Balanced Class Research Articles

Early Detection of Pressure Ulcers: Considering the Reperfusion

ObjectivesPressure ulcers are a great handicap for those who develop one. Pressure ulcers can take a long time to heal especially if detected late. These afflictions require a lot of time from the medical personnel and thus a great amount of money. We aim here to check the impact of continuous measurement on the performance of early pressure ulcer detection algorithms. Material and methodsTo detect pressure ulcers early on we use a simulation of a human buttocks to simulate the reaction of it to pressure. This simulation considers the most recent findings about pressure ulcers. In particular, the phenomenon of muscle stiffening when pressure is applied for a long period of time, and the reperfusion phenomenon. We can then simulate pressure captors on the outside interface of the buttocks to use these measurements for detection. We then determine the best threshold on the measured pressures to create standard algorithms that we compare to novel algorithms using an optimized threshold on a calculated damage based on the pressure measurement of the last 2 hours. ResultsWe compare these different algorithms for the early detection of pressure ulcers and show the need to take the measurement variation in time for a better detection. The detection error is improved by 7.3% for balanced classes and 2.7% for a dataset with a majority of healthy buttocks. ConclusionWe showed that taking the evolution of pressure instead of only instantaneous measurement can improve the early detection of pressure ulcer.

Feature selection for classification using WGCNA and Spread Sub-Sample for an imbalanced rheumatoid arthritis RNASEQ data

An imbalanced classification problem occurs when the distribution of samples among different classes is uneven or biased. Handling small and imbalanced training datasets poses a notable challenge in machine learning, especially in domains such as bioinformatics and medical research. These challenges can result in biased models, leading to poor performance on under-represented classes and an overemphasis on specific features, failing to capture the genuine patterns present in the data. The present study proposes a feature selection approach-based on genes connectivity and a class balancing technique for building a machine leaning model using imbalanced gene expression data. Rheumatic arthritis data composed of 28 normal samples and 152 rheumatic samples was used in testing our proposed model. Through the weighted gene co-expression network analysis (WGCNA) approach, features were reduced to 601 from 27,991 original features. The reduced features were used to build machine learning classification models with imbalanced and later balanced classes using Spread Sub-Sample technique. According to our findings, two classifiers reported higher accuracy with imbalanced data as compared to the balanced data set. This is an indication that most classifiers are biased when trained using imbalanced dataset. Logistic regression returned improved accuracy of 95%. The other two machine learning algorithms used in this study were decision tree and IBK returned reduced accuracy of 81% and 91% respectively. In conclusion, feature selection and class balancing approaches are important in reducing model execution time and accuracy especially for RNASeq gene expression data.

Bootstrapping Semi-supervised Medical Image Segmentation with Anatomical-Aware Contrastive Distillation.

Contrastive learning has shown great promise over annotation scarcity problems in the context of medical image segmentation. Existing approaches typically assume a balanced class distribution for both labeled and unlabeled medical images. However, medical image data in reality is commonly imbalanced (i.e., multi-class label imbalance), which naturally yields blurry contours and usually incorrectly labels rare objects. Moreover, it remains unclear whether all negative samples are equally negative. In this work, we present ACTION, an Anatomical-aware ConTrastive dIstillatiON framework, for semi-supervised medical image segmentation. Specifically, we first develop an iterative contrastive distillation algorithm by softly labeling the negatives rather than binary supervision between positive and negative pairs. We also capture more semantically similar features from the randomly chosen negative set compared to the positives to enforce the diversity of the sampled data. Second, we raise a more important question: Can we really handle imbalanced samples to yield better performance? Hence, the key innovation in ACTION is to learn global semantic relationship across the entire dataset and local anatomical features among the neighbouring pixels with minimal additional memory footprint. During the training, we introduce anatomical contrast by actively sampling a sparse set of hard negative pixels, which can generate smoother segmentation boundaries and more accurate predictions. Extensive experiments across two benchmark datasets and different unlabeled settings show that ACTION significantly outperforms the current state-of-the-art semi-supervised methods.

Performance metrics of AI-enhanced single lead EKG maintained after entry of organised clustered data

Abstract Background Our experience in creating innovative Artificial Intelligence-guided single lead EKG methodologies for ST-Elevation Myocardial Infarction (STEMI) detection within complex EKG records has been previously validated. Purpose By expanding the intricate variables of our previously tested algorithm input, we seek to further improve our STEMI detecting tool. Methods 11,567 12-lead EKG records (10-s length, 500 Hz sample frequency) derived from the Latin America Telemedicine Infarct Network database from April 2014 to December 2019. From these records, we included the following balanced classes: angiographically confirmed and unconfirmed STEMI (divided by wall affected), branch blocks, non-specific ST-T changes, normal, and abnormal (Remaining 200+ CPT codes). Cardiologist annotations ensured precision (Ground truth). Determined classes were “STEMI” and “Not-STEMI”. A 1-D Convolutional Neural Network model was trained and tested for each lead with dataset proportions of 90/10, respectively. The last dense layer outputs a probability for each record being STEMI/Not-STEMI. The analysis also included performance metrics and false-negative reports. Results Overall, the most promising Single lead for STEMI detection was V2 (91.2% Accuracy, 89.6% Sensitivity, and 92.9% Specificity). 55% of false negatives were inferior wall STEMI (Table 1). Conclusion Appreciable progress of our new methodology compared to our previous experiences in AI-guided Single Lead for STEMI detection, especially for lead V2. By performing a thorough analysis of false-negative reports, we aspire to identify potential areas of STEMI detection weakness which will become the focus of future ventures. Funding Acknowledgement Type of funding sources: None.

CHSMOTE: Convex hull-based synthetic minority oversampling technique for alleviating the class imbalance problem

The class imbalance problem, caused by skewed class distributions, is prevalent in real-world datasets. Generally, classifiers tend to bias toward the majority class, which may cause enormous misclassification costs. SMOTE is a powerful tool for this problem, which achieves a balanced class distribution by generating synthetic minority samples on the line segment between two minority samples. However, it still has some weaknesses, including 1) oversampling less-informative minority samples, 2) generating over-dense synthetic samples, and 3) generating noise. In this paper, we propose a novel and simple convex hull-based SMOTE (CHSMOTE) algorithm to overcome the three weaknesses of SMOTE simultaneously and further alleviate the class imbalance problem. Specifically, CHSMOTE selects the border minority samples as the initial minority samples for oversampling and identifies a sample synthesis area based on the convex hull by checking whether constructed convex hulls contain border majority samples. CHSMOTE can generate more samples with effective information by properly enlarging the generation range of synthetic samples. Extensive experiment results demonstrate the effectiveness and superiority of the proposed algorithm for alleviating the class imbalance problem.

TIME-USE AND WELL-BEING IN FAMILY AND OTHER UNPAID CAREGIVERS OF OLDER ADULTS

Abstract Due to the intensive time commitment for caregiving, caregivers report limited freedom to engage with others, participate in physical activities, pursue leisure activities, and have adequate time for sleep. Few studies have focused on caregivers’ time-use across different activities, particularly how different patterns of time-use are associated with well-being. This study aimed to: (1) identify time-use profiles of family caregivers of older adults and (2) examine associations between identified time-use profiles and caregiver well-being. We analyzed 1,640 family caregivers of community-dwelling older adults by combining secondary data from Round 7 (2017) of the National Study of Caregiving and the National Health and Aging Trends Study. We conducted latent profile analysis to estimate time-use profiles including covariates and outcomes. Three classes of caregivers emerged based on time-use patterns. The High Committed class (20%) spent the longest time in non-eldercare related committed activities, such as household activities and paid work. The High Discretionary class (49%) spent the highest amount of discretionary time, including social activities, physical activities, and other free-time activities. They also spent the least amount of non-eldercare committed time compared to the other two caregiver types. Lastly, the Balanced class (31%) allocated time relatively evenly in all activities. When comparing well-being outcomes between time-use profiles, caregivers in the High Discretionary class had worse self-rated health but lower levels of anxiety than the Balanced class. Research on time-use and caregiver well-being may help identify at-risk caregiver groups based on lifestyle profiles and develop targeted policies to promote better caregiver well-being.

Predicting Endocrine Disruption Using Conformal Prediction - APrioritization Strategy to Identify Hazardous Chemicals with Confidence.

Receptor-mediated molecular initiating events (MIEs) and their relevance in endocrine activity (EA) have been highlighted in literature. More than 15 receptors have been associated with neurodevelopmental adversity and metabolic disruption. MIEs describe chemical interactions with defined biological outcomes, a relationship that could be described with quantitative structure-activity relationship (QSAR) models. QSAR uncertainty can be assessed using the conformal prediction (CP) framework, which provides similarity (i.e., nonconformity) scores relative to the defined classes per prediction. CP calibration can indirectly mitigate data imbalance during model development, and the nonconformity scores serve as intrinsic measures of chemical applicability domain assessment during screening. The focus of this work was to propose an in silico predictive strategy for EA. First, 23 QSAR models for MIEs associated with EA were developed using high-throughput data for 14 receptors. To handle the data imbalance, five protocols were compared, and CP provided the most balanced class definition. Second, the developed QSAR models were applied to a large data set (∼55,000 chemicals), comprising chemicals representative of potential risk for human exposure. Using CP, it was possible to assess the uncertainty of the screening results and identify model strengths and out of domain chemicals. Last, two clustering methods, t-distributed stochastic neighbor embedding and Tanimoto similarity, were used to identify compounds with potential EA using known endocrine disruptors as reference. The cluster overlap between methods produced 23 chemicals with suspected or demonstrated EA potential. The presented models could be utilized for first-tier screening and identification of compounds with potential biological activity across the studied MIEs.

A small generating set for the balanced superelliptic handlebody group

The balanced superelliptic handlebody group is the normalizer of the transformation group of the balanced superelliptic covering space in the handlebody group of the total space. We prove that the balanced superelliptic mapping class group is generated by four elements. To prove this, we also proved that the liftable Hilden group is generated by three elements. This generating set for the liftable Hilden group is minimal except for some hyperelliptic cases and the generating set for the balanced superelliptic mapping class group above is also minimal for several cases.

An online continual object detector on VHR remote sensing images with class imbalance

It is a great challenge for traditional offline detectors to learn from continuous data streams, remember previous tasks and adapt to new-coming tasks in dynamic environments. To meet the challenge, online continual learning has recently attracted increasing attention, while the overwhelming majority of works focus only on classification with a balanced class distribution assumption. In this paper, we propose a replay-based approach called an online continual object detector (OCOD) for very-high-resolution (VHR) remote sensing images. First, we find that rehearsal imbalance is ubiquitous, and has more important impact on experimental results than class imbalance, which is contrary to the situation of offline learning (due to the limited memory). Here, rehearsal imbalance refers to significant difference among the number of images pertaining to various classes. Second, entropy is used to measure the degree of rehearsal imbalance in the memory, and an entropy reservoir sampling (ERS) strategy is proposed to maintain rehearsal balance in the online memory. Finally, a rehearsal-balancing priority assignment network (RBPAN) is proposed to adaptively select images from the memory for a rehearsal-balancing replay procedure. The experimental results obtained on three publicly available VHR satellite images from the NWPU VHR-10, DIOR and DOTA datasets, highlight the effectiveness and practicality of developed method.

A unique deep-learning-based model with chest x-ray image for diagnosing COVID-19

Later innovative advancements cleared the way for deep learning-based methods to be used in the therapeutic field due to its exactness for the detection and localization of different illnesses. Recently, the coronavirus widespread has put a parcel of weight on the health framework all around the world. Reverse Transcription- Polymerase Chain Reaction test and medical envisioning are both possible and effective techniques to determine the coronavirus infection. Since coronavirus is highly infection and Reverse Transcription- Polymerase Chain Reaction is time-consuming, determination utilizing a chest X-ray to early diagnosing the infection is considered secure in different situations. A preprocessing step is done first to balance classes inside the dataset and increase the training data. A deep learning-based method is proposed in this study to determine some human lung infections and classify coronavirus from other non-coronavirus diseases accordingly. The proposed model is used for multi-class classification which is more complicated than binary classification especially in the medical image due to the inter classes' large similarity. The proposed procedure effectively classifies four classes that incorporate coronavirus, lung opacity, normal lung, and viral pneumonia with an accuracy of 97.5 %. The proposed strategy appears excellent in terms of accuracy when compared with later strategies.

Anatomical labeling of intracranial arteries with deep learning in patients with cerebrovascular disease.

Brain arteries are routinely imaged in the clinical setting by various modalities, e.g., time-of-flight magnetic resonance angiography (TOF-MRA). These imaging techniques have great potential for the diagnosis of cerebrovascular disease, disease progression, and response to treatment. Currently, however, only qualitative assessment is implemented in clinical applications, relying on visual inspection. While manual or semi-automated approaches for quantification exist, such solutions are impractical in the clinical setting as they are time-consuming, involve too many processing steps, and/or neglect image intensity information. In this study, we present a deep learning-based solution for the anatomical labeling of intracranial arteries that utilizes complete information from 3D TOF-MRA images. We adapted and trained a state-of-the-art multi-scale Unet architecture using imaging data of 242 patients with cerebrovascular disease to distinguish 24 arterial segments. The proposed model utilizes vessel-specific information as well as raw image intensity information, and can thus take tissue characteristics into account. Our method yielded a performance of 0.89 macro F1 and 0.90 balanced class accuracy (bAcc) in labeling aggregated segments and 0.80 macro F1 and 0.83 bAcc in labeling detailed arterial segments on average. In particular, a higher F1 score than 0.75 for most arteries of clinical interest for cerebrovascular disease was achieved, with higher than 0.90 F1 scores in the larger, main arteries. Due to minimal pre-processing, simple usability, and fast predictions, our method could be highly applicable in the clinical setting.

Artificial intelligence-guided, single-lead EKG may be a game-changer for symptom-to-balloon time reduction in ST-elevated myocardial infarction

Abstract Background Over decades, efforts to shave off life-saving minutes from ST-Elevated Myocardial Infarction (STEMI) care centred on reducing door-to-needle and door-to-balloon times. We firmly believe that symptom-to-balloon time should prove a better focus to this end. Challenges come with this goal as it heavily relies on a patient's perception and initiative to seek care, which we deem intelligent and wearable Artificial Intelligence (AI)-driven Single Lead EKG technologies as an attractive solution in modern-day cardiology. Purpose To provide an accurate, accessible, and cost-effective AI-driven Single Lead STEMI detection algorithm that can be embedded into wearable devices and employed in a self-administered fashion. Methods Database: EKG records from Mexico, Colombia, Argentina, and Brazil from April 2014 to December 2019. Dataset: A total of 11,567 12-lead EKG records of 10[s] length with a sampling frequency of 500 Hz, including the following balanced classes: angiographically confirmed and unconfirmed STEMI, branch blocks, non-specific ST-T abnormalities, normal and abnormal (200+ CPT codes, excluding those mentioned above). Cardiologists manually checked the label of each record to ensure precision. Pre-processing: We discard the first and last 250 samples as they may contain a standardisation pulse. The study applied a digital low pass filter of order 5 with a frequency cut-off of 35 Hz. The mean was subtracted from each Lead. Classification: The determined classes were “STEMI” (Including STEMI in different locations of the myocardium – anterior, inferior, and lateral); and “Not-STEMI” (Combination of randomly sample, branch blocks, non-specific ST-T changes, and abnormal records – 25% of each). Training and Testing: A 1-D Convolutional Neural Network was trained and tested with a dataset proportion of 90/10, respectively. A different model was trained and tested for each Lead, using the central 4,500 samples of the records. The last dense layer outputs a probability for each report of being STEMI or Not-STEMI. Lead V2 showed the best overall results. The model was further tested through the same methodology using the best Lead with a subset of the previous data, excluding the unconfirmed STEMI EKG records (Total 7,230 12-lead EKG records for Confirmed Only STEMI dataset). Performance metrics were reported for each experiment and compared. Results Combined STEMI data: Accuracy: 91.2%; Sensitivity: 89.6%; Specificity: 92.9%. Confirmed STEMI Only dataset: Accuracy: 92.4%; Sensitivity: 93.4%; Specificity: 91.4% (Figure 1). Conclusion By assiduously improving the quality of the model's input, we continue to assess our algorithm's performance and reliability for future clinical validation as a potential remote monitoring and early STEMI detection device. Funding Acknowledgement Type of funding sources: None.

Performance metrics of AI-enhanced single lead EKG maintained after entry of organised clustered data

Abstract Background Our experience in creating innovative Artificial Intelligence-guided single lead EKG methodologies for ST-Elevation Myocardial Infarction (STEMI) detection within complex EKG records has been previously validated. Purpose By expanding the intricate variables of our previously tested algorithm input, we seek to further improve our STEMI detecting tool. Methods 11,567 12-lead EKG records (10-s length, 500 Hz sample frequency) derived from the Latin America Telemedicine Infarct Network database from April 2014 to December 2019. From these records, we included the following balanced classes: angiographically confirmed and unconfirmed STEMI (divided by wall affected), branch blocks, non-specific ST-T changes, normal, and abnormal (Remaining 200+ CPT codes). Cardiologist annotations ensured precision (Ground truth). Determined classes were “STEMI” and “Not-STEMI”. A 1-D Convolutional Neural Network model was trained and tested for each lead with dataset proportions of 90/10, respectively. The last dense layer outputs a probability for each record being STEMI/Not-STEMI. The analysis also included performance metrics and false-negative reports. Results Overall, the most promising Single lead for STEMI detection was V2 (91.2% Accuracy, 89.6% Sensitivity, and 92.9% Specificity). 55% of false negatives were inferior wall STEMI (Table 1). Conclusion Appreciable progress of our new methodology compared to our previous experiences in AI-guided Single Lead for STEMI detection, especially for lead V2. By performing a thorough analysis of false-negative reports, we aspire to identify potential areas of STEMI detection weakness which will become the focus of future ventures. Funding Acknowledgement Type of funding sources: None.

Curriculum Modernization Strategy in Salaf Islamic Boarding School Based on Indonesian National Qualification Framework

This article reveals the strategy of modernizing the curriculum of the Raudhatul Ulum Islamic Boarding School Jember based on the Indonesian national qualifications framework. With their salaf characteristics, Islamic boarding schools continue to innovate continuously in adapting the curriculum. This type of research is qualitative with a case study approach, data obtained from interviews, observations and documentation, then analyzed by condensing, displaying data and drawing conclusions. The findings of the study indicate that based on the results of the study, it can be concluded that first, pesantren remain consistent in exclusively studying classical books with a class system like modern schools, except that the subject matter of the study remains in the field of religion. Second, Islamic boarding schools incorporate the national curriculum into the cottage learning system so that the practice of teaching religion and science is realized in a balanced class form. Third, pesantren is consistent in traditional spending to study the yellow books and provide opportunities for santri to gain education provided by pesantren, which has been adapted to the Indonesian national qualification framework.

Artificial intelligence to evaluate diagnosed COVID-19 chest radiographs

We present a Machine Learning algorithm based on Python which can be used to aid COVID-19 diagnosis. This algorithm employs Convolutional Neural Networks (CNN) of ResNet-18 architecture from thoracic X-ray images to build a trained dataset that enables further comparisons between common pulmonary diseases and COVID-19 diagnosed patients to classify the radiological findings as being due the COVID-19 or other pathologies. We discuss the importance of setting the right parameters related to training and what they might represent in clinical procedures. We used a dataset containing 942 COVID-19 labeled radiographs from HCPA - Hospital das Clínicas de Porto Alegre and compared it to a public dataset from NIH Clinical Center containing images of pulmonary diseases. Lastly, our trained model had an accuracy of 81.76% for the imbalanced classes and an accuracy of 46.94% for the balanced classes, when compared to other pulmonary diseases such as pneumonia, edema, mass, consolidation, and fibrosis. These results disclose the difficulty of diagnosing COVID-19 from a chest radiograph as it resembles other pulmonary illnesses and makes room for further research in this matter.

OASIS: Only Adversarial Supervision for Semantic Image Synthesis

Despite their recent successes, generative adversarial networks (GANs) for semantic image synthesis still suffer from poor image quality when trained with only adversarial supervision. Previously, additionally employing the VGG-based perceptual loss has helped to overcome this issue, significantly improving the synthesis quality, but at the same time limited the progress of GAN models for semantic image synthesis. In this work, we propose a novel, simplified GAN model, which needs only adversarial supervision to achieve high quality results. We re-design the discriminator as a semantic segmentation network, directly using the given semantic label maps as the ground truth for training. By providing stronger supervision to the discriminator as well as to the generator through spatially- and semantically-aware discriminator feedback, we are able to synthesize images of higher fidelity and with a better alignment to their input label maps, making the use of the perceptual loss superfluous. Furthermore, we enable high-quality multi-modal image synthesis through global and local sampling of a 3D noise tensor injected into the generator, which allows complete or partial image editing. We show that images synthesized by our model are more diverse and follow the color and texture distributions of real images more closely. We achieve a strong improvement in image synthesis quality over prior state-of-the-art models across the commonly used ADE20K, Cityscapes, and COCO-Stuff datasets using only adversarial supervision. In addition, we investigate semantic image synthesis under severe class imbalance and sparse annotations, which are common aspects in practical applications but were overlooked in prior works. To this end, we evaluate our model on LVIS, a dataset originally introduced for long-tailed object recognition. We thereby demonstrate high performance of our model in the sparse and unbalanced data regimes, achieved by means of the proposed 3D noise and the ability of our discriminator to balance class contributions directly in the loss function. Our code and pretrained models are available at https://github.com/boschresearch/OASIS.

A New Large Scale SVM for Classification of Imbalanced Evolving Streams

Classification from imbalanced evolving streams possesses a combined challenge of class imbalance and concept drift (CI-CD). However, the state of imbalance is dynamic, a kind of virtual concept drift. The imbalanced distributions and concept drift hinder the online learner’s performance as a combined or individual problem. A weighted hybrid online oversampling approach,”weighted online oversampling large scale support vector machine (WOOLASVM),” is proposed in this work to address this combined problem. The WOOLASVM is an SVM active learning approach with new boundary weighing strategies such as (i) dynamically oversampling the current boundary and (ii) dynamic weighing of the cost parameter of the SVM objective function. Thus at any time step, WOOLASVM maintains balanced class distributions so that the CI-CD problem does not hinder the online learner performance. Over extensive experiments on synthetic and real-world streams with the static and dynamic state of imbalance, the WOOLASVM exhibits better online classification performances than other state-of-the-art methods.

Using different training strategies for urban land-use classification based on convolutional neural networks

Urban land-use scene classification from high-resolution remote-sensing imagery at high quality and accuracy is of paramount interest for urban planning, government policy-making and urban change detection. In recent years, urban land-use classification has become an ongoing task in areas addressable primarily by remote sensing, and numerous deep learning algorithms have achieved high performance on this task. However, both dataset and methodology problems still exist in the current approaches. Previous studies have relied on limited data sources, resulting in saturated classification results, and they have difficulty achieving comprehensive classification results. The previous methods based on convolutional neural networks (CNNs) focused primarily on model architecture rather than on the hyperparameters. Therefore, to achieve more accurate classification results, in this study, we constructed a new large dataset for urban land-use scene classification. More than thirty thousand remote sensing scene images were collected to create a dataset with balanced class samples that includes both higher intra-class variations and smaller inter-class dissimilarities than do the previously available public datasets. Then, we analysed two possible strategies for exploiting the capabilities of three existing popular CNNs on our datasets: full training and fine tuning. For each strategy, three types of learning rate decay were applied: fixed, exponential and polynomial. The experimental results indicate that fine tuning tends to be the best-performing strategy, and using ResNet-V1-50 and polynomial learning rate decay achieves the best results for the urban land-use scene classification task.

Multiclass Convolutional Neural Networks for Atrial Fibrillation Classification.

Atrial fibrillation (AF) is a common supraventricular arrhythmia. Its automatic identification by standard 12-lead electrocardiography (ECG) is still challenging. Recently, deep learning provided new instruments able to mimic the diagnostic ability of clinicians but only in case of binary classification (AF vs. normal sinus rhythm-NSR). However, binary classification is far from the real scenarios, where AF has to be discriminated also from several other physiological and pathological conditions. The aim of this work is to present a new AF multiclass classifier based on a convolutional neural network (CNN), able to discriminate AF from NSR, premature atrial contraction (PAC) and premature ventricular contraction (PVC). Overall, 2796 12-lead ECG recordings were selected from the open-source "PhysioNet/Computing in Cardiology Challenge 2021" database, to construct a dataset constituted by four balanced classes, namely AF class, PAC class, PVC class, and NSR class. Each lead of each ECG recording was decomposed into spectrogram by continuous wavelet transform and saved as 2D grayscale images, used to feed a 6-layers CNN. Considering the same CNN architecture, a multiclass classifiers (all classes) and three binary classifiers (AF class, PAC class, and PVC class vs. NSR class) were created and validated by a stratified shuffle split cross-validation of 10 splits. Performance was quantified in terms of area under the curve (AUC) of the receiver operating characteristic. Multiclass classifier performance was high (AF class: 96.6%; PAC class: 95.3%; PVC class: 92.8%; NSR class: 97.4%) and preferable to binary classifiers. Thus, our CNN AF multiclass classifier proved to be an efficient tool for AF discrimination from physiological and pathological confounders. Clinical Relevance-Our CNN AF multiclass classifier proved to be suitable for AF discrimination in real scenarios.

Reward criteria impact on the performance of reinforcement learning agent for autonomous navigation

In reinforcement learning, an agent takes action at every time step (follows a policy) in an environment to maximize the expected cumulative reward. Therefore, the shaping of a reward function plays a crucial role in an agent’s learning. Designing an optimal reward function is not a trivial task. In this article, we propose a reward criterion using which we develop different reward functions. The reward criterion chosen is based on the percentage of positive and negative rewards received by an agent. This reward criteria further gives rise to three different classes, ‘Balanced Class,’ ‘Skewed Positive Class,’ and ‘Skewed Negative Class.’ We train a Deep Q-Network agent on a point-goal based navigation task using the different reward classes. We also compare the performance of the proposed classes with a benchmark class. Based on the experiments, the skewed negative class outperforms the benchmark class by achieving very less variance. On the other hand, the benchmark class converges relatively faster than the skewed negative class.