state-of-the-art algorithms for time series sensor data classification tasks from publicly available UCR time series archive that includes different practical important problems like detection of CVDs from ECG data.

EVScout2.0 : Electric Vehicle Profiling Through Charging Profile

Electric Vehicles (EVs) represent a green alternative to traditional fuel-powered vehicles. To enforce their widespread use, both the technical development and the security of users shall be guaranteed. Users’ privacy represents a possible threat that impairs the adoption of EVs. In particular, recent works showed the feasibility of identifying EVs based on the current exchanged during the charging phase. In fact, while the resource negotiation phase runs over secure communication protocols, the signal exchanged during the actual charging contains features peculiar to each EV. In what is commonly known as profiling, a suitable feature extractor can associate such features to each EV.

In this paper, we propose EVScout2.0 , an extended and improved version of our previously proposed framework to profile EVs based on their charging behavior. By exploiting the current and pilot signals exchanged during the charging phase, our scheme can extract features peculiar for each EV, hence allowing their profiling. We implemented and tested EVScout2.0 over a set of real-world measurements considering over 7500 charging sessions from a total of 137 EVs. In particular, numerical results show the superiority of EVScout2.0 with respect to the previous version. EVScout2.0 can profile EVs, attaining a maximum of 0.88 for both recall and precision scores in the case of a balanced dataset. To the best of the authors’ knowledge, these results set a new benchmark for upcoming privacy research for large datasets of EVs.

Identification of robust deep neural network models of longitudinal clinical measurements

Deep learning (DL) from electronic health records holds promise for disease prediction, but systematic methods for learning from simulated longitudinal clinical measurements have yet to be reported. We compared nine DL frameworks using simulated body mass index (BMI), glucose, and systolic blood pressure trajectories, independently isolated shape and magnitude changes, and evaluated model performance across various parameters (e.g., irregularity, missingness). Overall, discrimination based on variation in shape was more challenging than magnitude. Time-series forest-convolutional neural networks (TSF-CNN) and Gramian angular field(GAF)-CNN outperformed other approaches (P < 0.05) with overall area-under-the-curve (AUCs) of 0.93 for both models, and 0.92 and 0.89 for variation in magnitude and shape with up to 50% missing data. Furthermore, in a real-world assessment, the TSF-CNN model predicted T2D with AUCs reaching 0.72 using only BMI trajectories. In conclusion, we performed an extensive evaluation of DL approaches and identified robust modeling frameworks for disease prediction based on longitudinal clinical measurements.

MultiRocket: multiple pooling operators and transformations for fast and effective time series classification

We propose MultiRocket, a fast time series classification (TSC) algorithm that achieves state-of-the-art accuracy with a tiny fraction of the time and without the complex ensembling structure of many state-of-the-art methods. MultiRocket improves on MiniRocket, one of the fastest TSC algorithms to date, by adding multiple pooling operators and transformations to improve the diversity of the features generated. In addition to processing the raw input series, MultiRocket also applies first order differences to transform the original series. Convolutions are applied to both representations, and four pooling operators are applied to the convolution outputs. When benchmarked using the University of California Riverside TSC benchmark datasets, MultiRocket is significantly more accurate than MiniRocket, and competitive with the best ranked current method in terms of accuracy, HIVE-COTE 2.0, while being orders of magnitude faster.

Attentional Gated Res2Net for Multivariate Time Series Classification

Multivariate time series classification is a critical problem in data mining with broad applications. It requires harnessing the inter-relationship of multiple variables and various ranges of temporal dependencies to assign the correct classification label of the time series. Multivariate time series may come from a wide range of sources and be used in various scenarios, bringing the classifier challenge of temporal representation learning. We propose a novel convolutional neural network architecture called Attentional Gated Res2Net for multivariate time series classification. Our model uses hierarchical residual-like connections to achieve multi-scale receptive fields and capture multi-granular temporal information. The gating mechanism enables the model to consider the relations between the feature maps extracted by receptive fields of multiple sizes for information fusion. Further, we propose two types of attention modules, channel-wise attention and block-wise attention, to better leverage the multi-granular temporal patterns. Our experimental results on 14 benchmark multivariate time-series datasets show that our model outperforms several baselines and state-of-the-art methods by a large margin. Our model outperforms the SOTA by a large margin, the classification accuracy of our model is 10.16% better than the SOTA model. Besides, we demonstrate that our model improves the performance of existing models when used as a plugin. Further, based on our experiments and analysis, we provide practical advice on applying our model to a new problem.

A Framework for Short Video Recognition Based on Motion Estimation and Feature Curves on SPD Manifolds

Given the prosperity of video media such as TikTok and YouTube, the requirement of short video recognition is becoming more and more urgent. A significant feature of short video is that there are few switches of scenes in short video, and the target (e.g., the face of the key person in the short video) often runs through the short video. This paper presents a new short video recognition algorithm framework that transforms a short video into a family of feature curves on symmetric positive definite (SPD) manifold as the basis of recognition. Thus far, no similar algorithm has been reported. The results of experiments suggest that our method performs better on three changeling databases than seven other related algorithms published in the top issues.

Near-Fall Detection in Unexpected Slips during Over-Ground Locomotion with Body-Worn Sensors among Older Adults

Slip-induced falls are a growing health concern for older adults, and near-fall events are associated with an increased risk of falling. To detect older adults at a high risk of slip-related falls, this study aimed to develop models for near-fall event detection based on accelerometry data collected by body-fixed sensors. Thirty-four healthy older adults who experienced 24 laboratory-induced slips were included. The slip outcomes were first identified as loss of balance (LOB) and no LOB (NLOB), and then the kinematic measures were compared between these two outcomes. Next, all the slip trials were split into a training set (90%) and a test set (10%) at sample level. The training set was used to train both machine learning models (n = 2) and deep learning models (n = 2), and the test set was used to evaluate the performance of each model. Our results indicated that the deep learning models showed higher accuracy for both LOB (>64%) and NLOB (>90%) classifications than the machine learning models. Among all the models, the Inception model showed the highest classification accuracy (87.5%) and the largest area under the receiver operating characteristic curve (AUC), indicating that the model is an effective method for near-fall (LOB) detection. Our approach can be helpful in identifying individuals at the risk of slip-related falls before they experience an actual fall.

Detection of Human Gait Phases Using Textile Pressure Sensors: A Low Cost and Pervasive Approach

Human gait analysis is a standard method used for detecting and diagnosing diseases associated with gait disorders. Wearable technologies, due to their low costs and high portability, are increasingly being used in gait and other medical analyses. This paper evaluates the use of low-cost homemade textile pressure sensors to recognize gait phases. Ten sensors were integrated into stretch pants, achieving an inexpensive and pervasive solution. Nevertheless, such a simple fabrication process leads to significant sensitivity variability among sensors, hindering their adoption in precision-demanding medical applications. To tackle this issue, we evaluated the textile sensors for the classification of gait phases over three machine learning algorithms for time-series signals, namely, random forest (RF), time series forest (TSF), and multi-representation sequence learner (Mr-SEQL). Training and testing signals were generated from participants wearing the sensing pants in a test run under laboratory conditions and from an inertial sensor attached to the same pants for comparison purposes. Moreover, a new annotation method to facilitate the creation of such datasets using an ordinary webcam and a pose detection model is presented, which uses predefined rules for label generation. The results show that textile sensors successfully detect the gait phases with an average precision of 91.2% and 90.5% for RF and TSF, respectively, only 0.8% and 2.3% lower than the same values obtained from the IMU. This situation changes for Mr-SEQL, which achieved a precision of 79% for the textile sensors and 36.8% for the IMU. The overall results show the feasibility of using textile pressure sensors for human gait recognition.

The Diverse Gait Dataset: Gait Segmentation Using Inertial Sensors for Pedestrian Localization with Different Genders, Heights and Walking Speeds

Stride length estimation is one of the most crucial aspects of Pedestrian Dead Reckoning (PDR). Due to the measurement noise of inertial sensors, individual variances of pedestrians, and the uncertainty in pedestrians walking, there is a substantial error in the assessment of stride length, which causes the accumulated deviation of Pedestrian Dead Reckoning (PDR). With the help of multi-gait analysis, which decomposes strides in time and space with greater detail and accuracy, a novel and revolutionary stride estimating model or scheme could improve the performance of PDR on different users. This paper presents a diverse stride gait dataset by using inertial sensors that collect foot movement data from people of different genders, heights, and walking speeds. The dataset contains 4690 walking strides data and 19,083 gait labels. Based on the dataset, we propose a threshold-independent stride segmentation algorithm called SDATW and achieve an F-measure of 0.835. We also provide the detailed results of recognizing four gaits under different walking speeds, demonstrating the utility of our dataset for helping train stride segmentation algorithms and gait detection algorithms.

Association of abnormal-glucose tolerance during pregnancy with exposure to PM2.5 components and sources

Maternal exposure to PM_2.5 has been associated with abnormal glucose tolerance during pregnancy, but little is known about which constituents and sources are most relevant to glycemic effects. We conducted a retrospective cohort study of 1148 pregnant women to investigate associations of PM_2.5 chemical components with gestational diabetes mellitus (GDM) and impaired glucose tolerance (IGT) and to identify the most harmful sources in Heshan, China from January 2015 to July 2016. We measured PM_2.5 using filter-based method and analyzed them for 28 constituents, including carbonaceous species, water-soluble ions and metal elements. Contributions of PM_2.5 sources were assessed by positive matrix factorization (PMF). Logistic regression model was used to estimate composition-specific and source-specific effects on GDM/IGT. Random forest algorithm was applied to evaluate the relative importance of components to GDM and IGT. PM_2.5 total mass and several chemical constituents were associated with GDM and IGT across the early to mid-gestation periods, as were the PM_2.5 sources fossil fuel/oil combustion, road dust, metal smelting, construction dust, electronic waster, vehicular emissions and industrial emissions. The trimester-specific associations differed among pollutants and sources. The third and highest quartile of elemental carbon, ammonium (NH₄⁺), iron (Fe) and manganese (Mn) across gestation were consistently associated with higher odds of GDM/IGT. Maternal exposures to zinc (Zn), titanium (Ti) and vehicular emissions during the first trimester, and vanadium (V), nickel (Ni), road dust and fossil fuel/oil combustion during the second trimester were more important for GDM/IGT. This study provides important new evidence that maternal exposure to PM_2.5 components and sources is significantly related to elevated risk for abnormal glucose tolerance during pregnancy.

Time Series Classification with InceptionFCN

Deep neural networks (DNN) have proven to be efficient in computer vision and data classification with an increasing number of successful applications. Time series classification (TSC) has been one of the challenging problems in data mining in the last decade, and significant research has been proposed with various solutions, including algorithm-based approaches as well as machine and deep learning approaches. This paper focuses on combining the two well-known deep learning techniques, namely the Inception module and the Fully Convolutional Network. The proposed method proved to be more efficient than the previous state-of-the-art InceptionTime method. We tested our model on the univariate TSC benchmark (the UCR/UEA archive), which includes 85 time-series datasets, and proved that our network outperforms the InceptionTime in terms of the training time and overall accuracy on the UCR archive.