Automated Recognition of Regional Wall Motion Abnormalities Through Deep Neural Network Interpretation of Transthoracic Echocardiography.

BACKGROUND: Automated interpretation of echocardiography by deep neural networks could support clinical reporting and improve efficiency. Whereas previous studies have evaluated spatial relationships using still frame images, we aimed to train and test a deep neural network for video analysis by combining spatial and temporal information, to automate the recognition of left ventricular regional wall motion abnormalities. METHODS: We collected a series of transthoracic echocardiography examinations performed between July 2017 and April 2018 in 2 tertiary care hospitals. Regional wall abnormalities were defined by experienced physiologists and confirmed by trained cardiologists. First, we developed a 3-dimensional convolutional neural network model for view selection ensuring stringent image quality control. Second, a U-net model segmented images to annotate the location of each left ventricular wall. Third, a final 3-dimensional convolutional neural network model evaluated echocardiographic videos from 4 standard views, before and after segmentation, and calculated a wall motion abnormality confidence level (0-1) for each segment. To evaluate model stability, we performed 5-fold cross-validation and external validation. RESULTS: In a series of 10 638 echocardiograms, our view selection model identified 6454 (61%) examinations with sufficient image quality in all standard views. In this training set, 2740 frames were annotated to develop the segmentation model, which achieved a Dice similarity coefficient of 0.756. External validation was performed in 1756 examinations from an independent hospital. A regional wall motion abnormality was observed in 8.9% and 4.9% in the training and external validation datasets, respectively. The final model recognized regional wall motion abnormalities in the cross-validation and external validation datasets with an area under the receiver operating characteristic curve of 0.912 (95% CI, 0.896-0.928) and 0.891 (95% CI, 0.834-0.948), respectively. In the external validation dataset, the sensitivity was 81.8% (95% CI, 73.8%-88.2%), and specificity was 81.6% (95% CI, 80.4%-82.8%). CONCLUSIONS: In echocardiographic examinations of sufficient image quality, it is feasible for deep neural networks to automate the recognition of regional wall motion abnormalities using temporal and spatial information from moving images. Further investigation is required to optimize model performance and evaluate clinical applications.

Application of an Artificial Intelligence Trilogy to Accelerate Processing of Suspected Patients With SARS-CoV-2 at a Smart Quarantine Station: Observational Study.

BACKGROUND: As the COVID-19 epidemic increases in severity, the burden of quarantine stations outside emergency departments (EDs) at hospitals is increasing daily. To address the high screening workload at quarantine stations, all staff members with medical licenses are required to work shifts in these stations. Therefore, it is necessary to simplify the workflow and decision-making process for physicians and surgeons from all subspecialties. OBJECTIVE: The aim of this paper is to demonstrate how the National Cheng Kung University Hospital artificial intelligence (AI) trilogy of diversion to a smart quarantine station, AI-assisted image interpretation, and a built-in clinical decision-making algorithm improves medical care and reduces quarantine processing times. METHODS: This observational study on the emerging COVID-19 pandemic included 643 patients. An "AI trilogy" of diversion to a smart quarantine station, AI-assisted image interpretation, and a built-in clinical decision-making algorithm on a tablet computer was applied to shorten the quarantine survey process and reduce processing time during the COVID-19 pandemic. RESULTS: The use of the AI trilogy facilitated the processing of suspected cases of COVID-19 with or without symptoms; also, travel, occupation, contact, and clustering histories were obtained with the tablet computer device. A separate AI-mode function that could quickly recognize pulmonary infiltrates on chest x-rays was merged into the smart clinical assisting system (SCAS), and this model was subsequently trained with COVID-19 pneumonia cases from the GitHub open source data set. The detection rates for posteroanterior and anteroposterior chest x-rays were 55/59 (93%) and 5/11 (45%), respectively. The SCAS algorithm was continuously adjusted based on updates to the Taiwan Centers for Disease Control public safety guidelines for faster clinical decision making. Our ex vivo study demonstrated the efficiency of disinfecting the tablet computer surface by wiping it twice with 75% alcohol sanitizer. To further analyze the impact of the AI application in the quarantine station, we subdivided the station group into groups with or without AI. Compared with the conventional ED (n=281), the survey time at the quarantine station (n=1520) was significantly shortened; the median survey time at the ED was 153 minutes (95% CI 108.5-205.0), vs 35 minutes at the quarantine station (95% CI 24-56; P<.001). Furthermore, the use of the AI application in the quarantine station reduced the survey time in the quarantine station; the median survey time without AI was 101 minutes (95% CI 40-153), vs 34 minutes (95% CI 24-53) with AI in the quarantine station (P<.001). CONCLUSIONS: The AI trilogy improved our medical care workflow by shortening the quarantine survey process and reducing the processing time, which is especially important during an emerging infectious disease epidemic.

Detecting Potential Adverse Drug Reactions Using a Deep Neural Network Model.

BACKGROUND: Adverse drug reactions (ADRs) are common and are the underlying cause of over a million serious injuries and deaths each year. The most familiar method to detect ADRs is relying on spontaneous reports. Unfortunately, the low reporting rate of spontaneous reports is a serious limitation of pharmacovigilance. OBJECTIVE: The objective of this study was to identify a method to detect potential ADRs of drugs automatically using a deep neural network (DNN). METHODS: We designed a DNN model that utilizes the chemical, biological, and biomedical information of drugs to detect ADRs. This model aimed to fulfill two main purposes: identifying the potential ADRs of drugs and predicting the possible ADRs of a new drug. For improving the detection performance, we distributed representations of the target drugs in a vector space to capture the drug relationships using the word-embedding approach to process substantial biomedical literature. Moreover, we built a mapping function to address new drugs that do not appear in the dataset. RESULTS: Using the drug information and the ADRs reported up to 2009, we predicted the ADRs of drugs recorded up to 2012. There were 746 drugs and 232 new drugs, which were only recorded in 2012 with 1325 ADRs. The experimental results showed that the overall performance of our model with mean average precision at top-10 achieved is 0.523 and the rea under the receiver operating characteristic curve (AUC) score achieved is 0.844 for ADR prediction on the dataset. CONCLUSIONS: Our model is effective in identifying the potential ADRs of a drug and the possible ADRs of a new drug. Most importantly, it can detect potential ADRs irrespective of whether they have been reported in the past.

Systematic Protein Prioritization for Targeted Proteomics Studies through Literature Mining.

There are more than 3.7 million published articles on the biological functions or disease implications of proteins, constituting an important resource of proteomics knowledge. However, it is difficult to summarize the millions of proteomics findings in the literature manually and quantify their relevance to the biology and diseases of interest. We developed a fully automated bioinformatics framework to identify and prioritize proteins associated with any biological entity. We used the 22 targeted areas of the Biology/Disease-driven (B/D)-Human Proteome Project (HPP) as examples, prioritized the relevant proteins through their Protein Universal Reference Publication-Originated Search Engine (PURPOSE) scores, validated the relevance of the score by comparing the protein prioritization results with a curated database, computed the scores of proteins across the topics of B/D-HPP, and characterized the top proteins in the common model organisms. We further extended the bioinformatics workflow to identify the relevant proteins in all organ systems and human diseases and deployed a cloud-based tool to prioritize proteins related to any custom search terms in real time. Our tool can facilitate the prioritization of proteins for any organ system or disease of interest and can contribute to the development of targeted proteomic studies for precision medicine.

Predicting Mortality Using Machine Learning Algorithms in Patients Who Require Renal Replacement Therapy in the Critical Care Unit.

Background: General severity of illness scores are not well calibrated to predict mortality among patients receiving renal replacement therapy (RRT) for acute kidney injury (AKI). We developed machine learning models to make mortality prediction and compared their performance to that of the Sequential Organ Failure Assessment (SOFA) and HEpatic failure, LactatE, NorepInephrine, medical Condition, and Creatinine (HELENICC) scores. Methods: We extracted routinely collected clinical data for AKI patients requiring RRT in the MIMIC and eICU databases. The development models were trained in 80% of the pooled dataset and tested in the rest of the pooled dataset. We compared the area under the receiver operating characteristic curves (AUCs) of four machine learning models (multilayer perceptron [MLP], logistic regression, XGBoost, and random forest [RF]) to that of the SOFA, nonrenal SOFA, and HELENICC scores and assessed calibration, sensitivity, specificity, positive (PPV) and negative (NPV) predicted values, and accuracy. Results: The mortality AUC of machine learning models was highest for XGBoost (0.823; 95% confidence interval [CI], 0.791−0.854) in the testing dataset, and it had the highest accuracy (0.758). The XGBoost model showed no evidence of lack of fit with the Hosmer−Lemeshow test (p > 0.05). Conclusion: XGBoost provided the highest performance of mortality prediction for patients with AKI requiring RRT compared with previous scoring systems.

Contextualizing Genes by Using Text-Mined Co-Occurrence Features for Cancer Gene Panel Discovery.

Developing a biomedical-explainable and validatable text mining pipeline can help in cancer gene panel discovery. We create a pipeline that can contextualize genes by using text-mined co-occurrence features. We apply Biomedical Natural Language Processing (BioNLP) techniques for literature mining in the cancer gene panel. A literature-derived 4,679 × 4,630 gene term-feature matrix was built. The EGFR L858R and T790M, and BRAF V600E genetic variants are important mutation term features in text mining and are frequently mutated in cancer. We validate the cancer gene panel by the mutational landscape of different cancer types. The cosine similarity of gene frequency between text mining and a statistical result from clinical sequencing data is 80.8%. In different machine learning models, the best accuracy for the prediction of two different gene panels, including MSK-IMPACT (Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets), and Oncomine cancer gene panel, is 0.959, and 0.989, respectively. The receiver operating characteristic (ROC) curve analysis confirmed that the neural net model has a better prediction performance (Area under the ROC curve (AUC) = 0.992). The use of text-mined co-occurrence features can contextualize each gene. We believe the approach is to evaluate several existing gene panels, and show that we can use part of the gene panel set to predict the remaining genes for cancer discovery.