Search

Saliency methods, which “explain” deep neural networks by producing heat maps that highlight the areas of the medical image that influence model prediction, are often presented to clinicians as an aid in diagnostic decision-making. Although many saliency methods have been proposed for medical imaging interpretation, rigorous investigation of the accuracy and reliability of these strategies is necessary before they are integrated into the clinical setting. In this work, we quantitatively evaluate seven saliency methods—including Grad-CAM, Grad-CAM++, and Integrated Gradients—across multiple neural network architectures using two evaluation metrics. We establish the first human benchmark for chest X-ray segmentation in a multilabel classification set up, and examine under what clinical conditions saliency maps might be more prone to failure in localizing important pathologies compared to a human expert benchmark. We find that (i) while Grad-CAM generally localized pathologies better than the other evaluated saliency methods, all seven performed significantly worse compared with the human benchmark; (ii) the gap in localization performance between Grad-CAM and the human benchmark was largest for pathologies that were smaller in size and had shapes that were more complex; (iii) model confidence was positively correlated with Grad-CAM localization performance. While it is difficult to know whether poor localization performance is attributable to the model or to the saliency method, our work demonstrates that several important limitations of saliency methods must be addressed before we can rely on them for deep learning explainability in medical imaging.

In tasks involving the interpretation of medical images, suitably trained machine-learning models often exceed the performance of medical experts. Yet such a high-level of performance typically requires that the models be trained with relevant datasets that have been painstakingly annotated by experts. Here we show that a self-supervised model trained on chest X-ray images that lack explicit annotations performs pathology-classification tasks with accuracies comparable to those of radiologists. On an external validation dataset of chest X-rays, the self-supervised model outperformed a fully supervised model in the detection of three pathologies (out of eight), and the performance generalized to pathologies that were not explicitly annotated for model training, to multiple image-interpretation tasks and to datasets from multiple institutions.

Improving the built environment to support walking is a popular strategy to increase urban sustainability and walkability. In the past decade alone, many US cities have implemented crosswalk visibility enhancement programs as part of road safety improvements and active transportation plans. However, there are no systematic ways of measuring and monitoring the presence of key built environment attributes that influence the safety and walkability of an area, such as marked crosswalks. Furthermore, little is known about how these attributes change over time at a national scale. In this paper, we introduce an innovative approach using a deep learning-based computer vision model on Street View images to identify changes in intersection-level marked crosswalks around more than 4,000 US transit stations over a 14-year period. We found an increase in the overall number of marked crosswalks at intersections. Furthermore, high-visibility crosswalks became more common, as they replaced existing parallel-line crosswalks. We further examine crosswalks around transit stations in New York City and San Francisco to illustrate geographic variations and compare associations with other characteristics of the built environment as reported in the Smart Location Database. Areas with increases in high-visibility crosswalks focused on high density residential areas and areas with a higher percent of zero-vehicle households. However, geographic variations exist. For example, in San Francisco, transit station areas outside downtown or major corridors (South and Southwest of the city) had the lower prevalence of marked crosswalks. This analysis confirms important gaps in crosswalk visibility that call for safety enhancements and opens the door for additional research involving these data. We conclude by discussing the limitations and future research opportunities using computer vision to automatically detect large-scale transportation infrastructure changes at a relatively low cost.

Climate change is one of the greatest challenges facing humanity, and we, as machine learning (ML) experts, may wonder how we can help. Here we describe how ML can be a powerful tool in reducing greenhouse gas emissions and helping society adapt to a changing climate. From smart grids to disaster management, we identify high impact problems where existing gaps can be filled by ML, in collaboration with other fields. Our recommendations encompass exciting research questions as well as promising business opportunities. We call on the ML community to join the global effort against climate change., ACM Computing Surveys, 55 (2), ISSN:0360-0300, ISSN:1557-7341

PURPOSE: Patients with pneumonia often present to the emergency department and require prompt diagnosis and treatment. Clinical decision support systems for the diagnosis and management of pneumonia are commonly utilized in emergency departments to improve patient care. The purpose of this study is to investigate whether a deep learning model for detecting radiographic pneumonia and pleural effusions can improve functionality of a clinical decision support system for pneumonia management (ePNa) operating in 20 emergency departments. MATERIALS AND METHODS: In this retrospective cohort study, a dataset of 7,434 prior chest radiographic studies from 6,551 emergency department patients was used to develop and validate a deep learning model to identify radiographic pneumonia, pleural effusions, and evidence of multilobar pneumonia. Model performance was evaluated against three radiologists’ adjudicated interpretation and compared to performance of the natural language processing of radiology reports used by ePNa. RESULTS: The deep learning model achieved an area under the receiver operating characteristic curve of 0.833 (95% CI 0.795, 0.868) for detecting radiographic pneumonia, 0.939 (95% CI 0.911, 0.962) for detecting pleural effusions, and 0.847 (95% CI 0.800, 0.890) for identifying multilobar pneumonia. On all three tasks, the model achieved higher agreement with the adjudicated radiologist interpretation compared to ePNa. CONCLUSIONS: A deep learning model demonstrated higher agreement with radiologists than the ePNa clinical decision support system in detecting radiographic pneumonia and related findings. Incorporating deep learning models into pneumonia clinical decision support systems could enhance diagnostic performance and improve pneumonia management.

Objective Chest pain is common, and current risk-stratification methods, requiring 12-lead electrocardiograms (ECGs) and serial biomarker assays, are static and restricted to highly resourced settings. Our objective was to predict myocardial injury using continuous single-lead ECG waveforms similar to those obtained from wearable devices and to evaluate the potential of transfer learning from labeled 12-lead ECGs to improve these predictions. Methods We studied 10 874 Emergency Department (ED) patients who received continuous ECG monitoring and troponin testing from 2020 to 2021. We defined myocardial injury as newly elevated troponin in patients with chest pain or shortness of breath. We developed deep learning models of myocardial injury using continuous lead II ECG from bedside monitors as well as conventional 12-lead ECGs from triage. We pretrained single-lead models on a pre-existing corpus of labeled 12-lead ECGs. We compared model predictions to those of ED physicians. Results A transfer learning strategy, whereby models for continuous single-lead ECGs were first pretrained on 12-lead ECGs from a separate cohort, predicted myocardial injury as accurately as models using patients’ own 12-lead ECGs: area under the receiver operating characteristic curve 0.760 (95% confidence interval [CI], 0.721–0.799) and area under the precision–recall curve 0.321 (95% CI, 0.251–0.397). Models demonstrated a high negative predictive value for myocardial injury among patients with chest pain or shortness of breath, exceeding the predictive performance of ED physicians, while attending to known stigmata of myocardial injury. Conclusions Deep learning models pretrained on labeled 12-lead ECGs can predict myocardial injury from noisy, continuous monitor data early in a patient’s presentation. The utility of continuous single-lead ECG in the risk stratification of chest pain has implications for wearable devices and preclinical settings, where external validation of the approach is needed.

Tuberculosis (TB) is the leading cause of preventable death in HIV-positive patients, and yet often remains undiagnosed and untreated. Chest x-ray is often used to assist in diagnosis, yet this presents additional challenges due to atypical radiographic presentation and radiologist shortages in regions where co-infection is most common. We developed a deep learning algorithm to diagnose TB using clinical information and chest x-ray images from 677 HIV-positive patients with suspected TB from two hospitals in South Africa. We then sought to determine whether the algorithm could assist clinicians in the diagnosis of TB in HIV-positive patients as a web-based diagnostic assistant. Use of the algorithm resulted in a modest but statistically significant improvement in clinician accuracy (p = 0.002), increasing the mean clinician accuracy from 0.60 (95% CI 0.57, 0.63) without assistance to 0.65 (95% CI 0.60, 0.70) with assistance. However, the accuracy of assisted clinicians was significantly lower (p

The development of deep learning algorithms for complex tasks in digital medicine has relied on the availability of large labeled training datasets, usually containing hundreds of thousands of examples. The purpose of this study was to develop a 3D deep learning model, AppendiXNet, to detect appendicitis, one of the most common life-threatening abdominal emergencies, using a small training dataset of less than 500 training CT exams. We explored whether pretraining the model on a large collection of natural videos would improve the performance of the model over training the model from scratch. AppendiXNet was pretrained on a large collection of YouTube videos called Kinetics, consisting of approximately 500,000 video clips and annotated for one of 600 human action classes, and then fine-tuned on a small dataset of 438 CT scans annotated for appendicitis. We found that pretraining the 3D model on natural videos significantly improved the performance of the model from an AUC of 0.724 (95% CI 0.625, 0.823) to 0.810 (95% CI 0.725, 0.895). The application of deep learning to detect abnormalities on CT examinations using video pretraining could generalize effectively to other challenging cross-sectional medical imaging tasks when training data is limited.

Time series of wetland methane fluxes measured by eddy covariance require gap-filling to estimate daily, seasonal, and annual emissions. Gap-filling methane fluxes is challenging because of high variability and complex responses to multiple drivers. To date, there is no widely established gap-filling standard for wetland methane fluxes, with regards both to the best model algorithms and predictors. This study synthesizes results of different gap-filling methods systematically applied at 17 wetland sites spanning boreal to tropical regions and including all major wetland classes and two rice paddies. Procedures are proposed for: 1) creating realistic artificial gap scenarios, 2) training and evaluating gap-filling models without overstating performance, and 3) predicting half-hourly methane fluxes and annual emissions with realistic uncertainty estimates. Performance is compared between a conventional method (marginal distribution sampling) and four machine learning algorithms. The conventional method achieved similar median performance as the machine learning models but was worse than the best machine learning models and relatively insensitive to predictor choices. Of the machine learning models, decision tree algorithms performed the best in cross-validation experiments, even with a baseline predictor set, and artificial neural networks showed comparable performance when using all predictors. Soil temperature was frequently the most important predictor whilst water table depth was important at sites with substantial water table fluctuations, highlighting the value of data on wetland soil conditions. Raw gap-filling uncertainties from the machine learning models were underestimated and we propose a method to calibrate uncertainties to observations. The python code for model development, evaluation, and uncertainty estimation is publicly available. This study outlines a modular and robust machine learning workflow and makes recommendations for, and evaluates an improved baseline of, methane gap-filling models that can be implemented in multi-site syntheses or standardized products from regional and global flux networks (e.g., FLUXNET). ISSN:0168-1923 ISSN:1873-2240

Background Respiratory virus infections are significant causes of morbidity and mortality, and may induce host metabolite alterations by infecting respiratory epithelial cells. We investigated the use of liquid chromatography quadrupole time-of-flight mass spectrometry (LC/Q-TOF) combined with machine learning for the diagnosis of influenza infection. Methods We analyzed nasopharyngeal swab samples by LC/Q-TOF to identify distinct metabolic signatures for diagnosis of acute illness. Machine learning models were performed for classification, followed by Shapley additive explanation (SHAP) analysis to analyze feature importance and for biomarker discovery. Findings A total of 236 samples were tested in the discovery phase by LC/Q-TOF, including 118 positive samples (40 influenza A 2009 H1N1, 39 influenza H3 and 39 influenza B) as well as 118 age and sex-matched negative controls with acute respiratory illness. Analysis showed an area under the receiver operating characteristic curve (AUC) of 1.00 (95% confidence interval [95% CI] 0.99, 1.00), sensitivity of 1.00 (95% CI 0.86, 1.00) and specificity of 0.96 (95% CI 0.81, 0.99). The metabolite most strongly associated with differential classification was pyroglutamic acid. Independent validation of a biomarker signature based on the top 20 differentiating ion features was performed in a prospective cohort of 96 symptomatic individuals including 48 positive samples (24 influenza A 2009 H1N1, 5 influenza H3 and 19 influenza B) and 48 negative samples. Testing performed using a clinically-applicable targeted approach, liquid chromatography triple quadrupole mass spectrometry, showed an AUC of 1.00 (95% CI 0.998, 1.00), sensitivity of 0.94 (95% CI 0.83, 0.98), and specificity of 1.00 (95% CI 0.93, 1.00). Limitations include lack of sample suitability assessment, and need to validate these findings in additional patient populations. Interpretation This metabolomic approach has potential for diagnostic applications in infectious diseases testing, including other respiratory viruses, and may eventually be adapted for point-of-care testing. Funding None.

Deep learning methods for chest X-ray interpretation typically rely on pretrained models developed for ImageNet. This paradigm assumes that better ImageNet architectures perform better on chest X-ray tasks and that ImageNet-pretrained weights provide a performance boost over random initialization. In this work, we compare the transfer performance and parameter efficiency of 16 popular convolutional architectures on a large chest X-ray dataset (CheXpert) to investigate these assumptions. First, we find no relationship between ImageNet performance and CheXpert performance for both models without pretraining and models with pretraining. Second, we find that, for models without pretraining, the choice of model family influences performance more than size within a family for medical imaging tasks. Third, we observe that ImageNet pretraining yields a statistically significant boost in performance across architectures, with a higher boost for smaller architectures. Fourth, we examine whether ImageNet architectures are unnecessarily large for CheXpert by truncating final blocks from pretrained models, and find that we can make models 3.25x more parameter-efficient on average without a statistically significant drop in performance. Our work contributes new experimental evidence about the relation of ImageNet to chest x-ray interpretation performance.

Recent advances in training deep learning models have demonstrated the potential to provide accurate chest X-ray interpretation and increase access to radiology expertise. However, poor generalization due to data distribution shifts in clinical settings is a key barrier to implementation. In this study, we measured the diagnostic performance for 8 different chest X-ray models when applied to (1) smartphone photos of chest X-rays and (2) external datasets without any finetuning. All models were developed by different groups and submitted to the CheXpert challenge, and re-applied to test datasets without further tuning. We found that (1) on photos of chest X-rays, all 8 models experienced a statistically significant drop in task performance, but only 3 performed significantly worse than radiologists on average, and (2) on the external set, none of the models performed statistically significantly worse than radiologists, and five models performed statistically significantly better than radiologists. Our results demonstrate that some chest X-ray models, under clinically relevant distribution shifts, were comparable to radiologists while other models were not. Future work should investigate aspects of model training procedures and dataset collection that influence generalization in the presence of data distribution shifts., Accepted to ACM Conference on Health, Inference, and Learning (ACM-CHIL) 2021. arXiv admin note: substantial text overlap with arXiv:2011.06129

Structured AbstractImportanceMachine learning (ML) models for allocating readmission-mitigating interventions are typically selected according to their discriminative ability, which may not necessarily translate into utility in allocation of resources.ObjectiveTo determine whether ML models for allocating readmission-mitigating interventions are ranked differently based on their overall utility and their discriminative ability.DesignA retrospective analysis of ML models using claims data acquired from the Optum Clinformatics Data Mart.SettingHealth plan claims from all 50 states for commercially-insured individuals.Participants513,495 patients who were admitted as inpatients over the period January 2016 through January 2017.Main Outcomes and MeasuresMaximum utility achieved by three machine learning models for allocating readmission-mitigating interventions, determined using cost accrued in the 90 days post-discharge of an index admission and estimated counterfactual cost. Data were analyzed between April 2019 and March 2020.ResultsThe study sample consisted of 513,495 patients (mean [SD] age 69 [19] years; 294,895 [57%] Female) mean 90 day cost of $11,552 for the study period. Allocating readmission-mitigating interventions based on a LightGBM model trained to predict readmissions achieved a maximum utility of $-12,645 per patient, and an AUC of 0.74 (95% CI 0.74, 0.75); allocating interventions based on a model trained to predict cost as a proxy achieved a higher maximum utility of $-12,472 per patient, and an AUC of 0.63 (95% CI 0.62, 0.63). A hybrid model combining both intervention strategies achieved a maximum utility of $-12,472, and an AUC of 0.71 (95% CI 0.71, 0.71), comparable with the best models on either metric.Conclusion and RelevanceWe demonstrate that machine learning models may be ranked differently based on overall utility and discriminative ability. Machine learning models for allocation of limited health resources should consider directly optimizing for utility.Key pointsQuestionDo machine learning models for allocating readmission-mitigating interventions rank differently based on overall utility and discriminative performance?FindingA machine learning model predicting a patient’s future cost of care was able to achieve higher utility than a readmission risk prediction model, even though it had a lower discriminative performance in predicting readmissions.MeaningOur study suggests that machine learning models guiding allocation of limited health resources may consider evaluating and optimizing on utility.

Automatic extraction of medical conditions from free-text radiology reports is critical for supervising computer vision models to interpret medical images. In this work, we show that radiologists labeling reports significantly disagree with radiologists labeling corresponding chest X-ray images, which reduces the quality of report labels as proxies for image labels. We develop and evaluate methods to produce labels from radiology reports that have better agreement with radiologists labeling images. Our best performing method, called VisualCheXbert, uses a biomedically-pretrained BERT model to directly map from a radiology report to the image labels, with a supervisory signal determined by a computer vision model trained to detect medical conditions from chest X-ray images. We find that VisualCheXbert outperforms an approach using an existing radiology report labeler by an average F1 score of 0.14 (95% CI 0.12, 0.17). We also find that VisualCheXbert better agrees with radiologists labeling chest X-ray images than do radiologists labeling the corresponding radiology reports by an average F1 score across several medical conditions of between 0.12 (95% CI 0.09, 0.15) and 0.21 (95% CI 0.18, 0.24)., Comment: Accepted to ACM Conference on Health, Inference, and Learning (ACM-CHIL) 2021

AimsDeep learning (DL), a sub-area of artificial intelligence, has demonstrated great promise at automating diagnostic tasks in pathology, yet its translation into clinical settings has been slow. Few studies have examined its impact on pathologist performance, when embedded into clinical workflows. The identification of H. pylori on H&E stain is a tedious, imprecise task which might benefit from DL assistance. Here, we developed a DL assistant for diagnosing H. pylori in gastric biopsies and tested its impact on pathologist diagnostic accuracy and turnaround time.Methods and resultsH&E-stained whole-slide images (WSI) of 303 gastric biopsies with ground truth confirmation by immunohistochemistry formed the study dataset; 47 and 126 WSI were respectively used to train and optimize our DL assistant to detect H. pylori, and 130 were used in a clinical experiment in which 3 experienced GI pathologists reviewed the same test set with and without assistance. On the test set, the assistant achieved high performance, with a WSI-level area-under-the-receiver-operating-characteristic curve (AUROC) of 0.965 (95% CI 0.934–0.987). On H. pylori-positive cases, assisted diagnoses were faster (, the fixed effect size for assistance = –0.557, p = 0.003) and much more accurate (OR = 13.37, p< 0.001) than unassisted diagnoses. However, assistance increased diagnostic uncertainty on H. pylori- negative cases, resulting in an overall decrease in assisted accuracy (OR = 0.435, p = 0.016) and negligible impact on overall turnaround time ( for assistance = 0.010, p = 0.860).ConclusionsDL can assist pathologists with H. pylori diagnosis, but its integration into clinical workflows requires optimization to mitigate diagnostic uncertainty as a potential consequence of assistance.

Pulmonary embolism (PE) is a life-threatening clinical problem and computed tomography pulmonary angiography (CTPA) is the gold standard for diagnosis. Prompt diagnosis and immediate treatment are critical to avoid high morbidity and mortality rates, yet PE remains among the diagnoses most frequently missed or delayed. In this study, we developed a deep learning model-PENet, to automatically detect PE on volumetric CTPA scans as an end-to-end solution for this purpose. The PENet is a 77-layer 3D convolutional neural network (CNN) pretrained on the Kinetics-600 dataset and fine-tuned on a retrospective CTPA dataset collected from a single academic institution. The PENet model performance was evaluated in detecting PE on data from two different institutions: one as a hold-out dataset from the same institution as the training data and a second collected from an external institution to evaluate model generalizability to an unrelated population dataset. PENet achieved an AUROC of 0.84 [0.82-0.87] on detecting PE on the hold out internal test set and 0.85 [0.81-0.88] on external dataset. PENet also outperformed current state-of-the-art 3D CNN models. The results represent successful application of an end-to-end 3D CNN model for the complex task of PE diagnosis without requiring computationally intensive and time consuming preprocessing and demonstrates sustained performance on data from an external institution. Our model could be applied as a triage tool to automatically identify clinically important PEs allowing for prioritization for diagnostic radiology interpretation and improved care pathways via more efficient diagnosis.

Workshop paper from the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) conference 14. - 19. june, 2020, Seattle, USA. The first Agriculture-Vision Challenge aims to encourage research in developing novel and effective algorithms for agricultural pattern recognition from aerial images, especially for the semantic segmentation task associated with our challenge dataset. Around 57 participating teams from various countries compete to achieve state-of-the-art in aerial agriculture semantic segmentation. The Agriculture-Vision Challenge Dataset was employed, which comprises of 21,061 aerial and multi-spectral farmland images. This paper provides a summary of notable methods and results in the challenge. Our submission server and leaderboard will continue to open for researchers that are interested in this challenge dataset and task; the link can be found here.

Key Points Question Can machine learning models predict improvement of various depressive symptoms with antidepressant treatment based on pretreatment symptom scores and electroencephalographic measures? Findings In this prognostic study, using the machine learning approach of gradient-boosted decision trees, the ElecTreeScore algorithm could reliably distinguish the patients who responded to treatment from those who did not based on various depressive symptoms using pretreatment symptom scores and electroencephalographic features (using the cross-validation approach on 518 patients). Meaning Machine learning approaches that include pretreatment symptom scores and electroencephalographic features may help predict which depressive symptoms will improve with antidepressants., Importance Despite the high prevalence and potential outcomes of major depressive disorder, whether and how patients will respond to antidepressant medications is not easily predicted. Objective To identify the extent to which a machine learning approach, using gradient-boosted decision trees, can predict acute improvement for individual depressive symptoms with antidepressants based on pretreatment symptom scores and electroencephalographic (EEG) measures. Design, Setting, and Participants This prognostic study analyzed data collected as part of the International Study to Predict Optimized Treatment in Depression, a randomized, prospective open-label trial to identify clinically useful predictors and moderators of response to commonly used first-line antidepressant medications. Data collection was conducted at 20 sites spanning 5 countries and including 518 adult outpatients (18-65 years of age) from primary care or specialty care practices who received a diagnosis of current major depressive disorder between December 1, 2008, and September 30, 2013. Patients were antidepressant medication naive or willing to undergo a 1-week washout period of any nonprotocol antidepressant medication. Statistical analysis was conducted from January 5 to June 30, 2019. Exposures Participants with major depressive disorder were randomized in a 1:1:1 ratio to undergo 8 weeks of treatment with escitalopram oxalate (n = 162), sertraline hydrochloride (n = 176), or extended-release venlafaxine hydrochloride (n = 180). Main Outcomes and Measures The primary objective was to predict improvement in individual symptoms, defined as the difference in score for each of the symptoms on the 21-item Hamilton Rating Scale for Depression from baseline to week 8, evaluated using the C index. Results The resulting data set contained 518 patients (274 women; mean [SD] age, 39.0 [12.6] years; mean [SD] 21-item Hamilton Rating Scale for Depression score improvement, 13.0 [7.0]). With the use of 5-fold cross-validation for evaluation, the machine learning model achieved C index scores of 0.8 or higher on 12 of 21 clinician-rated symptoms, with the highest C index score of 0.963 (95% CI, 0.939-1.000) for loss of insight. The importance of any single EEG feature was higher than 5% for prediction of 7 symptoms, with the most important EEG features being the absolute delta band power at the occipital electrode sites (O1, 18.8%; Oz, 6.7%) for loss of insight. Over and above the use of baseline symptom scores alone, the use of both EEG and baseline symptom features was associated with a significant increase in the C index for improvement in 4 symptoms: loss of insight (C index increase, 0.012 [95% CI, 0.001-0.020]), energy loss (C index increase, 0.035 [95% CI, 0.011-0.059]), appetite changes (C index increase, 0.017 [95% CI, 0.003-0.030]), and psychomotor retardation (C index increase, 0.020 [95% CI, 0.008-0.032]). Conclusions and Relevance This study suggests that machine learning may be used to identify independent associations of symptoms and EEG features to predict antidepressant-associated improvements in specific symptoms of depression. The approach should next be prospectively validated in clinical trials and settings. Trial Registration ClinicalTrials.gov Identifier: NCT00693849, This prognostic study examines the extent to which a machine learning approach, using gradient-boosted decision trees, can predict acute improvement in individual depressive symptoms with antidepressants based on pretreatment symptom scores and electroencephalographic measures.

Artificial intelligence (AI) algorithms continue to rival human performance on a variety of clinical tasks, while their actual impact on human diagnosticians, when incorporated into clinical workflows, remains relatively unexplored. In this study, we developed a deep learning-based assistant to help pathologists differentiate between two subtypes of primary liver cancer, hepatocellular carcinoma and cholangiocarcinoma, on hematoxylin and eosin-stained whole-slide images (WSI), and evaluated its effect on the diagnostic performance of 11 pathologists with varying levels of expertise. Our model achieved accuracies of 0.885 on a validation set of 26 WSI, and 0.842 on an independent test set of 80 WSI. Although use of the assistant did not change the mean accuracy of the 11 pathologists (p = 0.184, OR = 1.281), it significantly improved the accuracy (p = 0.045, OR = 1.499) of a subset of nine pathologists who fell within well-defined experience levels (GI subspecialists, non-GI subspecialists, and trainees). In the assisted state, model accuracy significantly impacted the diagnostic decisions of all 11 pathologists. As expected, when the model’s prediction was correct, assistance significantly improved accuracy (p = 0.000, OR = 4.289), whereas when the model’s prediction was incorrect, assistance significantly decreased accuracy (p = 0.000, OR = 0.253), with both effects holding across all pathologist experience levels and case difficulty levels. Our results highlight the challenges of translating AI models into the clinical setting, and emphasize the importance of taking into account potential unintended negative consequences of model assistance when designing and testing medical AI-assistance tools.

Osteoclasts are bone-resorbing cells differentiated from macrophage/monocyte precursors in response to macrophage colony-stimulating factor (M-CSF) and receptor activator of NF-κB ligand (RANKL). In vitro models are principally based on primary bone marrow macrophages (BMMs), but RAW 264.7 cells are frequently used because they are widely available, easy to culture, and more amenable to genetic manipulation than primary cells. Increasing evidence, however, has shown that the vastly different origins of these two cell types may have important effects on cell behavior. In particular, M-CSF is a prerequisite for the differentiation of BMMs, by promoting survival and proliferation and priming the cells for RANKL induction. RAW 264.7 cells readily form osteoclasts in the presence of RANKL, but M-CSF is not required. Based on these key differences, we sought to understand their functional implications and how it might affect osteoclast differentiation and related signaling pathways. Using a robust and high-throughput proteomics strategy, we quantified the global protein changes in osteoclasts derived from BMMs and RAW 264.7 cells at 1, 3, and 5 days of differentiation. Data are available via ProteomeXchange with the identifier PXD009610. Correlation analysis of the proteomes demonstrated low concordance between the two cell types (R2 ≈ 0.13). Bioinformatics analysis indicate that RANKL-dependent signaling was intact in RAW 264.7 cells, but biological processes known to be dependent on M-CSF were significantly different, including cell cycle control, cytoskeletal organization, and apoptosis. RAW 264.7 cells exhibited constitutive activation of Erk and Akt that was dependent on the activity of Abelson tyrosine kinase, and the timing of Erk and Akt activation was significantly different between BMMs and RAW 264.7 cells. Our findings provide the first evidence for major discrepancies between BMMs and RAW 264.7 cells, indicating that careful consideration is needed when using the RAW 264.7 cell line for studying M-CSF-dependent signaling and functions. © 2018 American Society for Bone and Mineral Research. © 2018 The Authors. JBMR Plus published by Wiley Periodicals, Inc. on behalf of American Society for Bone and Mineral Research.

Summary Data labeling is often the limiting step in machine learning because it requires time from trained experts. To address the limitation on labeled data, contrastive learning, among other unsupervised learning methods, leverages unlabeled data to learn representations of data. Here, we propose a contrastive learning framework that utilizes metadata for selecting positive and negative pairs when training on unlabeled data. We demonstrate its application in the healthcare domain on heart and lung sound recordings. The increasing availability of heart and lung sound recordings due to adoption of digital stethoscopes lends itself as an opportunity to demonstrate the application of our contrastive learning method. Compared to contrastive learning with augmentations, the contrastive learning model leveraging metadata for pair selection utilizes clinical information associated with lung and heart sound recordings. This approach uses shared context of the recordings on the patient level using clinical information including age, sex, weight, location of sounds, etc. We show improvement in downstream tasks for diagnosing heart and lung sounds when leveraging patient-specific representations in selecting positive and negative pairs. This study paves the path for medical applications of contrastive learning that leverage clinical information. We have made our code available here: https://github.com/stanfordmlgroup/selfsupervised-lungandheartsounds., Graphical abstract, Highlights • Contrastive learning uses unlabeled data to learn representations • A new contrastive learning framework is proposed for metadata pair selection • We show its application in medical heart and lung sound data and metadata • The contrastive learning strategy only needs 10% of labeled training data, The bigger picture Annotating data at scale is time consuming, especially in specialized domains, such as healthcare, agriculture, and autonomous driving. The scarcity of labeled data can limit the effectiveness of supervised learning. In contrast, there is usually access to more unlabeled data. Unlabeled data can be used through unsupervised learning. One type of unsupervised learning is self-supervised learning, where representations of data are learned from unlabeled data through pretext tasks and are later used for supervised learning tasks. We propose a new contrastive learning framework that leverages metadata in selecting pairs during contrastive learning. We demonstrate the application of the framework in diagnosing heart and lung diseases through heart and lung sound recordings and associated clinical metadata. Our strategy could also be applied in other medical settings such as electronic health records and medical imaging, as well as beyond medicine., The project focuses on a common problem in machine learning in specialized fields: a lack of labeled data. We propose a solution with a contrastive learning framework to leverage unlabeled data and use associated metadata for pair selection to learn data representations. With this framework, we are able to improve downstream tasks performances with limited labeled data. We choose to apply this framework in healthcare, where unlabeled data are abundant and data labeling is expensive.

How can we effectively leverage the domain knowledge from remote sensing to better segment agriculture land cover from satellite images? In this paper, we propose a novel, model-agnostic, data-fusion approach for vegetation-related computer vision tasks. Motivated by the various Vegetation Indices (VIs), which are introduced by domain experts, we systematically reviewed the VIs that are widely used in remote sensing and their feasibility to be incorporated in deep neural networks. To fully leverage the Near-Infrared channel, the traditional Red-Green-Blue channels, and Vegetation Index or its variants, we propose a Generalized Vegetation Index (GVI), a lightweight module that can be easily plugged into many neural network architectures to serve as an additional information input. To smoothly train models with our GVI, we developed an Additive Group Normalization (AGN) module that does not require extra parameters of the prescribed neural networks. Our approach has improved the IoUs of vegetation-related classes by 0.9-1.3 percent and consistently improves the overall mIoU by 2 percent on our baseline., Comment: CVPR 2020 - Vision for Agriculture; https://www.agriculture-vision.com

Diffuse Large B-Cell Lymphoma (DLBCL) is the most common non-Hodgkin lymphoma. Though histologically DLBCL shows varying morphologies, no morphologic features have been consistently demonstrated to correlate with prognosis. We present a morphologic analysis of histology sections from 209 DLBCL cases with associated clinical and cytogenetic data. Duplicate tissue core sections were arranged in tissue microarrays (TMAs), and replicate sections were stained with H&E and immunohistochemical stains for CD10, BCL6, MUM1, BCL2, and MYC. The TMAs are accompanied by pathologist-annotated regions-of-interest (ROIs) that identify areas of tissue representative of DLBCL. We used a deep learning model to segment all tumor nuclei in the ROIs, and computed several geometric features for each segmented nucleus. We fit a Cox proportional hazards model to demonstrate the utility of these geometric features in predicting survival outcome, and found that it achieved a C-index (95% CI) of 0.635 (0.574,0.691). Our finding suggests that geometric features computed from tumor nuclei are of prognostic importance, and should be validated in prospective studies., Corrections to folder structure figure

The extraction of labels from radiology text reports enables large-scale training of medical imaging models. Existing approaches to report labeling typically rely either on sophisticated feature engineering based on medical domain knowledge or manual annotations by experts. In this work, we introduce a BERT-based approach to medical image report labeling that exploits both the scale of available rule-based systems and the quality of expert annotations. We demonstrate superior performance of a biomedically pretrained BERT model first trained on annotations of a rule-based labeler and then finetuned on a small set of expert annotations augmented with automated backtranslation. We find that our final model, CheXbert, is able to outperform the previous best rules-based labeler with statistical significance, setting a new SOTA for report labeling on one of the largest datasets of chest x-rays.

International audience; We present an 802.15.4 compatible transceiver that operates without any off-chip frequency reference. With integrated Cortex-M0, the chip can also transmit BLE beacons with only three external connections (power, ground, and antenna). The RF transmitter operates with >10% system efficiency at-10 dBm output power from a regulated supply. The entire chip, including the microprocessor, can operate below 1 mW peak power when transmitting. The analog receiver power consumption is 1.03 mW from a 1.5V battery.

INTRODUCTION AND OBJECTIVES:The role of multiparametric MRI in clinical care is rapidly expanding due to its ability to improve prostate cancer detection. However, MRI interpretation suffers from a...

Background: The absolute risk reduction (ARR) in cardiovascular events from therapy is generally assumed to be proportional to baseline risk—such that high-risk patients benefit most. Yet newer analyses have proposed using randomized trial data to develop models that estimate individual treatment effects. We tested 2 hypotheses: first, that models of individual treatment effects would reveal that benefit from intensive blood pressure therapy is proportional to baseline risk; and second, that a machine learning approach designed to predict heterogeneous treatment effects—the X-learner meta-algorithm—is equivalent to a conventional logistic regression approach. Methods and Results: We compared conventional logistic regression to the X-learner approach for prediction of 3-year cardiovascular disease event risk reduction from intensive (target systolic blood pressure Conclusions: Predictions for individual treatment effects from trial data reveal that patients may experience ARRs not simply proportional to baseline cardiovascular disease risk. Machine learning methods may improve discrimination and calibration of individualized treatment effect estimates from clinical trial data. Clinical Trial Registration: URL: https://www.clinicaltrials.gov . Unique identifiers: NCT01206062; NCT00000620.

Importance Physicians are required to work with rapidly growing amounts of medical data. Approximately 62% of time per patient is devoted to reviewing electronic health records (EHRs), with clinical data review being the most time-consuming portion. Objective To determine whether an artificial intelligence (AI) system developed to organize and display new patient referral records would improve a clinician's ability to extract patient information compared with the current standard of care. Design, setting, and participants In this prognostic study, an AI system was created to organize patient records and improve data retrieval. To evaluate the system on time and accuracy, a nonblinded, prospective study was conducted at a single academic medical center. Recruitment emails were sent to all physicians in the gastroenterology division, and 12 clinicians agreed to participate. Each of the clinicians participating in the study received 2 referral records: 1 AI-optimized patient record and 1 standard (non-AI-optimized) patient record. For each record, clinicians were asked 22 questions requiring them to search the assigned record for clinically relevant information. Clinicians reviewed records from June 1 to August 30, 2020. Main outcomes and measures The time required to answer each question, along with accuracy, was measured for both records, with and without AI optimization. Participants were asked to assess overall satisfaction with the AI system, their preferred review method (AI-optimized vs standard), and other topics to assess clinical utility. Results Twelve gastroenterology physicians/fellows completed the study. Compared with standard (non-AI-optimized) patient record review, the AI system saved first-time physician users 18% of the time used to answer the clinical questions (10.5 [95% CI, 8.5-12.6] vs 12.8 [95% CI, 9.4-16.2] minutes; P = .02). There was no significant decrease in accuracy when physicians retrieved important patient information (83.7% [95% CI, 79.3%-88.2%] with the AI-optimized vs 86.0% [95% CI, 81.8%-90.2%] without the AI-optimized record; P = .81). Survey responses from physicians were generally positive across all questions. Eleven of 12 physicians (92%) preferred the AI-optimized record review to standard review. Despite a learning curve pointed out by respondents, 11 of 12 physicians believed that the technology would save them time to assess new patient records and were interested in using this technology in their clinic. Conclusions and relevance In this prognostic study, an AI system helped physicians extract relevant patient information in a shorter time while maintaining high accuracy. This finding is particularly germane to the ever-increasing amounts of medical data and increased stressors on clinicians. Increased user familiarity with the AI system, along with further enhancements in the system itself, hold promise to further improve physician data extraction from large quantities of patient health records.

Data augmentation has led to substantial improvements in the performance and generalization of deep models, and remains a highly adaptable method to evolving model architectures and varying amounts of data—in particular, extremely scarce amounts of available training data. In this paper, we present a novel method of applying Möbius transformations to augment input images during training. Möbius transformations are bijective conformal maps that generalize image translation to operate over complex inversion in pixel space. As a result, Möbius transformations can operate on the sample level and preserve data labels. We show that the inclusion of Möbius transformations during training enables improved generalization over prior sample-level data augmentation techniques such as cutout and standard crop-and-flip transformations, most notably in low data regimes.

Aims Deep learning (DL), a sub-area of artificial intelligence, has demonstrated great promise at automating diagnostic tasks in pathology, yet its translation into clinical settings has been slow. Few studies have examined its impact on pathologist performance, when embedded into clinical workflows. The identification of H. pylori on HE 47 and 126 WSI were respectively used to train and optimize the DL assistant to detect H. pylori, and 130 were used in a clinical experiment in which 3 experienced GI pathologists reviewed the same test set with and without assistance. On the test set, the assistant achieved high performance, with a WSI-level area under the receiver-operating-characteristic curve (AUROC) of 0.965 (95% CI 0.934–0.987). On H. pylori-positive cases, assisted diagnoses were faster ( β ˆ , the fixed effect size for assistance ​= ​−0.557, p ​= ​0.003) and much more accurate (OR ​= ​13.37, p ​ β ˆ for assistance ​= ​0.010, p ​= ​0.860). Conclusions DL can assist pathologists with H. pylori diagnosis, but its integration into clinical workflows requires optimization to mitigate diagnostic uncertainty as a potential consequence of assistance.

Large, labeled datasets have driven deep learning methods to achieve expert-level performance on a variety of medical imaging tasks. We present CheXpert, a large dataset that contains 224,316 chest radiographs of 65,240 patients. We design a labeler to automatically detect the presence of 14 observations in radiology reports, capturing uncertainties inherent in radiograph interpretation. We investigate different approaches to using the uncertainty labels for training convolutional neural networks that output the probability of these observations given the available frontal and lateral radiographs. On a validation set of 200 chest radiographic studies which were manually annotated by 3 board-certified radiologists, we find that different uncertainty approaches are useful for different pathologies. We then evaluate our best model on a test set composed of 500 chest radiographic studies annotated by a consensus of 5 board-certified radiologists, and compare the performance of our model to that of 3 additional radiologists in the detection of 5 selected pathologies. On Cardiomegaly, Edema, and Pleural Effusion, the model ROC and PR curves lie above all 3 radiologist operating points. We release the dataset to the public as a standard benchmark to evaluate performance of chest radiograph interpretation models. The dataset is freely available at https://stanfordmlgroup.github.io/competitions/chexpert ., Published in AAAI 2019

Computerized electrocardiogram (ECG) interpretation plays a critical role in the clinical ECG workflow1. Widely available digital ECG data and the algorithmic paradigm of deep learning2 present an opportunity to substantially improve the accuracy and scalability of automated ECG analysis. However, a comprehensive evaluation of an end-to-end deep learning approach for ECG analysis across a wide variety of diagnostic classes has not been previously reported. Here, we develop a deep neural network (DNN) to classify 12 rhythm classes using 91,232 single-lead ECGs from 53,549 patients who used a single-lead ambulatory ECG monitoring device. When validated against an independent test dataset annotated by a consensus committee of board-certified practicing cardiologists, the DNN achieved an average area under the receiver operating characteristic curve (ROC) of 0.97. The average F1 score, which is the harmonic mean of the positive predictive value and sensitivity, for the DNN (0.837) exceeded that of average cardiologists (0.780). With specificity fixed at the average specificity achieved by cardiologists, the sensitivity of the DNN exceeded the average cardiologist sensitivity for all rhythm classes. These findings demonstrate that an end-to-end deep learning approach can classify a broad range of distinct arrhythmias from single-lead ECGs with high diagnostic performance similar to that of cardiologists. If confirmed in clinical settings, this approach could reduce the rate of misdiagnosed computerized ECG interpretations and improve the efficiency of expert human ECG interpretation by accurately triaging or prioritizing the most urgent conditions. Analysis of electrocardiograms using an end-to-end deep learning approach can detect and classify cardiac arrhythmia with high accuracy, similar to that of cardiologists.

The Regulator of G-protein Signaling 12 (Rgs12) is important for osteoclast (OC) differentiation, and its deletion in vivo protected mice against pathological bone loss. To characterize its mechanism in osteoclastogenesis, we selectively deleted Rgs12 in OC precursors using the LysM-Cre transgenic line or overexpressed the gene in RAW264.7 cells. Rgs12 deletion led to increased bone mass with decreased OC numbers, whereas its overexpression increased OC number and size. Proteomics analysis of Rgs12-deficient OCs identified an upregulation of antioxidant enzymes under the transcriptional regulation of Nrf2, the master regulator of oxidative stress. We confirmed an increase of Nrf2 activity and impaired production in Rgs12-deficient cells. Conversely, Rgs12 overexpression suppressed Nrf2 through a mechanism dependent on the 26S proteasome, and promoted RANKL-induced phosphorylation of ERK1/2 and NFκB, which was abrogated by antioxidant treatment. We therefore identified a novel role of Rgs12 in regulating Nrf2, thereby controlling cellular redox state and OC differentiation.

Background Magnetic resonance imaging (MRI) of the knee is the preferred method for diagnosing knee injuries. However, interpretation of knee MRI is time-intensive and subject to diagnostic error and variability. An automated system for interpreting knee MRI could prioritize high-risk patients and assist clinicians in making diagnoses. Deep learning methods, in being able to automatically learn layers of features, are well suited for modeling the complex relationships between medical images and their interpretations. In this study we developed a deep learning model for detecting general abnormalities and specific diagnoses (anterior cruciate ligament [ACL] tears and meniscal tears) on knee MRI exams. We then measured the effect of providing the model’s predictions to clinical experts during interpretation. Methods and findings Our dataset consisted of 1,370 knee MRI exams performed at Stanford University Medical Center between January 1, 2001, and December 31, 2012 (mean age 38.0 years; 569 [41.5%] female patients). The majority vote of 3 musculoskeletal radiologists established reference standard labels on an internal validation set of 120 exams. We developed MRNet, a convolutional neural network for classifying MRI series and combined predictions from 3 series per exam using logistic regression. In detecting abnormalities, ACL tears, and meniscal tears, this model achieved area under the receiver operating characteristic curve (AUC) values of 0.937 (95% CI 0.895, 0.980), 0.965 (95% CI 0.938, 0.993), and 0.847 (95% CI 0.780, 0.914), respectively, on the internal validation set. We also obtained a public dataset of 917 exams with sagittal T1-weighted series and labels for ACL injury from Clinical Hospital Centre Rijeka, Croatia. On the external validation set of 183 exams, the MRNet trained on Stanford sagittal T2-weighted series achieved an AUC of 0.824 (95% CI 0.757, 0.892) in the detection of ACL injuries with no additional training, while an MRNet trained on the rest of the external data achieved an AUC of 0.911 (95% CI 0.864, 0.958). We additionally measured the specificity, sensitivity, and accuracy of 9 clinical experts (7 board-certified general radiologists and 2 orthopedic surgeons) on the internal validation set both with and without model assistance. Using a 2-sided Pearson’s chi-squared test with adjustment for multiple comparisons, we found no significant differences between the performance of the model and that of unassisted general radiologists in detecting abnormalities. General radiologists achieved significantly higher sensitivity in detecting ACL tears (p-value = 0.002; q-value = 0.019) and significantly higher specificity in detecting meniscal tears (p-value = 0.003; q-value = 0.019). Using a 1-tailed t test on the change in performance metrics, we found that providing model predictions significantly increased clinical experts’ specificity in identifying ACL tears (p-value < 0.001; q-value = 0.006). The primary limitations of our study include lack of surgical ground truth and the small size of the panel of clinical experts. Conclusions Our deep learning model can rapidly generate accurate clinical pathology classifications of knee MRI exams from both internal and external datasets. Moreover, our results support the assertion that deep learning models can improve the performance of clinical experts during medical imaging interpretation. Further research is needed to validate the model prospectively and to determine its utility in the clinical setting., Nicholas Bien and colleagues present an automated system for interpreting knee magnetic resonance (MR) images to prioritize high-risk patients and assist clinicians in making diagnoses., Author summary Why was this study done? We wanted to see if a deep learning model could succeed in the clinically important task of detecting disorders in knee magnetic resonance imaging (MRI) scans. We wanted to determine whether a deep learning model could improve the diagnostic accuracy, specificity, or sensitivity of clinical experts, including general radiologists and orthopedic surgeons. What did the researchers do and find? Our deep learning model predicted 3 outcomes for knee MRI exams (anterior cruciate ligament [ACL] tears, meniscal tears, and general abnormalities) in a matter of seconds and with similar performance to that of general radiologists. We experimented with providing model outputs to general radiologists and orthopedic surgeons during interpretation and observed statistically significant improvement in diagnosis of ACL tears with model assistance. When externally validated on a dataset from a different institution, the model picked up ACL tears with high discriminative ability. What do these findings mean? Deep learning has the potential to provide rapid preliminary results following MRI exams and improve access to quality MRI diagnoses in the absence of specialist radiologists. Providing clinical experts with predictions from a deep learning model could improve the quality and consistency of MRI interpretation.

Background Chest radiograph interpretation is critical for the detection of thoracic diseases, including tuberculosis and lung cancer, which affect millions of people worldwide each year. This time-consuming task typically requires expert radiologists to read the images, leading to fatigue-based diagnostic error and lack of diagnostic expertise in areas of the world where radiologists are not available. Recently, deep learning approaches have been able to achieve expert-level performance in medical image interpretation tasks, powered by large network architectures and fueled by the emergence of large labeled datasets. The purpose of this study is to investigate the performance of a deep learning algorithm on the detection of pathologies in chest radiographs compared with practicing radiologists. Methods and findings We developed CheXNeXt, a convolutional neural network to concurrently detect the presence of 14 different pathologies, including pneumonia, pleural effusion, pulmonary masses, and nodules in frontal-view chest radiographs. CheXNeXt was trained and internally validated on the ChestX-ray8 dataset, with a held-out validation set consisting of 420 images, sampled to contain at least 50 cases of each of the original pathology labels. On this validation set, the majority vote of a panel of 3 board-certified cardiothoracic specialist radiologists served as reference standard. We compared CheXNeXt’s discriminative performance on the validation set to the performance of 9 radiologists using the area under the receiver operating characteristic curve (AUC). The radiologists included 6 board-certified radiologists (average experience 12 years, range 4–28 years) and 3 senior radiology residents, from 3 academic institutions. We found that CheXNeXt achieved radiologist-level performance on 11 pathologies and did not achieve radiologist-level performance on 3 pathologies. The radiologists achieved statistically significantly higher AUC performance on cardiomegaly, emphysema, and hiatal hernia, with AUCs of 0.888 (95% confidence interval [CI] 0.863–0.910), 0.911 (95% CI 0.866–0.947), and 0.985 (95% CI 0.974–0.991), respectively, whereas CheXNeXt’s AUCs were 0.831 (95% CI 0.790–0.870), 0.704 (95% CI 0.567–0.833), and 0.851 (95% CI 0.785–0.909), respectively. CheXNeXt performed better than radiologists in detecting atelectasis, with an AUC of 0.862 (95% CI 0.825–0.895), statistically significantly higher than radiologists' AUC of 0.808 (95% CI 0.777–0.838); there were no statistically significant differences in AUCs for the other 10 pathologies. The average time to interpret the 420 images in the validation set was substantially longer for the radiologists (240 minutes) than for CheXNeXt (1.5 minutes). The main limitations of our study are that neither CheXNeXt nor the radiologists were permitted to use patient history or review prior examinations and that evaluation was limited to a dataset from a single institution. Conclusions In this study, we developed and validated a deep learning algorithm that classified clinically important abnormalities in chest radiographs at a performance level comparable to practicing radiologists. Once tested prospectively in clinical settings, the algorithm could have the potential to expand patient access to chest radiograph diagnostics., In their study, Pranav Rajpurkar and colleagues test a deep learning algorithm that classifies clinically important abnormalities in chest radiographs., Author summary Why was this study done? Chest radiographs are the most common medical imaging test in the world and critical for diagnosing common thoracic diseases. Radiograph interpretation is a time-consuming task, and there is shortage of qualified trained radiologists in many healthcare systems. Deep learning algorithms that have been developed to provide diagnostic chest radiograph interpretation have not been compared to expert human radiologist performance. What did the researchers do and find? We developed a deep learning algorithm to concurrently detect 14 clinically important pathologies in chest radiographs. The algorithm can also localize parts of the image most indicative of each pathology. We evaluated the algorithm against 9 practicing radiologists on a validation set of 420 images for which the majority vote of 3 cardiothoracic specialty radiologists served as ground truth. The algorithm achieved performance equivalent to the practicing radiologists on 10 pathologies, better on 1 pathology, and worse on 3 pathologies. Radiologists labeled the 420 images in 240 minutes on average, and the algorithm labeled them in 1.5 minutes. What do these findings mean? Deep learning algorithms can diagnose certain pathologies in chest radiographs at a level comparable to practicing radiologists on a single institution dataset. After clinical validation, algorithms such as the one presented in this work could be used to increase access to rapid, high-quality chest radiograph interpretation.

Translation-based methods for grammar correction that directly map noisy, ungrammatical text to their clean counterparts are able to correct a broad range of errors; however, such techniques are bottlenecked by the need for a large parallel corpus of noisy and clean sentence pairs. In this paper, we consider synthesizing parallel data by noising a clean monolingual corpus. While most previous approaches introduce perturbations using features computed from local context windows, we instead develop error generation processes using a neural sequence transduction model trained to translate clean examples to their noisy counterparts. Given a corpus of clean examples, we propose beam search noising procedures to synthesize additional noisy examples that human evaluators were nearly unable to discriminate from nonsynthesized examples. Surprisingly, when trained on additional data synthesized using our best-performing noising scheme, our model approaches the same performance as when trained on additional nonsynthesized data.

We develop an algorithm that accurately detects Atrial Fibrillation (AF) episodes from photoplethysmograms (PPG) recorded in ambulatory free-living conditions. We collect and annotate a dataset containing more than 4000 hours of PPG recorded from a wrist-worn device. Using a 50-layer convolutional neural network, we achieve a test AUC of 95% and show robustness to motion artifacts inherent to PPG signals. Continuous and accurate detection of AF from PPG has the potential to transform consumer wearable devices into clinically useful medical monitoring tools., Comment: Equal contribution by Maxime Voisin and Yichen Shen

Improving the quality of end-of-life care for hospitalized patients is a priority for healthcare organizations. Studies have shown that physicians tend to over-estimate prognoses, which in combination with treatment inertia results in a mismatch between patients wishes and actual care at the end of life. We describe a method to address this problem using Deep Learning and Electronic Health Record (EHR) data, which is currently being piloted, with Institutional Review Board approval, at an academic medical center. The EHR data of admitted patients are automatically evaluated by an algorithm, which brings patients who are likely to benefit from palliative care services to the attention of the Palliative Care team. The algorithm is a Deep Neural Network trained on the EHR data from previous years, to predict all-cause 3-12 month mortality of patients as a proxy for patients that could benefit from palliative care. Our predictions enable the Palliative Care team to take a proactive approach in reaching out to such patients, rather than relying on referrals from treating physicians, or conduct time consuming chart reviews of all patients. We also present a novel interpretation technique which we use to provide explanations of the model's predictions., IEEE International Conference on Bioinformatics and Biomedicine 2017