Search

International audience; This paper focuses on the problem of the hierarchical multi‐label classification of research papers, which is the task of assigning the set of relevant labels for a paper from a hierarchy, using reduced amounts of labelled training data. Specifically, we study leveraging unlabelled data, which are usually plentiful and easy to collect, in addition to the few available labelled ones in a semi‐supervised learning framework for achieving better performance results. Thus, in this paper, we propose a semi‐supervised approach for the hierarchical multi‐label classification task of research papers based on the well‐known Co‐training algorithm, which exploit content and bibliographic coupling information as two distinct papers' views. In our approach, two hierarchical multi‐label classifiers, are learnt on different views of the labelled data, and iteratively select their most confident unlabelled samples, which are further added to the labelled set. The success of our suggested Co‐training‐based approach lies in two main components. The first is the use of two suggested selection criteria (i.e., Maximum Agreement and Labels Cardinality Consistency) that enforce selecting confident unlabelled samples. The second is the appliance of an oversampling method that rebalances the labels distribution of the initial labelled set, which reduces the reinforcement of the label imbalance issue during the Co‐training learning. The proposed approach is evaluated using a collection of scientific papers extracted from the ACM digital library. Performed experiments show the effectiveness of our approach with regards to several baseline methods.

To explore whether the brain contains pattern differences in the rock–paper–scissors (RPS) imagery task, this paper attempts to classify this task using fNIRS and deep learning. In this study, we designed an RPS task with a total duration of 25 min and 40 s, and recruited 22 volunteers for the experiment. We used the fNIRS acquisition device (FOIRE-3000) to record the cerebral neural activities of these participants in the RPS task. The time series classification (TSC) algorithm was introduced into the time-domain fNIRS signal classification. Experiments show that CNN-based TSC methods can achieve 97% accuracy in RPS classification. CNN-based TSC method is suitable for the classification of fNIRS signals in RPS motor imagery tasks, and may find new application directions for the development of brain–computer interfaces (BCI).

Thermal papers are replacing the conventional form of printing and are being extensively used across the globe. This study encompasses a non-destructive approach to examine thermal papers by using ATR-FTIR spectroscopy and Video Spectral Comparator (VSC), where the former technique helps in characterizing and discriminating different samples and the latter helps in deciphering the faded prints on thermal paper. The qualitative analysis of the spectroscopic data based on peak to peak comparison and quantitative analysis using chemometrics has been done to obtain high discriminating power. Multivariate analysis using HCA gave a discriminating power of 83.82% and PCA showed a variance of 95.64%. The strength of the study is portrayed through the decipherment of artificially and naturally faded thermal papers using VSC and analyzing the effect of different storing conditions on their rate of fading.

With the arrival of the Internet and social media, at the same time as masses of humans have benefitted from the full-size reassets of records available, there was an full-size boom with inside the upward push of cyber-crimes, mainly targeted closer to women. According to a 2019 file with inside the Economics Times, India has witnessed a 457% upward push in cybercrime with inside the 5 years span among 2011 and 2016. Most speculate that that is because of effect of social media inclusive of Facebook, Instagram and Twitter on our day by day lives. While those simply assist in growing a legitimate social network, advent of consumer debts in those websites normally desires simply an email-id. A actual lifestyles man or woman can create more than one fake IDs and for this reason impostors can effortlessly be made. Unlike the actual international state of affairs in which more than one policies and guidelines are imposed to become aware of oneself in a completely unique manner (as an instance at the same time as issuing one’s passport or driver’s license), with inside the digital international of social media, admission does now no longer require this kind of checks. In this paper, we study the one-of-a-kind debts of Instagram, specifically and try and verify an account as fake or actual the use of Machine Learning strategies specifically Logistic Regression and Random Forest Algorithm.

The development of microfluidic devices for point-of-care (POC) testing has been widely applied for food quality control, environmental monitoring, and clinical diagnoses. Regarding the latter, microfluidic paper-based analytical devices (µPADs) have a great potential in developing portable and disposable platforms for fast disease-relevant biomarker detection in remote areas. In this work, k-nearest neighbors (k-NN) classifiers were applied to demonstrate the reliability of using chemical data obtained from µPADs for diagnosis of diseases like diabetes mellitus and hyperuricemia. Predictors based on glucose, lactate, and uric acid biomarkers were evaluated to reach this goal. Healthy adults were differentiated from the diagnosed patients with a perfect success rate. Metabolic disorders were suitably predicted with 95.0% overall accuracy. This model also showed sensitivities and specificities in the ranges 75–100% and 93–100%, respectively, indicating that the classification protocol is robust. The k-NN modeling proved the potential of the µPADs to be applied in telemedicine for reliable off-site diagnoses.

This paper presents a critical review of laser pyrolysis. Although this technology is almost 60 years old, in literature many researchers, both from academia and industry, are still developing and improving it. On the contrary industrial applications are struggling to take off, if not in very restricted areas, although the technology has undoubted advantages that justify future development. The aim of this work consists in analysing a representative pool of scientific papers (230) and patents (121), from the last 20 years, to have an overview about the evolution of the method and try to understand the efforts spent to improve this technology effectively in academia and in industry. This study is important to provide a complete review about the argument, still missing in the literature. The objective is to provide an overview sufficiently broad and representative in the sources and to capture all the main ways in which laser pyrolysis has been used and with what distribution. The main focuses of the study are the analyses of the functions carried out by laser technologies, the application fields, and the types of used laser (i.e. models, power and fluence). Among the main results, the study showed that the main use of laser pyrolysis is to produce nanoparticles and coatings, the main materials worked by laser pyrolysis are silicon and carbon dioxide and the main searched properties in the products of laser pyrolysis are catalysts activity and electrical conductivity. CO2 lasers are the most used and the have high versatility compared to others. In conclusion, the study showed that laser pyrolysis is a consolidated technology within its main application fields (nanoparticles and coatings) for several years. Within this context, the technology has been developed on very different sizes and processes, obtaining a very wide range of results. Finally, these results may also have stimulated new areas of experimentation that emerged mainly in recent years and which concern biomedical applications, additive manufacturing, and waste disposal. Graphical abstract

Flying insect detection, identification, and counting are the key components of agricultural pest management. Insect identification is also one of the most challenging tasks in agricultural image processing. With the aid of machine vision and machine learning, traditional (manual) identification and counting can be automated. To achieve this goal, a particular data acquisition device and an accurate insect recognition algorithm (model) is necessary. In this work, we propose a new embedded system-based insect trap with an OpenMV Cam H7 microcontroller board, which can be used anywhere in the field without any restrictions (AC power supply, WIFI coverage, human interaction, etc.). In addition, we also propose a deep learning-based insect-counting method where we offer solutions for problems such as the “lack of data” and “false insect detection”. By means of the proposed trap and insect-counting method, spraying (pest swarming) could then be accurately scheduled.

Due to the increasing size of s (s), their design and optimization phases (i.e., ) grow increasingly complex. At design time, a large design space needs to be explored to find an implementation that fulfills all specifications and then optimizes metrics like energy, area, delay, reliability, etc. At run time, a large configuration space needs to be searched to find the best set of parameters (e.g., voltage/frequency) to further optimize the system. Both spaces are infeasible for exhaustive search typically leading to heuristic optimization algorithms that find some trade-off between design quality and computational overhead. ML can build powerful models that have successfully been employed in related domains. In this survey, we categorize how () may be used and is used for design-time and run-time optimization and exploration strategies of s. A meta-study of published techniques unveils areas in that are well-explored and underexplored with, as well as trends in the employed algorithms. We present a comprehensive categorization and summary of the state of the art on for. Finally, we summarize remaining challenges and promising open research directions.

Function-as-a-Service (FaaS) has raised a growing interest in how to "tame" serverless to enable domain-specific use cases such as data-intensive applications and machine learning (ML), to name a few. Recently, several systems have been implemented for training ML models. Certainly, these research articles are significant steps in the correct direction. However, they do not completely answer the nagging question of when serverless ML training can be more cost-effective compared to traditional "serverful" computing. To help in this task, we propose MLLess, a FaaS-based ML training prototype built atop IBM Cloud Functions. To boost cost-efficiency, MLLess implements two key optimizations: a significance filter and a scale-in auto-tuner, and leverages them to specialize model training to the FaaS model. Our results certify that MLLess can be 15X faster than serverful ML systems [24] at a lower cost for ML models (such as sparse logistic regression and matrix factorization) that exhibit fast convergence.

In modern academic social network, it is very difficult for scholars to find academic papers consistent with their research direction. Time is a critical factor in paper recommendation. As time goes on, the impact of an academic paper would gradually fade. Likewise, the research interests of users may also change. Therefore, we propose a temporal perceptual neural collaborative filtering model that integrates the multi-dimensional features of papers. We conducted our experiments on the dataset from CiteULike, comparing the recommended results by using four time-decay functions and evaluating our model with multiple evaluation indicators. The satisfactory results show that our model is effective in filtering out the expired papers by considering the characteristics of papers and the changes of scholars' interests.

Automatically detecting the positions of key-points (e.g., facial key-points or finger key-points) in an image is an essential problem in many applications, such as driver's gaze detection and drowsiness detection in automated driving systems. With the recent advances of Deep Neural Networks (DNNs), Key-Points detection DNNs (KP-DNNs) have been increasingly employed for that purpose. Nevertheless, KP-DNN testing and validation have remained a challenging problem because KP-DNNs predict many independent key-points at the same time -- where each individual key-point may be critical in the targeted application -- and images can vary a great deal according to many factors. In this paper, we present an approach to automatically generate test data for KP-DNNs using many-objective search. In our experiments, focused on facial key-points detection DNNs developed for an industrial automotive application, we show that our approach can generate test suites to severely mispredict, on average, more than 93% of all key-points. In comparison, random search-based test data generation can only severely mispredict 41% of them. Many of these mispredictions, however, are not avoidable and should not therefore be considered failures. We also empirically compare state-of-the-art, many-objective search algorithms and their variants, tailored for test suite generation. Furthermore, we investigate and demonstrate how to learn specific conditions, based on image characteristics (e.g., head posture and skin color), that lead to severe mispredictions. Such conditions serve as a basis for risk analysis or DNN retraining., to appear in ISSTA 2021

This paper is review on image fusion for different technique i.e. Non-subsampled Contourlet Transform (NSCT) and Neural Network (NN). NSCT is the pyramid structure and gives the multiscale feature of the image. NSCT used to directional filtering structure and used to filter bank. Neural network is a progression of calculations that tries to perceive fundamental connections in a bunch of information through an interaction that impersonates the manner in which the human mind works. The implemented algorithm will applied to MRI image and with the help of NN to calculated accuracy. The previous algorithm was work of different technique i.e. principal component analysis, discrete wavelet transorm and curvelet transform. The implemented algorithm will implement MATLAB software and calculate mean square error and peak signal to noise ratio.

Low energy surface coatings have found wide range of applications for generating hydrophobic and superhydrophobic surfaces. Most of the studies have been related to use of a single coating material over a single substrate or using a single technique. The degree of hydrophobicity is highly dependent on fabrication processes as well as materials being coated and as such warrants a high-level study using experimental optimization leading to the evaluation of the parametric behavior of coatings and their application techniques. Also, a single platform or system which can predict the required set of parameters for generating hydrophobic surface of required nature for given substrate is of requirement. This work applies the powerful machine learning algorithms (Levenberg Marquardt using Gauss Newton and Gradient methods) to evaluate the various processes affecting the anti-wetting behavior of coated printable paper substrates with the capability to predict the most optimized method of coating and materials that may lead to a desirable surface contact angle. The major application techniques used for this study pertain to dip coating, spray coating, spin coating and inkjet printing and silane and sol–gel base coating materials.

Bacteria identification has predominantly been conducted using specific bioreceptors such as antibodies or nucleic acid sequences. This approach may be inappropriate for environmental monitoring when the user does not know the target bacterial species and for screening complex water samples with many unknown bacterial species. In this work, we investigate the supervised machine learning of the bacteria-particle aggregation pattern induced by the peptide sets identified from the biofilm-bacteria interface. Each peptide is covalently conjugated to polystyrene particles and loaded together with bacterial suspensions onto paper microfluidic chips. Each peptide interacts with bacterial species to a different extent, leading to varying sizes of particle aggregation. This aggregation changes the surface tension and viscosity of the liquid flowing through the paper pores, altering the flow velocity at different extents. A smartphone camera captures this flow velocity without being affected by ambient and environmental conditions, towards a low-cost, rapid, and field-ready assay. A collection of such flow velocity data generates a unique fingerprinting profile for each bacterial species. Support vector machine is utilized to classify the species. At optimized conditions, the training model can predict the species at 93.3% accuracy out of five bacteria: Escherichia coli, Staphylococcus aureus, Salmonella Typhimurium, Enterococcus faecium, and Pseudomonas aeruginosa. Flow rates are monitored for less than 6 s and the sample-to-answer assay time is less than 10 min. The demonstrated method can open a new way of analyzing complex biological and environmental samples in a biomimetic manner with machine learning classification.

Creatinine is a clinically significant analyte used to diagnose kidney condition. However, the literature still lacks in creatinine sensors fulfilling point-of-care testing requirements. In this paper, we have developed colorimetric paper-based creatinine sensors adhering to point-of-care testing principles. The signal readout is accomplished with a smartphone modified with 3D-printed elements and processed with a self-written application compromising computer vision algorithm for automatic detection of the colored zone. Two colorimetric methods – routinely used Jaffe method and an alternative one with 3,5-dinitrobenzoic acid were both tested and compared. Hue channel intensity from HSV color space and green channel from RGB color space was used as the analytical signal in Jaffe and 3,5-dinitrobenzoate method, respectively. For both kinds of sensors the linear range of the response covered the range significant for urinary analysis, with precision, expressed as RSD, below 5%. Limit of quantification for Jaffe method was 1.05 mmol·L−1 whereas it was 0.82 mmol·L−1 for 3,5-dinitrobenzoate method. The utility of the developed sensors to selectively quantify creatinine in undiluted urine was proved using artificial urine samples and the obtained recoveries were in the range from 70 to 129 %.

Robotics and artificial intelligence (AI) are revolutionizing all spheres of human life. From industrial processes to graphic design, the implementation of automated intelligent systems is changing how industries work. The spread of robots and AI systems has triggered academic institutions to closely examine how these technologies may affect the humanity—this is how the fields of roboethics and AI ethics have been born. The identification of ethical issues for robotics and AI and creation of ethical frameworks were the first steps to creating a regulatory environment for these technologies. In this paper, we focus on regulatory efforts in Europe and North America to create enforceable regulation for AI and robotics. We describe and compare ethical principles, policies, and regulations that have been proposed by government organizations for the design and use of robots and AI. We also discuss proposed international regulation for robotics and AI. This paper tries to highlight the need for a comprehensive, enforceable, and agile policy to ethically regulate technology today and in the future. Through reviewing existing policies, we conclude that the European Unition currently leads the way in defining roboethics and AI ethical principles and implementing them into policy. Our findings suggest that governments in Europe and North America are aware of the ethical risks that robotics and AI pose, and are engaged in policymaking to create regulatory policies for these new technologies.

Due to the highly infectious nature of the novel coronavirus (COVID‐19) disease, excessive number of patients waits in the line for chest X‐ray examination, which overloads the clinicians and radiologists and negatively affects the patient's treatment, prognosis and control of the pandemic. Now that the clinical facilities such as the intensive care units and the mechanical ventilators are very limited in the face of this highly contagious disease, it becomes quite important to classify the patients according to their severity levels. This paper presents a novel implementation of convolutional neural network (CNN) approach for COVID‐19 disease severity classification (assessment). An automated CNN model is designed and proposed to divide COVID‐19 patients into four severity classes as mild, moderate, severe, and critical with an average accuracy of 95.52% using chest X‐ray images as input. Experimental results on a sufficiently large number of chest X‐ray images demonstrate the effectiveness of CNN model produced with the proposed framework. To the best of the author's knowledge, this is the first COVID‐19 disease severity assessment study with four stages (mild vs. moderate vs. severe vs. critical) using a sufficiently large number of X‐ray images dataset and CNN whose almost all hyper‐parameters are automatically tuned by the grid search optimiser.

Atomic force microscopy (AFM) has been an important tool for nanoscale imaging and characterization with atomic and subatomic resolution. Theoretical investigations are getting highly important for the interpretation of AFM images. Researchers have used molecular simulation to examine the AFM imaging mechanism. With a recent flurry of researches applying machine learning to AFM, AFM images obtained from molecular simulation have also been used as training data. However, the simulation is incredibly time consuming. In this paper, we apply super-resolution methods, including compressed sensing and deep learning methods, to reconstruct simulated images and to reduce simulation time. Several molecular simulation energy maps under different conditions are presented to demonstrate the performance of reconstruction algorithms. Through the analysis of reconstructed results, we find that both presented algorithms could complete the reconstruction with good quality and greatly reduce simulation time. Moreover, the super-resolution methods can be used to speed up the generation of training data and vary simulation resolution for AFM machine learning.

Recently, virus diseases, such as SARS-CoV, MERS-CoV, and COVID-19, continue to emerge and pose a severe public health problem. These diseases threaten the lives of many people and cause serious social and economic losses. Recent developments in information technology (IT) and connectivity have led to the emergence of Internet of Things (IoT) and Artificial Intelligence (AI) applications in many industries. These industries, where IoT and AI together are making significant impacts, are the healthcare and the diagnosis department. In addition, by actively communicating with smart devices and various biometric sensors, it is expanding its application fields to telemedicine, healthcare, and disease prevention. Even though existing IoT and AI technologies can enhance disease detection, monitoring, and quarantine, their impact is very limited because they are not integrated or applied rapidly to the emergence of a sudden epidemic. Especially in the situation where infectious diseases are rapidly spreading, the conventional methods fail to prevent large-scale infections and block global spreads through prediction, resulting in great loss of lives. Therefore, in this paper, various sources of infection information with local limitations are collected through virus disease information collector, and AI analysis and severity matching are performed through AI broker. Finally, through the Integrated Disease Control Center, risk alerts are issued, proliferation block letters are sent, and post-response services are provided quickly. Suppose we further develop the proposed integrated virus disease control model. In that case, it will be possible to proactively detect and warn of risk factors in response to infectious diseases that are rapidly spreading worldwide and strengthen measures to prevent spreading of infection in no time.

A prediction model is an indispensable tool in business, helping to make decisions, whether in the short, medium, or long term. In this context, the implementation of machine learning techniques in time series forecasting models has a notorious relevance, as information processing and efficient and dynamic knowledge uncovering are increasingly demanded. This paper develops a model called Variable step-size evolving Participatory Learning with Kernel Recursive Least Squares, VS-ePL-KRLS, applied to the forecast of weekly prices for S500 and S10 diesel oil, at the Brazilian level, for biweekly and monthly horizons. The presented model demonstrates a better accuracy compared with analogous models in the literature, without loss of computational performance for all time series analyzed.

In the era of autonomous robots, multi-targets search methods inspired researchers to develop adapted algorithms to robot constraints, and with the rising of Swarm Intelligence (SI) approaches, Swarm Robotics (SRs) became a very popular topic. In this paper, the problem of searching for an exponentially increasing number of targets in a complex and unknown environment is addressed. Our main objective is to propose a Robotic target search strategy based on the EHO (Elephants Herding Optimization) algorithm, namely Robotic-EHO (REHO). The main additions were the collision-free path planning strategy, the velocity limitation, and the extension to the multi-target version in discrete environments. The proposed method has been the subject of many experiments, emulating the search of infected individuals by COVID-19 in a context of containment within complex and unknown random environments, as well as in the real case study of USA. The particularity of these environments is their increasing targets' number and the dynamic Containment Rate (CR) that we propose. The experimental results show that REHO reacts much better in high Containment Rate, early start search mission, and where the robots' speed is higher than the virus spread speed.

COVID-19 is an acronym for coronavirus disease 2019. Initially, it was called 2019-nCoV, and later International Committee on Taxonomy of Viruses (ICTV) termed it SARS-CoV-2. On 30th January 2020, the World Health Organization (WHO) declared it a pandemic. With an increasing number of COVID-19 cases, the available medical infrastructure is essential to detect the suspected cases. Medical imaging techniques such as Computed Tomography (CT), chest radiography can play an important role in the early screening and detection of COVID-19 cases. It is important to identify and separate the cases to stop the further spread of the virus. Artificial Intelligence can play an important role in COVID-19 detection and decreases the workload on collapsing medical infrastructure. In this paper, a deep convolutional neural network-based architecture is proposed for the COVID-19 detection using chest radiographs. The dataset used to train and test the model is available on different public repositories. Despite having the high accuracy of the model, the decision on COVID-19 should be made in consultation with the trained medical clinician.

In this paper, linear regression (LR), multi-linear regression (MLR) and polynomial regression (PR) techniques are applied to propose a model Li-MuLi-Poly. The model predicts COVID-19 deaths happening in the United States of America. The experiment was carried out on machine learning model, minimum mean square error model, and maximum likelihood ratio model. The best-fitting model was selected according to the measures of mean square error, adjusted mean square error, mean square error, root mean square error (RMSE) and maximum likelihood ratio, and the statistical t-test was used to verify the results. Data sets are analyzed, cleaned up and debated before being applied to the proposed regression model. The correlation of the selected independent parameters was determined by the heat map and the Carl Pearson correlation matrix. It was found that the accuracy of the LR model best-fits the dataset when all the independent parameters are used in modeling, however, RMSE and mean absolute error (MAE) are high as compared to PR models. The PR models of a high degree are required to best-fit the dataset when not much independent parameter is considered in modeling. However, the PR models of low degree best-fits the dataset when independent parameters from all dimensions are considered in modeling.

The paper structure of historical prints is sort of a unique fingerprint. Paper with the same origin shows similar chain line distances. As the manual measurement of chain line distances is time consuming, the automatic detection of chain lines is beneficial. We propose an end-to-end trainable deep learning method for segmentation and parameterization of chain lines in transmitted light images of German prints from the 16th Century. We trained a conditional generative adversarial network with a multitask loss for line segmentation and line parameterization. We formulated a fully differentiable pipeline for line coordinates’ estimation that consists of line segmentation, horizontal line alignment, and 2D Fourier filtering of line segments, line region proposals, and differentiable line fitting. We created a dataset of high-resolution transmitted light images of historical prints with manual line coordinate annotations. Our method shows superior qualitative and quantitative chain line detection results with high accuracy and reliability on our historical dataset in comparison to competing methods. Further, we demonstrated that our method achieves a low error of less than 0.7 mm in comparison to manually measured chain line distances.

This paper presents a new active electrode design for electroencephalogram (EEG) and electrocardiogram (ECG) sensors based on inertial measurement units to remove motion artefacts during signal acquisition. Rather than measuring motion data from a single source for the entire recording unit, inertial measurement units are attached to each individual EEG or ECG electrode to collect local movement data. This data is then used to remove the motion artefact by using normalised least mean square adaptive filtering. Results show that the proposed active electrode design can reduce motion contamination from EEG and ECG signals in chest movement and head swinging motion scenarios. However, it is found that the performance varies, necessitating the need for the algorithm to be paired with more sophisticated signal processing to identify scenarios where it is beneficial in terms of improving signal quality. The new instrumentation hardware allows data driven artefact removal to be performed, providing a new data driven approach compared to widely used blind-source separation methods, and helps enable in the wild EEG recordings to be performed.

Since the end of 2019, computed tomography (CT) images have been used as an important substitute for the time‐consuming Reverse Transcriptase polymerase chain reaction (RT‐PCR) test; a new coronavirus 2019 (COVID‐19) disease has been detected and has quickly spread through many countries across the world. Medical imaging such as computed tomography provides great potential due to growing skepticism toward the sensitivity of RT‐PCR as a screening tool. For this purpose, automated image segmentation is highly desired for a clinical decision aid and disease monitoring. However, there is limited publicly accessible COVID‐19 image knowledge, leading to the overfitting of conventional approaches. To address this issue, the present paper focuses on data augmentation techniques to create synthetic data. Further, a framework has been proposed using WoT and traditional U‐Net with EfficientNet B0 to segment the COVID Radiopedia and Medseg datasets automatically. The framework achieves an F‐score of 0.96, which is best among state‐of‐the‐art methods. The performance of the proposed framework also computed using Sensitivity, Specificity, and Dice‐coefficient, achieves 84.5%, 93.9%, and 65.0%, respectively. Finally, the proposed work is validated using three quality of service (QoS) parameters such as server latency, response time, and network latency which improves the performance by 8%, 7%, and 10%, respectively.

Background Due to the COVID-19 pandemic, the demand for remote electrocardiogram (ECG) monitoring has increased drastically in an attempt to prevent the spread of the virus and keep vulnerable individuals with less severe cases out of hospitals. Enabling clinicians to set up remote patient ECG monitoring easily and determining how to classify the ECG signals accurately so relevant alerts are sent in a timely fashion is an urgent problem to be addressed for remote patient monitoring (RPM) to be adopted widely. Hence, a new technique is required to enable routine and widespread use of RPM, as is needed due to COVID-19. Objective The primary aim of this research is to create a robust and easy-to-use solution for personalized ECG monitoring in real-world settings that is precise, easily configurable, and understandable by clinicians. Methods In this paper, we propose a Personalized Monitoring Model (PMM) for ECG data based on motif discovery. Motif discovery finds meaningful or frequently recurring patterns in patient ECG readings. The main strategy is to use motif discovery to extract a small sample of personalized motifs for each individual patient and then use these motifs to predict abnormalities in real-time readings of that patient using an artificial logical network configured by a physician. Results Our approach was tested on 30 minutes of ECG readings from 32 patients. The average diagnostic accuracy of the PMM was always above 90% and reached 100% for some parameters, compared to 80% accuracy for the Generalized Monitoring Models (GMM). Regardless of parameter settings, PMM training models were generated within 3-4 minutes, compared to 1 hour (or longer, with increasing amounts of training data) for the GMM. Conclusions Our proposed PMM almost eliminates many of the training and small sample issues associated with GMMs. It also addresses accuracy and computational cost issues of the GMM, caused by the uniqueness of heartbeats and training issues. In addition, it addresses the fact that doctors and nurses typically do not have data science training and the skills needed to configure, understand, and even trust existing black box machine learning models.

A deep learning neural network has been developed to efficiently and accurately clean the background noise in experimental lab-based X-ray diffraction images. A better spot segmentation is obtained and thus grain mapping in 3D is improved compared with the conventional method., Laboratory X-ray diffraction contrast tomography (LabDCT) is a novel imaging technique for non-destructive 3D characterization of grain structures. An accurate grain reconstruction critically relies on precise segmentation of diffraction spots in the LabDCT images. The conventional method utilizing various filters generally satisfies segmentation of sharp spots in the images, thereby serving as a standard routine, but it also very often leads to over or under segmentation of spots, especially those with low signal-to-noise ratios and/or small sizes. The standard routine also requires a fine tuning of the filtering parameters. To overcome these challenges, a deep learning neural network is presented to efficiently and accurately clean the background noise, thereby easing the spot segmentation. The deep learning network is first trained with input images, synthesized using a forward simulation model for LabDCT in combination with a generic approach to extract features of experimental backgrounds. Then, the network is applied to remove the background noise from experimental images measured under different geometrical conditions for different samples. Comparisons of both processed images and grain reconstructions show that the deep learning method outperforms the standard routine, demonstrating significantly better grain mapping.

Using the ‘deep image prior’ characteristic of generative neural networks, a post-processing method that mitigates the inter-slice crosstalk issue in multislice ptychography has been developed. Under certain scenarios, crosstalk can be suppressed effectively in slice images reconstructed using ptychographic diffraction data alone., Multislice ptychography is a high-resolution microscopy technique used to image multiple separate axial planes using a single illumination direction. However, multislice ptychography reconstructions are often degraded by crosstalk, where some features on one plane erroneously contribute to the reconstructed image of another plane. Here, the use of a modified ‘double deep image prior’ (DDIP) architecture is demonstrated in mitigating crosstalk artifacts in multislice ptychography. Utilizing the tendency of generative neural networks to produce natural images, a modified DDIP method yielded good results on experimental data. For one of the datasets, it is shown that using DDIP could remove the need of using additional experimental data, such as from X-ray fluorescence, to suppress the crosstalk. This method may help X-ray multislice ptychography work for more general experimental scenarios.

We address the problem of exposure correction of dark, blurry and noisy images captured in low-light conditions in the wild. Classical image-denoising filters work well in the frequency space but are constrained by several factors such as the correct choice of thresholds, frequency estimates etc. On the other hand, traditional deep networks are trained end-to-end in the RGB space by formulating this task as an image-translation problem. However, that is done without any explicit constraints on the inherent noise of the dark images and thus produce noisy and blurry outputs. To this end we propose a DCT/FFT based multi-scale loss function, which when combined with traditional losses, trains a network to translate the important features for visually pleasing output. Our loss function is end-to-end differentiable, scale-agnostic, and generic; i.e., it can be applied to both RAW and JPEG images in most existing frameworks without additional overhead. Using this loss function, we report significant improvements over the state-of-the-art using quantitative metrics and subjective tests., Comment: Published version 8 pages. SIViP (2021)

With the development of high-throughput sequencing technology, biological sequence data reflecting life information becomes increasingly accessible. Particularly on the background of the COVID-19 pandemic, biological sequence data play an important role in detecting diseases, analyzing the mechanism and discovering specific drugs. In recent years, pretraining models that have emerged in natural language processing have attracted widespread attention in many research fields not only to decrease training cost but also to improve performance on downstream tasks. Pretraining models are used for embedding biological sequence and extracting feature from large biological sequence corpus to comprehensively understand the biological sequence data. In this survey, we provide a broad review on pretraining models for biological sequence data. Moreover, we first introduce biological sequences and corresponding datasets, including brief description and accessible link. Subsequently, we systematically summarize popular pretraining models for biological sequences based on four categories: CNN, word2vec, LSTM and Transformer. Then, we present some applications with proposed pretraining models on downstream tasks to explain the role of pretraining models. Next, we provide a novel pretraining scheme for protein sequences and a multitask benchmark for protein pretraining models. Finally, we discuss the challenges and future directions in pretraining models for biological sequences.

Accurate detection of hate speech against politicians, policy making and political ideas is crucial to maintain democracy and free speech. Unfortunately, the amount of labelled data necessary for training models to detect hate speech are limited and domain-dependent. In this paper, we address the issue of classification of hate speech against policy makers from Twitter in Italian, producing the first resource of this type in this language. We collected and annotated 1264 tweets, examined the cases of disagreements between annotators, and performed in-domain and cross-domain hate speech classifications with different features and algorithms. We achieved a performance of ROC AUC 0.83 and analyzed the most predictive attributes, also finding the different language features in the anti-policymakers and anti-immigration domains. Finally, we visualized networks of hashtags to capture the topics used in hateful and normal tweets., to appear in SN social sciences - special issue on hate speech

Public transport has been identified as high risk as the corona-virus carrying droplets generated by the infected passengers could be distributed to other passengers. Therefore, predicting the patterns of droplet spreading in public transport environment is of primary importance. This paper puts forward a novel computational and artificial intelligence (AI) framework for fast prediction of the spread of droplets produced by a sneezing passenger in a bus. The formation of droplets of salvia is numerically modelled using a volume of fluid methodology applied to the mouth and lips of an infected person during the sneezing process. This is followed by a large eddy simulation of the resultant two phase flow in the vicinity of the person while the effects of droplet evaporation and ventilation in the bus are considered. The results are subsequently fed to an AI tool that employs deep learning to predict the distribution of droplets in the entire volume of the bus. This combined framework is two orders of magnitude faster than the pure computational approach. It is shown that the droplets with diameters less than 250 micrometers are most responsible for the transmission of the virus, as they can travel the entire length of the bus., Graphical Abstract ga1

Natural Language Processing (NLP) is one of the most captivating applications of Deep Learning. In this survey, we consider how the Data Augmentation training strategy can aid in its development. We begin with the major motifs of Data Augmentation summarized into strengthening local decision boundaries, brute force training, causality and counterfactual examples, and the distinction between meaning and form. We follow these motifs with a concrete list of augmentation frameworks that have been developed for text data. Deep Learning generally struggles with the measurement of generalization and characterization of overfitting. We highlight studies that cover how augmentations can construct test sets for generalization. NLP is at an early stage in applying Data Augmentation compared to Computer Vision. We highlight the key differences and promising ideas that have yet to be tested in NLP. For the sake of practical implementation, we describe tools that facilitate Data Augmentation such as the use of consistency regularization, controllers, and offline and online augmentation pipelines, to preview a few. Finally, we discuss interesting topics around Data Augmentation in NLP such as task-specific augmentations, the use of prior knowledge in self-supervised learning versus Data Augmentation, intersections with transfer and multi-task learning, and ideas for AI-GAs (AI-Generating Algorithms). We hope this paper inspires further research interest in Text Data Augmentation.

Hybrid organic–inorganic perovskites are one of the promising candidates for the next‐generation semiconductors due to their superlative optoelectronic properties. However, one of the limiting factors for potential applications is their chemical and structural instability in different environments. Herein, the stability of (FAPbI3)0.85(MAPbBr3)0.15 perovskite solar cell is explored in different atmospheres using impedance spectroscopy. An equivalent circuit model and distribution of relaxation times (DRTs) are used to effectively analyze impedance spectra. DRT is further analyzed via machine learning workflow based on the non‐negative matrix factorization of reconstructed relaxation time spectra. This exploration provides the interplay of charge transport dynamics and recombination processes under environment stimuli and illumination. The results reveal that in the dark, oxygen atmosphere induces an increased hole concentration with less ionic character while ionic motion is dominant under ambient air. Under 1 Sun illumination, the environment‐dependent impedance responses show a more striking effect compared with dark conditions. In this case, the increased transport resistance observed under oxygen atmosphere in equivalent circuit analysis arises due to interruption of photogenerated hole carriers. The results not only shed light on elucidating transport mechanisms of perovskite solar cells in different environments but also offer an effective interpretation of impedance responses., The stability of (FAPbI3)0.85(MAPbBr3)0.15 perovskite solar cell is explored in different atmospheres using impedance spectroscopy. An equivalent circuit model and distribution of relaxation times are used to analyze the impedance spectra supported by an unsupervised machine learning method. The transport behavior at the interface and bulk is strongly dependent on the environmental conditions particularly under 1 Sun illumination.

BackgroundWith the development of biomedicine, the number of biomedical documents has increased rapidly bringing a great challenge for researchers trying to retrieve the information they need. Information retrieval aims to meet this challenge by searching relevant documents from abundant documents based on the given query. However, sometimes the relevance of search results needs to be evaluated from multiple aspects in specific retrieval tasks, thereby increasing the difficulty of biomedical information retrieval.ObjectiveThis study aimed to find a more systematic method for retrieving relevant scientific literature for a given patient.MethodsIn the initial retrieval stage, we supplemented query terms through query expansion strategies and applied query boosting to obtain an initial ranking list of relevant documents. In the re-ranking phase, we employed a text classification model and relevance matching model to evaluate documents from different dimensions and then combined the outputs through logistic regression to re-rank all the documents from the initial ranking list.ResultsThe proposed ensemble method contributed to the improvement of biomedical retrieval performance. Compared with the existing deep learning–based methods, experimental results showed that our method achieved state-of-the-art performance on the data collection provided by the Text Retrieval Conference 2019 Precision Medicine Track.ConclusionsIn this paper, we proposed a novel ensemble method based on deep learning. As shown in the experiments, the strategies we used in the initial retrieval phase such as query expansion and query boosting are effective. The application of the text classification model and relevance matching model better captured semantic context information and improved retrieval performance.

Incidents of hydraulic or oil spills in the oceans/seas or ports occur with some regularity during the exploitation, production and transportation of petroleum products. Immediate, safe, effective and environmentally friendly measures must be adopted to reduce the impact of the oil spill on marine life. Due to the difficulty to detect and clean these areas, semi-autonomous vehicles can make a significant contribution by implementing a cooperative and coordinated response. The paper proposes a concept study of Hybrid Monitoring Detection and Cleaning System (HMDCS-UV) for a maritime region using semi-autonomous unmanned vehicles. This system is based on a cooperative decision architecture for an unmanned aerial vehicle to monitor and detect dirty zones (i.e., hydraulic spills), and clean them up using a swarm of unmanned surface vehicles. The proposed solutions were implemented in a real cloud and were evaluated using different simulation scenarios. Experimental results show that the proposed HMDCS-UV can detect and reduce the level of hydraulic pollution in maritime regions with a significant gain in terms of energy consumption.

Background Coronavirus disease 2019 (COVID-19) has become a pandemic since its first appearance in late 2019. Deaths caused by COVID-19 are still increasing day by day and early diagnosis has become crucial. Since current diagnostic methods have many disadvantages, new investigations are needed to improve the performance of diagnosis. Methods A novel method is proposed to automatically diagnose COVID-19 by using Electrocardiogram (ECG) data with deep learning for the first time. Moreover, a new and effective method called hexaxial feature mapping is proposed to represent 12-lead ECG to 2D colorful images. Gray-Level Co-Occurrence Matrix (GLCM) method is used to extract features and generate hexaxial mapping images. These generated images are then fed into a new Convolutional Neural Network (CNN) architecture to diagnose COVID-19. Results Two different classification scenarios are conducted on a publicly available paper-based ECG image dataset to reveal the diagnostic capability and performance of the proposed approach. In the first scenario, ECG data labeled as COVID-19 and No-Findings (normal) are classified to evaluate COVID-19 classification ability. According to results, the proposed approach provides encouraging COVID-19 detection performance with an accuracy of 96.20% and F1-Score of 96.30%. In the second scenario, ECG data labeled as Negative (normal, abnormal, and myocardial infarction) and Positive (COVID-19) are classified to evaluate COVID-19 diagnostic ability. The experimental results demonstrated that the proposed approach provides satisfactory COVID-19 prediction performance with an accuracy of 93.00% and F1-Score of 93.20%. Furthermore, different experimental studies are conducted to evaluate the robustness of the proposed approach. Conclusion Automatic detection of cardiovascular changes caused by COVID-19 can be possible with a deep learning framework through ECG data. This not only proves the presence of cardiovascular changes caused by COVID-19 but also reveals that ECG can potentially be used in the diagnosis of COVID-19. We believe the proposed study may provide a crucial decision-making system for healthcare professionals. Source code All source codes are made publicly available at: https://github.com/mkfzdmr/COVID-19-ECG-Classification

We present an automated method to determine oil release from lipid-based materials. Oil-release tests can provide information regarding the ability to retain oil within the structuring network of lipid-based materials. This test provides a first insight into the stability of these materials and their possible applications in food, cosmetic, and pharmaceutical products. The method presented features a simple setup comprising a camera that automatically captures images of the evolution of the oil stain released from lipid-based materials placed on a filter paper. Image postprocessing is performed with two custom-made scripts developed for the freeware application ImageJ. The scripts allow direct calculation of the oil-stain area from all images stored in a folder returning as output numerical values in a table. This method was shown to be:•inexpensive, as the employed tools and equipment are available in most laboratories both in academia and industry,•self-running, as the method automatically captures images at predefined time intervals for a certain time span,•practical, as manual-image analysis is unnecessary (200 images can be automatically analyzed in 3 min)., Graphical abstract Image, graphical abstract

Identifying a forged printed document with scanned evidence can be a challenge. Microscopic printing is showing random shape which depends on the printing source as well as printing material. This paper presents a statistical analysis of the printing patterns under a microscopic scale, analyses the effect of printing direction, printing substrate (uncoated and coated paper), and printing technology (conventional offset, waterless offset, and electrophotography). The analysis shows a negligible effect of printing direction, yet, using the shape descriptor indexes, the printing materials and technologies are distinguishable under a microscopic scale. As a result, the algorithms based on Support Vectors Machine and Random Forest are developed, with shape descriptor indexes as features, for printing source identification. Both proposed algorithms, equally, achieve a high classification accuracy rate, over 92% accuracy with complex geometric-shape patterns. Thanks to the lightweight and efficiency of the Support Vectors Machine, the study shows promising applications for real-world and potential implementation in the Internet of Things devices.

International audience; Automation throughout history has caused profound changes in employment dynamics. With the advent of the fourth industrial revolution, a new threat may affect employability, asrobots and AI-based processes can now assume tasks considered exclusive to humans. This position paperaims to motivate the study of the effects of AI and automation on employability, extending it into a collaborative network perspective. The problem is firstly observed from a historical perspective. The collaboration aspects are considered through the analysis of two casestudies. Results suggestthata latent element of collaborative networks, complexity, may have effects in terms of employability.

Falls are unusual actions that cause a significant health risk among older people. The growing percentage of people of old age requires urgent development of fall detection and prevention systems. The emerging technology focuses on developing such systems to improve quality of life, especially for the elderly. A fall prevention system tries to predict and reduce the risk of falls. In contrast, a fall detection system observes the fall and generates a help notification to minimize the consequences of falls. A plethora of technical and review papers exist in the literature with a primary focus on fall detection. Similarly, several studies are relatively old, with a focus on wearables only, and use statistical and threshold-based approaches with a high false alarm rate. Therefore, this paper presents the latest research trends in fall detection and prevention systems using Machine Learning (ML) algorithms. It uses recent studies and analyzes datasets, age groups, ML algorithms, sensors, and location. Additionally, it provides a detailed discussion of the current trends of fall detection and prevention systems with possible future directions. This overview can help researchers understand the current systems and propose new methodologies by improving the highlighted issues. ispartof: SENSORS vol:21 issue:15 ispartof: location:Switzerland status: published

Background Foodborne disease is a common threat to human health worldwide, leading to millions of deaths every year. Thus, the accurate prediction foodborne disease risk is very urgent and of great importance for public health management. Objective We aimed to design a spatial–temporal risk prediction model suitable for predicting foodborne disease risks in various regions, to provide guidance for the prevention and control of foodborne diseases. Methods We designed a novel end-to-end framework to predict foodborne disease risk by using a multigraph structural long short-term memory neural network, which can utilize an encoder–decoder to achieve multistep prediction. In particular, to capture multiple spatial correlations, we divided regions by administrative area and constructed adjacent graphs with metrics that included region proximity, historical data similarity, regional function similarity, and exposure food similarity. We also integrated an attention mechanism in both spatial and temporal dimensions, as well as external factors, to refine prediction accuracy. We validated our model with a long-term real-world foodborne disease data set, comprising data from 2015 to 2019 from multiple provinces in China. Results Our model can achieve F1 scores of 0.822, 0.679, 0.709, and 0.720 for single-month forecasts for the provinces of Beijing, Zhejiang, Shanxi and Hebei, respectively, and the highest F1 score was 20% higher than the best results of the other models. The experimental results clearly demonstrated that our approach can outperform other state-of-the-art models, with a margin. Conclusions The spatial–temporal risk prediction model can take into account the spatial–temporal characteristics of foodborne disease data and accurately determine future disease spatial–temporal risks, thereby providing support for the prevention and risk assessment of foodborne disease.

We report the aperiodic titanate Ba10Y6Ti4O27 with a room‐temperature thermal conductivity that equals the lowest reported for an oxide. The structure is characterised by discontinuous occupancy modulation of each of the sites and can be considered as a quasicrystal. The resulting localisation of lattice vibrations suppresses phonon transport of heat. This new lead material for low‐thermal‐conductivity oxides is metastable and located within a quaternary phase field that has been previously explored. Its isolation thus requires a precisely defined synthetic protocol. The necessary narrowing of the search space for experimental investigation was achieved by evaluation of titanate crystal chemistry, prediction of unexplored structural motifs that would favour synthetically accessible new compositions, and assessment of their properties with machine‐learning models., Ba10Y6Ti4O27 has a complex aperiodic structure with the lowest thermal conductivity of any first‐transition‐series oxide. It is metastable, which makes it challenging to identify. Experimental exploration was focused sufficiently by structure prediction and property‐targeted machine learning to enable the isolation of this material (see picture), whose structural motifs provide new insight into the control of heat transport.

Satellite imagery is changing the way we understand and predict economic activity in the world. Advancements in satellite hardware and low-cost rocket launches have enabled near-real-time, high-resolution images covering the entire Earth. It is too labour-intensive, time-consuming and expensive for human annotators to analyse petabytes of satellite imagery manually. Current computer vision research exploring this problem still lack accuracy and prediction speed, both significantly important metrics for latency-sensitive automatized industrial applications. Here we address both of these challenges by proposing a set of improvements to the object recognition model design, training and complexity regularisation, applicable to a range of neural networks. Furthermore, we propose a fully convolutional neural network (FCN) architecture optimised for accurate and accelerated object recognition in multispectral satellite imagery. We show that our FCN exceeds human-level performance with state-of-the-art 97.67% accuracy over multiple sensors, it is able to generalize across dispersed scenery and outperforms other proposed methods to date. Its computationally light architecture delivers a fivefold improvement in training time and a rapid prediction, essential to real-time applications. To illustrate practical model effectiveness, we analyse it in algorithmic trading environment. Additionally, we publish a proprietary annotated satellite imagery dataset for further development in this research field. Our findings can be readily implemented for other real-time applications too.

Background Traditional Chinese medicine (TCM) clinical records contain the symptoms of patients, diagnoses, and subsequent treatment of doctors. These records are important resources for research and analysis of TCM diagnosis knowledge. However, most of TCM clinical records are unstructured text. Therefore, a method to automatically extract medical entities from TCM clinical records is indispensable. Objective Training a medical entity extracting model needs a large number of annotated corpus. The cost of annotated corpus is very high and there is a lack of gold-standard data sets for supervised learning methods. Therefore, we utilized distantly supervised named entity recognition (NER) to respond to the challenge. Methods We propose a span-level distantly supervised NER approach to extract TCM medical entity. It utilizes the pretrained language model and a simple multilayer neural network as classifier to detect and classify entity. We also designed a negative sampling strategy for the span-level model. The strategy randomly selects negative samples in every epoch and filters the possible false-negative samples periodically. It reduces the bad influence from the false-negative samples. Results We compare our methods with other baseline methods to illustrate the effectiveness of our method on a gold-standard data set. The F1 score of our method is 77.34 and it remarkably outperforms the other baselines. Conclusions We developed a distantly supervised NER approach to extract medical entity from TCM clinical records. We estimated our approach on a TCM clinical record data set. Our experimental results indicate that the proposed approach achieves a better performance than other baselines.

Background Machine learning (ML) is now widely deployed in our everyday lives. Building robust ML models requires a massive amount of data for training. Traditional ML algorithms require training data centralization, which raises privacy and data governance issues. Federated learning (FL) is an approach to overcome this issue. We focused on applying FL on vertically partitioned data, in which an individual’s record is scattered among different sites. Objective The aim of this study was to perform FL on vertically partitioned data to achieve performance comparable to that of centralized models without exposing the raw data. Methods We used three different datasets (Adult income, Schwannoma, and eICU datasets) and vertically divided each dataset into different pieces. Following the vertical division of data, overcomplete autoencoder-based model training was performed for each site. Following training, each site’s data were transformed into latent data, which were aggregated for training. A tabular neural network model with categorical embedding was used for training. A centrally based model was used as a baseline model, which was compared to that of FL in terms of accuracy and area under the receiver operating characteristic curve (AUROC). Results The autoencoder-based network successfully transformed the original data into latent representations with no domain knowledge applied. These altered data were different from the original data in terms of the feature space and data distributions, indicating appropriate data security. The loss of performance was minimal when using an overcomplete autoencoder; accuracy loss was 1.2%, 8.89%, and 1.23%, and AUROC loss was 1.1%, 0%, and 1.12% in the Adult income, Schwannoma, and eICU dataset, respectively. Conclusions We proposed an autoencoder-based ML model for vertically incomplete data. Since our model is based on unsupervised learning, no domain-specific knowledge is required in individual sites. Under the circumstances where direct data sharing is not available, our approach may be a practical solution enabling both data protection and building a robust model.

Background The use of artificial intelligence has revolutionized every area of life such as business and trade, social and electronic media, education and learning, manufacturing industries, medicine and sciences, and every other sector. The new reforms and advanced technologies of artificial intelligence have enabled data analysts to transmute raw data generated by these sectors into meaningful insights for an effective decision-making process. Health care is one of the integral sectors where a large amount of data is generated daily, and making effective decisions based on these data is therefore a challenge. In this study, cases related to childbirth either by the traditional method of vaginal delivery or cesarean delivery were investigated. Cesarean delivery is performed to save both the mother and the fetus when complications related to vaginal birth arise. Objective The aim of this study was to develop reliable prediction models for a maternity care decision support system to predict the mode of delivery before childbirth. Methods This study was conducted in 2 parts for identifying the mode of childbirth: first, the existing data set was enriched and second, previous medical records about the mode of delivery were investigated using machine learning algorithms and by extracting meaningful insights from unseen cases. Several prediction models were trained to achieve this objective, such as decision tree, random forest, AdaBoostM1, bagging, and k-nearest neighbor, based on original and enriched data sets. Results The prediction models based on enriched data performed well in terms of accuracy, sensitivity, specificity, F-measure, and receiver operating characteristic curves in the outcomes. Specifically, the accuracy of k-nearest neighbor was 84.38%, that of bagging was 83.75%, that of random forest was 83.13%, that of decision tree was 81.25%, and that of AdaBoostM1 was 80.63%. Enrichment of the data set had a good impact on improving the accuracy of the prediction process, which supports maternity care practitioners in making decisions in critical cases. Conclusions Our study shows that enriching the data set improves the accuracy of the prediction process, thereby supporting maternity care practitioners in making informed decisions in critical cases. The enriched data set used in this study yields good results, but this data set can become even better if the records are increased with real clinical data.

BACKGROUND The emergence of SARS-CoV-2 in late 2019 and its subsequent spread worldwide continues to be a global health crisis. Many governments consider contact tracing of citizens through apps installed on mobile phones as a key mechanism to contain the spread of SARS-CoV-2. OBJECTIVE In this study, we sought to explore the suitability of artificial intelligence (AI)–enabled social media analyses using Facebook and Twitter to understand public perceptions of COVID-19 contact tracing apps in the United Kingdom. METHODS We extracted and analyzed over 10,000 relevant social media posts across an 8-month period, from March 1 to October 31, 2020. We used an initial filter with COVID-19–related keywords, which were predefined as part of an open Twitter-based COVID-19 dataset. We then applied a second filter using contract tracing app–related keywords and a geographical filter. We developed and utilized a hybrid, rule-based ensemble model, combining state-of-the-art lexicon rule-based and deep learning–based approaches. RESULTS Overall, we observed 76% positive and 12% negative sentiments, with the majority of negative sentiments reported in the North of England. These sentiments varied over time, likely influenced by ongoing public debates around implementing app-based contact tracing by using a centralized model where data would be shared with the health service, compared with decentralized contact-tracing technology. CONCLUSIONS Variations in sentiments corroborate with ongoing debates surrounding the information governance of health-related information. AI-enabled social media analysis of public attitudes in health care can help facilitate the implementation of effective public health campaigns.

BackgroundIn clinical diagnostic interviews, mental health professionals (MHPs) implement a care practice that involves asking open questions (eg, “What do you want from your life?” “What have you tried before to bring change in your life?”) while listening empathetically to patients. During these interviews, MHPs attempted to build a trusting human-centered relationship while collecting data necessary for professional medical and psychiatric care. Often, because of the social stigma of mental health disorders, patient discomfort in discussing their presenting problem may add additional complexities and nuances to the language they use, that is, hidden signals among noisy content. Therefore, a focused, well-formed, and elaborative summary of clinical interviews is critical to MHPs in making informed decisions by enabling a more profound exploration of a patient’s behavior, especially when it endangers life.ObjectiveThe aim of this study is to propose an unsupervised, knowledge-infused abstractive summarization (KiAS) approach that generates summaries to enable MHPs to perform a well-informed follow-up with patients to improve the existing summarization methods built on frequency heuristics by creating more informative summaries.MethodsOur approach incorporated domain knowledge from the Patient Health Questionnaire-9 lexicon into an integer linear programming framework that optimizes linguistic quality and informativeness. We used 3 baseline approaches: extractive summarization using the SumBasic algorithm, abstractive summarization using integer linear programming without the infusion of knowledge, and abstraction over extractive summarization to evaluate the performance of KiAS. The capability of KiAS on the Distress Analysis Interview Corpus-Wizard of Oz data set was demonstrated through interpretable qualitative and quantitative evaluations.ResultsKiAS generates summaries (7 sentences on average) that capture informative questions and responses exchanged during long (58 sentences on average), ambiguous, and sparse clinical diagnostic interviews. The summaries generated using KiAS improved upon the 3 baselines by 23.3%, 4.4%, 2.5%, and 2.2% for thematic overlap, Flesch Reading Ease, contextual similarity, and Jensen Shannon divergence, respectively. On the Recall-Oriented Understudy for Gisting Evaluation-2 and Recall-Oriented Understudy for Gisting Evaluation-L metrics, KiAS showed an improvement of 61% and 49%, respectively. We validated the quality of the generated summaries through visual inspection and substantial interrater agreement from MHPs.ConclusionsOur collaborator MHPs observed the potential utility and significant impact of KiAS in leveraging valuable but voluminous communications that take place outside of normally scheduled clinical appointments. This study shows promise in generating semantically relevant summaries that will help MHPs make informed decisions about patient status.