Binary Labels Research Articles

Automating avalanche detection in ground-based photographs with deep learning

Automated avalanche detection has previously relied on satellite imagery, which is typically unsuitable for real-time monitoring due to long revisit times. To address this, we propose automating avalanche detection using photographs taken from the ground. This paper introduces image classification and avalanche segmentation tasks on a publicly-released dataset of 4090 photographs annotated by experts. Using the ResNet and YOLO architectures, we achieve avalanche detection F1 scores of 94.4% per image and 65.4% per avalanche region, demonstrating the potential that this method offers for avalanche monitoring.In contrast with existing approaches, we label images by avalanche type into four distinct categories: glide, loose-snow, slab, and no avalanche. This labelling scheme provides more detail on avalanche events than binary labels and is shown to improve model F1 scores. Moreover, our models do not require a digital elevation model, simplifying application to new areas. Trained models can be used for real-time avalanche monitoring and to gather temporally continuous data for the improvement of existing avalanche forecasting models.The code and dataset are available at github.com/j-f-ox/avalanche-detection.

Ch-Oracle A Unified Statistical Framework for Churn Prediction

The User churn stands as a consequential challenge within the realm of online services, posing a substantial threat to the vitality and financial viability of such services. Traditionally, endeavors in churn prediction have transformed the issue into a binary classification task, wherein users are categorized as either churned or non-churned. More recently, a shift towards a more pragmatic approach has been witnessed in the domain of online services, wherein the focus has transitioned from predicting a binary churn label to anticipating the users' return times. This method, aligning more closely with the dynamics of real-world online services, involves the model predicting the specific time of user return at each temporal step, eschewing the simplistic churn label. Nevertheless, antecedent works within this paradigm have grappled with issues of limited generality and imposing computational complexities. This paper introduces ChOracle, an innovative oracle that prognosticates user churn by modeling user return times through the amalgamation of Temporal Point Processes and Recurrent Neural Networks. Furthermore, our approach incorporates latent variables into the proposed recurrent neural network, effectively capturing the latent user loyalty to the system. An efficient approximate variational algorithm, leveraging backpropagation through time, is developed for the purpose of learning parameters within the proposed RNN.

Lung Ultrasound Reproducibly Outperforms Computed Tomography in the Detection of Extravascular Lung Water in Patients Undergoing Haemodialysis.

Lung ultrasound (LUS) is increasingly used as an extension of physical examination, informing clinical diagnosis, and decision making. There is particular interest in the assessment of patients with pulmonary congestion and extravascular lung water, although gaps remain in the evidence base underpinning this practice as a result of the limited evaluation of its inter-rater reliability and comparison with more established radiologic tests. 30 patients undergoing haemodialysis were prospectively recruited to an observational cohort study (NCT01949402). Patients underwent standardised LUS assessment before, during and after haemodialysis; their total LUS B-line score was generated, alongside a binary label of whether appearances were consistent with an interstitial syndrome. LUS video clips were recorded and independently scored by two blinded expert clinician sonographers. Low-dose non-contrast thoracic CT, pre- and post dialysis, was used as a "gold standard" radiologic comparison. LUS detected a progressive reduction in B-line scores in almost all patients undergoing haemodialysis, correlating with the volume of fluid removed once individuals with no or minimal B-lines upon pre-dialysis examination were discounted. When comparing CT scans pre- and post dialysis, radiologic evidence of the change in fluid status was only identified in a single patient. This is the first study to demonstrate that LUS detects changes in extravascular lung water caused by changing fluid status during haemodialysis using a blinded outcome assessment and that LUS appears to be more sensitive than CT for this purpose. Further research is needed to better understand the role of LUS in this and similar patient populations, with the aim of improving clinical care and outcomes.

Synthesizing class labels for highly imbalanced credit card fraud detection data

Acquiring labeled datasets often incurs substantial costs primarily due to the requirement of expert human intervention to produce accurate and reliable class labels. In the modern data landscape, an overwhelming proportion of newly generated data is unlabeled. This paradigm is especially evident in domains such as fraud detection and datasets for credit card fraud detection. These types of data have their own difficulties associated with being highly class imbalanced, which poses its own challenges to machine learning and classification. Our research addresses these challenges by extensively evaluating a novel methodology for synthesizing class labels for highly imbalanced credit card fraud data. The methodology uses an autoencoder as its underlying learner to effectively learn from dataset features to produce an error metric for use in creating new binary class labels. The methodology aims to automatically produce new labels with minimal expert input. These class labels are then used to train supervised classifiers for fraud detection. Our empirical results show that the synthesized labels are of high enough quality to produce classifiers that significantly outperform a baseline learner comparison when using area under the precision-recall curve (AUPRC). We also present results of varying levels of positive-labeled instances and their effect on classifier performance. Results show that AUPRC performance improves as more instances are labeled positive and belong to the minority class. Our methodology thereby effectively addresses the concerns of high class imbalance in machine learning by creating new and effective class labels.

Pro-cycling team cyclist assignment for an upcoming race.

Professional bicycle racing is a popular sport that has attracted significant attention in recent years. The evolution and ubiquitous use of sensors allow cyclists to measure many metrics including power, heart rate, speed, cadence, and more in training and racing. In this paper we explore for the first time assignment of a subset of a team's cyclists to an upcoming race. We introduce RaceFit, a model that recommends, based on recent workouts and past assignments, cyclists for participation in an upcoming race. RaceFit consists of binary classifiers that are trained on pairs of a cyclist and a race, described by their relevant properties (features) such as the cyclist's demographic properties, as well as features extracted from his workout data from recent weeks; as well additional properties of the race, such as its distance, elevation gain, and more. Two main approaches are introduced in recommending on each stage in a race and aggregate from it to the race, or on the entire race. The model training is based on binary label which represent participation of cyclist in a race (or in a stage) in past events. We evaluated RaceFit rigorously on a large dataset of three pro-cycling teams' cyclists and race data achieving up to 80% precision@i. The first experiment had shown that using TP or STRAVA data performs the same. Then the best-performing parameters of the framework are using 5 weeks time window, imputation was effective, and the CatBoost classifier performed best. However, the model with any of the parameters performed always better than the baselines, in which the cyclists are assigned based on their popularity in historical data. Additionally, we present the top-ranked predictive features.

Discriminative elastic-net broad learning systems for visual classification

The broad learning system (BLS) has garnered significant attention in the realm of visual classification due to its exceptional balance between accuracy and efficiency. However, the supervision mechanism in BLS typically relies on strict binary labels, limiting the approximation freedom and failing to represent the data distribution adequately. Furthermore, the inadequacy of the guidance mechanism for the output weights hinders the precise approximation of the target spaces by the input features. To tackle these issues, in this paper, we propose two discriminative elastic-net regularized BLS models with label enhancement, achieving the following significant objectives: Firstly, the proposed label enhancement technologies can substantially augment the margins between different classes while enhancing the diversity within label spaces. Secondly, the compactness and effectiveness of the output weights can be further improved via the guidance of elastic-net regularization of singular values. Thirdly, our proposed algorithms can be efficiently optimized with the augmented Lagrangian method, whose convergence and calculation complexity can be guaranteed well with solid theoretical analysis. Extensive experiments are intended on various popular databases to compare our proposed models with numerous other state-of-the-art recognition algorithms. The numerical results indicate that our proposed algorithms can achieve the best face recognition accuracy of 99.67%, and 97.54% for object recognition. Even in challenging recognition tasks, our methods can still yield an average improvement of 0.8 percentage points.

Optimizer algorithms and convolutional neural networks for text classification

<span lang="EN-US">Lately, deep learning has improved the algorithms and the architectures of several natural language processing (NLP) tasks. In spite of that, the performance of any deep learning model is widely impacted by the used optimizer algorithm; which allows updating the model parameters, finding the optimal weights, and minimizing the value of the loss function. Thus, this paper proposes a new convolutional neural network (CNN) architecture for text classification (TC) and sentiment analysis and uses it with various optimizer algorithms in the literature. Actually, in NLP, and particularly for sentiment classification concerns, the need for more empirical experiments increases the probability of selecting the pertinent optimizer. Hence, we have evaluated various optimizers on three types of text review datasets: small, medium, and large. Thereby, we examined the optimizers regarding the data amount and we have implemented our CNN model on three different sentiment analysis datasets so as to binary label text reviews. The experimental results illustrate that the adaptive optimization algorithms Adam and root mean square propagation (RMSprop) have surpassed the other optimizers. Moreover, our best CNN model which employed the RMSprop optimizer has achieved 90.48% accuracy and surpassed the state-of-the-art CNN models for binary sentiment classification problems.</span>

Prediction of Effectiveness and Toxicities of Immune Checkpoint Inhibitors Using Real-World Patient Data.

Although immune checkpoint inhibitors (ICIs) have improved outcomes in certain patients with cancer, they can also cause life-threatening immunotoxicities. Predicting immunotoxicity risks alongside response could provide a personalized risk-benefit profile, inform therapeutic decision making, and improve clinical trial cohort selection. We aimed to build a machine learning (ML) framework using routine electronic health record (EHR) data to predict hepatitis, colitis, pneumonitis, and 1-year overall survival. Real-world EHR data of more than 2,200 patients treated with ICI through December 31, 2018, were used to develop predictive models. Using a prediction time point of ICI initiation, a 1-year prediction time window was applied to create binary labels for the four outcomes for each patient. Feature engineering involved aggregating laboratory measurements over appropriate time windows (60-365 days). Patients were randomly partitioned into training (80%) and test (20%) sets. Random forest classifiers were developed using a rigorous model development framework. The patient cohort had a median age of 63 years and was 61.8% male. Patients predominantly had melanoma (37.8%), lung cancer (27.3%), or genitourinary cancer (16.4%). They were treated with PD-1 (60.4%), PD-L1 (9.0%), and CTLA-4 (19.7%) ICIs. Our models demonstrate reasonably strong performance, with AUCs of 0.739, 0.729, 0.755, and 0.752 for the pneumonitis, hepatitis, colitis, and 1-year overall survival models, respectively. Each model relies on an outcome-specific feature set, though some features are shared among models. To our knowledge, this is the first ML solution that assesses individual ICI risk-benefit profiles based predominantly on routine structured EHR data. As such, use of our ML solution will not require additional data collection or documentation in the clinic.

Multiple-instance Learning from Triplet Comparison Bags

Multiple-instance learning (MIL) solves the problem where training instances are grouped in bags, and a binary (positive or negative) label is provided for each bag. Most of the existing MIL studies need fully labeled bags for training an effective classifier, while it could be quite hard to collect such data in many real-world scenarios, due to the high cost of data labeling process. Fortunately, unlike fully labeled data, triplet comparison data can be collected in a more accurate and human-friendly way. Therefore, in this article, we for the first time investigate MIL from only triplet comparison bags , where a triplet (X a , X b , X c ) contains the weak supervision information that bag X a is more similar to X b than to X c . To solve this problem, we propose to train a bag-level classifier by the empirical risk minimization framework and theoretically provide a generalization error bound. We also show that a convex formulation can be obtained only when specific convex binary losses such as the square loss and the double hinge loss are used. Extensive experiments validate that our proposed method significantly outperforms other baselines.

MVIndEmo: a dataset for micro video public-induced emotion prediction on social media

Distinct from the realm of perceived emotion research, induced emotion pertains to the emotional responses engendered within content consumers. This facet has garnered considerable attention and finds extensive application in the analysis of public social media. However, the advent of micro videos presents unique challenges when attempting to discern the induced emotional patterns exhibited by content consumers, owing to their free-style representation and other factors. Consequently, we have put forth two novel tasks concerning the recognition of public-induced emotion on micro videos: emotion polarity and emotion classification. Additionally, we have introduced a accessible dataset specifically tailored for the analysis of public-induced emotion on micro videos. The data corpus has been meticulously collected from Tiktok, a burgeoning social media platform renowned for its trendsetting content. To construct the dataset, we have selected eight captivating topics that elicit vibrant social discussions. In devising our label generation strategy, we have employed an automated approach characterized by the fusion of multiple expert models. This strategy incorporates a confidence measure method that relies on three distinct models for effectively aggregating user comments. To accommodate adaptable benchmark configurations, we provide both binary classification labels and probability distribution labels. The dataset encompasses a vast collection of 7,153 labeled micro videos. We have undertaken an extensive statistical analysis of the dataset to provide a comprehensive overview composition. It is our earnest aspiration that this dataset will serve as a catalyst for pioneering research avenues in the analysis of emotional patterns and the understanding of multi-modal information.

A Machine Learning Ensemble Model for the Detection of Cyberbullying

The pervasive use of social media platforms, such as Facebook, Instagram, and X, has significantly amplified our electronic interconnectedness. Moreover, these platforms are now easily accessible from any location at any given time. However, the increased popularity of social media has also led to cyberbullying.It is imperative to address the need for finding, monitoring, and mitigating cyberbullying posts on social media platforms. Motivated by this necessity, we present this paper to contribute to developing an automated system for detecting binary labels of aggressive tweets.Our study has demonstrated remarkable performance compared to previous experiments on the same dataset. We employed the stacking ensemble machine learning method, utilizing four various feature extraction techniques to optimize performance within the stacking ensemble learning framework. Combining five machine learning algorithms,Decision Trees, Random Forest, Linear Support Vector Classification, Logistic Regression, and K-Nearest Neighbors into an ensemble method, we achieved superior results compared to traditional machine learning classifier models. The stacking classifier achieved a high accuracy rate of 94.00%, outperforming traditional machine learning models and surpassing the results of prior experiments that utilized the same dataset. The outcomes of our experiments showcased an accuracy rate of 0.94% in detection tweets as aggressive or non-aggressive.

Vision Transformer-based Decision Support for Neurosurgical Intervention in Acute Traumatic Brain Injury: Automated Surgical Intervention Support Tool.

Purpose To develop an automated triage tool to predict neurosurgical intervention for patients with traumatic brain injury (TBI). Materials and Methods A provincial trauma registry was reviewed to retrospectively identify patients with TBI from 2005 to 2022 treated at a specialized Canadian trauma center. Model training, validation, and testing were performed using head CT scans with binary reference standard patient-level labels corresponding to whether the patient received neurosurgical intervention. Performance and accuracy of the model, the Automated Surgical Intervention Support Tool for TBI (ASIST-TBI), were also assessed using a held-out consecutive test set of all patients with TBI presenting to the center between March 2021 and September 2022. Results Head CT scans from 2806 patients with TBI (mean age, 57 years ± 22 [SD]; 1955 [70%] men) were acquired between 2005 and 2021 and used for training, validation, and testing. Consecutive scans from an additional 612 patients (mean age, 61 years ± 22; 443 [72%] men) were used to assess the performance of ASIST-TBI. There was accurate prediction of neurosurgical intervention with an area under the receiver operating characteristic curve (AUC) of 0.92 (95% CI: 0.88, 0.94), accuracy of 87% (491 of 562), sensitivity of 87% (196 of 225), and specificity of 88% (295 of 337) on the test dataset. Performance on the held-out test dataset remained robust with an AUC of 0.89 (95% CI: 0.85, 0.91), accuracy of 84% (517 of 612), sensitivity of 85% (199 of 235), and specificity of 84% (318 of 377). Conclusion A novel deep learning model was developed that could accurately predict the requirement for neurosurgical intervention using acute TBI CT scans. Keywords: CT, Brain/Brain Stem, Surgery, Trauma, Prognosis, Classification, Application Domain, Traumatic Brain Injury, Triage, Machine Learning, Decision Support Supplemental material is available for this article. © RSNA, 2024 See also commentary by Haller in this issue.

SPREd: a simulation-supervised neural network tool for gene regulatory network reconstruction.

Reconstruction of gene regulatory networks (GRNs) from expression data is a significant open problem. Common approaches train a machine learning (ML) model to predict a gene's expression using transcription factors' (TFs') expression as features and designate important features/TFs as regulators of the gene. Here, we present an entirely different paradigm, where GRN edges are directly predicted by the ML model. The new approach, named "SPREd," is a simulation-supervised neural network for GRN inference. Its inputs comprise expression relationships (e.g. correlation, mutual information) between the target gene and each TF and between pairs of TFs. The output includes binary labels indicating whether each TF regulates the target gene. We train the neural network model using synthetic expression data generated by a biophysics-inspired simulation model that incorporates linear as well as non-linear TF-gene relationships and diverse GRN configurations. We show SPREd to outperform state-of-the-art GRN reconstruction tools GENIE3, ENNET, PORTIA, and TIGRESS on synthetic datasets with high co-expression among TFs, similar to that seen in real data. A key advantage of the new approach is its robustness to relatively small numbers of conditions (columns) in the expression matrix, which is a common problem faced by existing methods. Finally, we evaluate SPREd on real data sets in yeast that represent gold-standard benchmarks of GRN reconstruction and show it to perform significantly better than or comparably to existing methods. In addition to its high accuracy and speed, SPREd marks a first step toward incorporating biophysics principles of gene regulation into ML-based approaches to GRN reconstruction. Data and code are available from https://github.com/iiiime/SPREd.

Rethinking English in Education: The Paradoxes of Labelling the “Other”

ABSTRACT Research into how to support student teachers to work with diverse school students frequently uses a narrow, Anglocentric lens based on binary language speaker labels. This lens limits understandings of the complex factors impacting any individual’s ability to teach inclusively. Given the increases in diversity in the tertiary sector, we therefore sought to explore four teacher educators' perceptions of two inclusive literacy activities they taught that drew on their students’ rich knowledge in, and of, English to understand multilingual classrooms. An experiential checklist was employed to thematically analyse the psychological and sociocultural classroom experiences together with Bourdieu’s habitus and field theories with key findings revealing key aspects in activity design both affirmed and challenged some participants’ thinking. However, critical in disrupting deficit binaries that position us all as “others” was the need to understand how staff and students see themselves as English language speakers.

Flexible Label-Induced Manifold Broad Learning System for Multiclass Recognition.

Broad learning system (BLS), which emerges as a lightweight network paradigm, has recently attracted great attention for recognition problems due to its good balance between efficiency and accuracy. However, the supervision mechanism in BLS and its variants generally relies on the strict binary label matrix, which imposes limitations on approximation and fails to adequately align with the data distribution. To address this issue, in this article, two novel flexible label-induced BLS models with the manifold manner are proposed, whose notable characteristics are as follows. First, two proposed label relaxation strategies can both enlarge the margins between different categories and simultaneously enhance the diversity within labels. Second, the integration of manifold geometrical criterion enables the models to capture local feature structures, ensuring the obtained flexible labels align better with the similarity between samples. Third, the proposed models can be optimized efficiently with the alternating direction method of multipliers. Each iteration benefits from a closed-form solution, facilitating the optimization process. Extensive experiments and thorough theoretical analysis are intended to show the advantages of our proposed models compared to other state-of-the-art recognition algorithms.

Performance Evaluation and Comparative Analysis of Machine Learning Models on the UNSW-NB15 Dataset: A Contemporary Approach to Cyber Threat Detection

This research work utilizes the University of New South Wales Network Based 2015 (UNSW-NB15) dataset to investigate the dynamic nature of cyber threats, departing from the obsolete Knowledge Discovery and Data Mining competition 1999 (KDD Cup99) dataset. The data preparation pipeline consists of essential procedures aimed at ensuring the integrity and appropriateness of the data for analysis. The method begins by removing null values, thereafter, applying one-hot encoding to categorical features, min-max scaling for data normalization, and label encoding for efficient management of binary labels. The process of feature selection is conducted utilizing the Pearson coefficient correlation. An exhaustive evaluation is conducted on six machine learning models for the purpose of binary classification. The evaluation takes into account key performance measures like accuracy, precision, recall, and F1 score. The Random Forest model demonstrated exceptional performance, with a remarkable accuracy of 99% and a robust F1 score of 98%. Additionally, it exhibited a well-balanced precision and recall at 98%. The Support Vector Machine, Gradient Boosting, Logistic Regression, Decision Tree, and K-Nearest Neighbors models exhibit notable performance, achieving accuracy and F1 scores around at the 98% level. During our investigation into multi-class classification research, we thoroughly examined numerous machine learning models, all of which exhibited robust performance, with accuracy rates ranging from 97% to 98%. The aforementioned results highlight the efficacy of these models in accurately classifying data, regularly achieving high levels of precision, recall, and F1 scores for positive case predictions. This study offers a current viewpoint on the identification of cyber threats and emphasizes the appropriateness of several machine learning models in this rapidly changing field.

Discriminative Dictionary Pair Learning With Scale-Constrained Structured Representation for Image Classification.

The dictionary pair learning (DPL) model aims to design a synthesis dictionary and an analysis dictionary to accomplish the goal of rapid sample encoding. In this article, we propose a novel structured representation learning algorithm based on the DPL for image classification. It is referred to as discriminative DPL with scale-constrained structured representation (DPL-SCSR). The proposed DPL-SCSR utilizes the binary label matrix of dictionary atoms to project the representation into the corresponding label space of the training samples. By imposing a non-negative constraint, the learned representation adaptively approximates a block-diagonal structure. This innovative transformation is also capable of controlling the scale of the block-diagonal representation by enforcing the sum of within-class coefficients of each sample to 1, which means that the dictionary atoms of each class compete to represent the samples from the same class. This implies that the requirement of similarity preservation is considered from the perspective of the constraint on the sum of coefficients. More importantly, the DPL-SCSR does not need to design a classifier in the representation space as the label matrix of the dictionary can also be used as an efficient linear classifier. Finally, the DPL-SCSR imposes the l2,p -norm on the analysis dictionary to make the process of feature extraction more interpretable. The DPL-SCSR seamlessly incorporates the scale-constrained structured representation learning, within-class similarity preservation of representation, and the linear classifier into one regularization term, which dramatically reduces the complexity of training and parameter tuning. The experimental results on several popular image classification datasets show that our DPL-SCSR can deliver superior performance compared with the state-of-the-art (SOTA) dictionary learning methods. The MATLAB code of this article is available at https://github.com/chenzhe207/DPL-SCSR.

Factors affecting the labelling accuracy of brain MRI studies relevant for deep learning abnormality detection.

Unlocking the vast potential of deep learning-based computer vision classification systems necessitates large data sets for model training. Natural Language Processing (NLP)-involving automation of dataset labelling-represents a potential avenue to achieve this. However, many aspects of NLP for dataset labelling remain unvalidated. Expert radiologists manually labelled over 5,000 MRI head reports in order to develop a deep learning-based neuroradiology NLP report classifier. Our results demonstrate that binary labels (normal vs. abnormal) showed high rates of accuracy, even when only two MRI sequences (T2-weighted and those based on diffusion weighted imaging) were employed as opposed to all sequences in an examination. Meanwhile, the accuracy of more specific labelling for multiple disease categories was variable and dependent on the category. Finally, resultant model performance was shown to be dependent on the expertise of the original labeller, with worse performance seen with non-expert vs. expert labellers.

High-Capacity Reversible Data Hiding in Encrypted Images Based on Pixel Prediction and QuadTree Decomposition

Over the past few years, a considerable number of researchers have shown great interest in reversible data hiding for encrypted images (RDHEI). One popular category among various RDHEI methods is the reserving room before encryption (RRBE) approach, which leverages data redundancy in the original image before encryption to create space for data hiding and to achieve high embedding rates (ERs). This paper introduces an RRBE-based RDHEI method that employs pixel prediction, quadtree decomposition, and bit plane reordering to provide high embedding capacity and error-free reversibility. Initially, the content owner predicts the error image using a prediction method, followed by mapping it to a new error image with positive pixel values and a compressed binary label map is generated for overhead pixels. Subsequently, quadtree decomposition is applied to each bit plane of the mapped prediction error image to identify homogeneous blocks, which are then reordered to create room for data embedding. After generating the encrypted image with the encryption key, the data hider employs the data hiding key to embed the data based on the auxiliary information added to each embeddable bit plane’s beginning. Finally, the receiver is able to retrieve the secret message without any error, decrypt the image, and restore it without any loss or distortion. The experimental results demonstrate that the proposed RDHEI method achieves significantly higher ERs than previous competitors, with an average ER exceeding 3.6 bpp on the BOSSbase and BOWS-2 datasets.

Abstract 18718: A Deep Learning Model for Classifying Left Ventricular Enlargement Using Transthoracic and Point-of-Care Echocardiograms

Background: Enlargement or dilation of the left ventricle (LV) is associated with increased cardiovascular morbidity and mortality, and is an established precursor of ventricular dysfunction and heart failure in almost all major cardiovascular conditions. Although there are existing artificial intelligence frameworks that make use of deep learning to detect LV enlargement, many entail LV segmentation, which can be error-prone especially for point-of-care ultrasound (POCUS) images. Goal: To develop a fully end-to-end deep learning framework that takes as input either transthoracic echocardiographic (TTE) or POCUS 2D image clips, and classifies directly (i.e., without first segmenting the LV) whether a patient has an enlarged LV. Methods: Using a sample of retrospective TTE data enriched for LV enlargement (out of 13,868 patients total, 3,351 had an enlarged LV defined using sex-specific guideline-directed criteria), we trained a classifier that takes as input apical 2-chamber and parasternal long axis videos, and provides as output a binary label denoting whether a patient has an enlarged LV. Then, we evaluated the trained model using a holdout testing dataset of retrospective TTE data (2,087 patients) as well as data from a prospective cohort of 428 patients from whom TTE and POCUS were collected simultaneously. Results: For the retrospective TTE testing dataset, the AUC was 0.851 (95% confidence interval (CI) of 0.830-871). For the prospective cohort, the model performed comparably for both TTE and POCUS video clips (TTE: AUC of 0.865, 95% CI of 0.823-0.906; POCUS: AUC of 0.840, 95% CI of 0.797-0.883). Conclusions: The current deep learning model performed comparably well for both TTE and POCUS echocardiograms. We regard these results as a proof-of-concept of the potential utility of this model in helping point-of-care clinicians make decisions regarding whether patients should receive further evaluation for LV enlargement and associated conditions.