Ensemble Deep Learning Model Research Articles

An ensemble deep-learning framework for landslide susceptibility assessment using multiple blocks: a case study of Wenchuan area, China

The improvement of the landslide susceptibility mapping (LSM) is a long-standing problem, as it provides basics for hazard mitigation. Recently, hybrid ensemble deep learning (DL) techniques have witnessed the potential for this purpose. In this paper, we proposed a novel ensemble DL model, namely GL-ResNet, which employs the conventional ResNet blocks for landslide feature extraction, long-short term memory (LSTM) structures for information storage, and a proposed GoogLeNet block (GBlk) to broaden model perception ability. To validate the model performance, a landslide inventory containing 1147 historical landslide polygons and the data of 12 landslide factors in the Wenchuan area in southwestern China, was presented and separated into training and validating dataset using a 7:3 randomly sampling ratio strategy. Based on AUC and Accuracy, GL-ResNet (0.96 and 0.909) outperformed logistic regression (0.92 and 0.851), support vector machines (0.94 and 0.884), deep belief networks (0.95 and 0.884), gated recurrent unit (0.94 and 0.884) and ResNet (0.95 and 0.894). We also explored the robustness of GL-ResNet for LSM. The results suggested that although GL-ResNet is sensitive to initial training conditions, it showed good robustness to model training method and sample ratios. In detail, GL-ResNet outperformed the conventional models in terms of fitting power and prediction performance by 0.03-0.04 and 0.02 respectively in most cases, with even greater differences in the limited training dataset.

An ensemble deep learning model for risk stratification of invasive lung adenocarcinoma using thin-slice CT

Lung cancer screening using computed tomography (CT) has increased the detection rate of small pulmonary nodules and early-stage lung adenocarcinoma. It would be clinically meaningful to accurate assessment of the nodule histology by CT scans with advanced deep learning algorithms. However, recent studies mainly focus on predicting benign and malignant nodules, lacking of model for the risk stratification of invasive adenocarcinoma. We propose an ensemble multi-view 3D convolutional neural network (EMV-3D-CNN) model to study the risk stratification of lung adenocarcinoma. We include 1075 lung nodules (≤30 mm and ≥4 mm) with preoperative thin-section CT scans and definite pathology confirmed by surgery. Our model achieves a state-of-art performance of 91.3% and 92.9% AUC for diagnosis of benign/malignant and pre-invasive/invasive nodules, respectively. Importantly, our model outperforms senior doctors in risk stratification of invasive adenocarcinoma with 77.6% accuracy [i.e., Grades 1, 2, 3]). It provides detailed predictive histological information for the surgical management of pulmonary nodules. Finally, for user-friendly access, the proposed model is implemented as a web-based system (https://seeyourlung.com.cn).

Ensemble deep learning model for optical character recognition

Ensemble deep learning model for optical character recognition

Computer-aided diagnosis using embedded ensemble deep learning for multiclass drug-resistant tuberculosis classification.

This study aims to develop a web application, TB-DRD-CXR, for the categorization of tuberculosis (TB) patients into subgroups based on their level of drug resistance. The application utilizes an ensemble deep learning model that classifies TB strains into five subtypes: drug sensitive tuberculosis (DS-TB), drug resistant TB (DR-TB), multidrug-resistant TB (MDR-TB), pre-extensively drug-resistant TB (pre-XDR-TB), and extensively drug-resistant TB (XDR-TB). The ensemble deep learning model employed in the TB-DRD-CXR web application incorporates novel fusion techniques, image segmentation, data augmentation, and various learning rate strategies. The performance of the proposed model is compared with state-of-the-art techniques and standard homogeneous CNN architectures documented in the literature. Computational results indicate that the suggested method outperforms existing methods reported in the literature, providing a 4.0%-33.9% increase in accuracy. Moreover, the proposed model demonstrates superior performance compared to standard CNN models, including DenseNet201, NASNetMobile, EfficientNetB7, EfficientNetV2B3, EfficientNetV2M, and ConvNeXtSmall, with accuracy improvements of 28.8%, 93.4%, 2.99%, 48.0%, 4.4%, and 7.6% respectively. The TB-DRD-CXR web application was developed and tested with 33 medical staff. The computational results showed a high accuracy rate of 96.7%, time-based efficiency (ET) of 4.16 goals/minutes, and an overall relative efficiency (ORE) of 100%. The system usability scale (SUS) score of the proposed application is 96.7%, indicating user satisfaction and a likelihood of recommending the TB-DRD-CXR application to others based on previous literature.

Order-up-to-level inventory optimization model using time-series demand forecasting with ensemble deep learning

Inventory control aims to meet customer demands at a given service level while minimizing cost. As a result of market volatility, customer demand is generally changing, and ignoring this uncertainty could lead to under or over-estimation of inventories resulting in shortages or inefficiencies. Inventory managers need batch ordering such that the ordered items arrive before the depletion of stocks due to the lead time between the ordering point and delivery. Therefore, to meet demand while optimizing the cost of the inventory system, firms must forecast future demands to address ordering uncertainties. Traditionally, it was challenging to predict such uncertainties with high accuracy. The availability of high volumes of historical data and big data analytics have made it easier to overcome such a challenge. This study aims to predict future demand in the case of an online retail industry using ensemble deep learning-based forecasting methods with a comparison of their performance. Compared to single-model learning, ensemble learning could improve the accuracy of predictions by combining the best performance of each model. Also, the advantages of deep learning and ensemble learning are combined in ensemble deep learning models, allowing the final model to be more generalizable. Finally, safety stocks are estimated using the forecasted demand distribution, optimizing the inventory system under a cycle service level objective.

An Ensemble Deep Learning Model for Provincial Load Forecasting Based on Reduced Dimensional Clustering and Decomposition Strategies

The accurate prediction of short-term load is crucial for the grid dispatching department in developing power generation plans, regulating unit output, and minimizing economic losses. However, due to the variability in customers’ electricity consumption behaviour and the randomness of load fluctuations, it is challenging to achieve high prediction accuracy. To address this issue, we propose an ensemble deep learning model that utilizes reduced dimensional clustering and decomposition strategies to mitigate large prediction errors caused by non-linearity and unsteadiness of load sequences. The proposed model consists of three steps: Firstly, the selected load features are dimensionally reduced using singular value decomposition (SVD), and the principal features are used for clustering different loads. Secondly, variable mode decomposition (VMD) is applied to decompose the total load of each class into intrinsic mode functions of different frequencies. Finally, an ensemble deep learning model is developed by combining the strengths of LSTM and CNN-GRU deep learning algorithms to achieve accurate load forecasting. To validate the effectiveness of our proposed model, we employ actual residential electricity load data from a province in northwest China. The results demonstrate that the proposed algorithm performs better than existing methods in terms of predictive accuracy.

Attention-Enabled Ensemble Deep Learning Models and Their Validation for Depression Detection: A Domain Adoption Paradigm

Depression is increasingly prevalent, leading to higher suicide risk. Depression detection and sentimental analysis of text inputs in cross-domain frameworks are challenging. Solo deep learning (SDL) and ensemble deep learning (EDL) models are not robust enough. Recently, attention mechanisms have been introduced in SDL. We hypothesize that attention-enabled EDL (aeEDL) architectures are superior compared to attention-not-enabled SDL (aneSDL) or aeSDL models. We designed EDL-based architectures with attention blocks to build eleven kinds of SDL model and five kinds of EDL model on four domain-specific datasets. We scientifically validated our models by comparing “seen” and “unseen” paradigms (SUP). We benchmarked our results against the SemEval (2016) sentimental dataset and established reliability tests. The mean increase in accuracy for EDL over their corresponding SDL components was 4.49%. Regarding the effect of attention block, the increase in the mean accuracy (AUC) of aeSDL over aneSDL was 2.58% (1.73%), and the increase in the mean accuracy (AUC) of aeEDL over aneEDL was 2.76% (2.80%). When comparing EDL vs. SDL for non-attention and attention, the mean aneEDL was greater than aneSDL by 4.82% (3.71%), and the mean aeEDL was greater than aeSDL by 5.06% (4.81%). For the benchmarking dataset (SemEval), the best-performing aeEDL model (ALBERT+BERT-BiLSTM) was superior to the best aeSDL (BERT-BiLSTM) model by 3.86%. Our scientific validation and robust design showed a difference of only 2.7% in SUP, thereby meeting the regulatory constraints. We validated all our hypotheses and further demonstrated that aeEDL is a very effective and generalized method for detecting symptoms of depression in cross-domain settings.

Ensemble deep learning in speech signal tasks: A review

Machine learning methods are extensively used for processing and analysing speech signals by virtue of their performance gains over multiple domains. Deep learning and ensemble learning are the two most commonly used techniques, which results in benchmark performance across different downstream tasks. Ensemble deep learning is a recent development which combines these two techniques to result in a robust architecture having substantial performance gains, as well as better generalization performance over the individual techniques. In this paper, we extensively review the use of ensemble deep learning methods for different speech signal related tasks, ranging from general objectives such as automatic speech recognition and voice activity detection, to more specific areas such as biomedical applications involving the detection of pathological speech or music genre detection. We provide a discussion on the use of different ensemble strategies such as bagging, boosting and stacking in the context of speech signals, and identify the various salient features and advantages from a broader perspective when coupled with deep learning architectures. The main objective of this study is to comprehensively evaluate existing works in the area of ensemble deep learning, and highlight the future directions that may be explored to further develop it as a tool for several speech related tasks. To the best of our knowledge, this is the first review study which primarily focuses on ensemble deep learning for speech applications. This study aims to serve as a valuable resource for researchers in academia and in industry working with speech signals, supporting advanced novel applications of ensemble deep learning models towards solving challenges in existing speech processing systems.

Brain Tumor Classification based on Improved Stacked Ensemble Deep Learning Methods.

Brain Tumor diagnostic prediction is essential for assisting radiologists and other healthcare professionals in identifying and classifying brain tumors. For the diagnosis and treatment of cancer diseases, prediction and classification accuracy are crucial. The aim of this study was to improve ensemble deep learning models for classifing brain tumor and increase the performance of structure models by combining different model of deep learning to develop a model with more accurate predictions than the individual models. Convolutional neural networks (CNNs), which are made up of a single algorithm called CNN model, are the foundation of most current methods for classifying cancer illness images. The model CNN is combined with other models to create other methods of classification called ensemble method. However, compared to a single machine learning algorithm, ensemble machine learning models are more accurate. This study used stacked ensemble deep learning technology. The data set used in this study was obtained from Kaggle and included two categories: abnormal & normal brains. The data set was trained with three models: VGG19, Inception v3, and Resnet 10. The 96.6% accuracy for binary classification (0,1) have been achieved by stacked ensemble deep learning model with Loss binary cross entropy, and Adam optimizer take into consideration with stacking models. The stacked ensemble deep learning model can be improved over a single framework.

Artificial intelligence-assisted ultrasound image analysis to discriminate early breast cancer in Chinese population: a retrospective, multicentre, cohort study.

Early diagnosis of breast cancer has always been a difficult clinical challenge. We developed a deep-learning model EDL-BC to discriminate early breast cancer with ultrasound (US) benign findings. This study aimed to investigate how the EDL-BC model could help radiologists improve the detection rate of early breast cancer while reducing misdiagnosis. In this retrospective, multicentre cohort study, we developed an ensemble deep learning model called EDL-BC based on deep convolutional neural networks. The EDL-BC model was trained and internally validated on B-mode and color Doppler US image of 7955 lesions from 6795 patients between January 1, 2015 and December 31, 2021 in the First Affiliated Hospital of Army Medical University (SW), Chongqing, China. The model was assessed by internal and external validations, and outperformed radiologists. The model performance was validated in two independent external validation cohorts included 448 lesions from 391 patients between January 1 to December 31, 2021 in the Tangshan People's Hospital (TS), Chongqing, China, and 245 lesions from 235 patients between January 1 to December 31, 2021 in the Dazu People's Hospital (DZ), Chongqing, China. All lesions in the training and total validation cohort were US benign findings during screening and biopsy-confirmed malignant, benign, and benign with 3-year follow-up records. Six radiologists performed the clinical diagnostic performance of EDL-BC, and six radiologists independently reviewed the retrospective datasets on a web-based rating platform. The area under the receiver operating characteristic curve (AUC) of the internal validation cohort and two independent external validation cohorts for EDL-BC was 0.950 (95% confidence interval [CI]: 0.909-0.969), 0.956 (95% [CI]: 0.939-0.971), and 0.907 (95% [CI]: 0.877-0.938), respectively. The sensitivity values were 94.4% (95% [CI]: 72.7%-99.9%), 100% (95% [CI]: 69.2%-100%), and 80% (95% [CI]: 28.4%-99.5%), respectively, at 0.76. The AUC for accurate diagnosis of EDL-BC (0.945 [95% [CI]: 0.933-0.965]) and radiologists with artificial intelligence (AI) assistance (0.899 [95% [CI]: 0.883-0.913]) was significantly higher than that of the radiologists without AI assistance (0.716 [95% [CI]: 0.693-0.738]; p<0.0001). Furthermore, there were no significant differences between the EDL-BC model and radiologists with AI assistance (p=0.099). EDL-BC can identify subtle but informative elements on US images of breast lesions and can significantly improve radiologists' diagnostic performance for identifying patients with early breast cancer and benefiting the clinical practice. The National Key R&D Program of China.

Predicting benefit from immune checkpoint inhibitors in patients with non-small-cell lung cancer by CT-based ensemble deep learning: a retrospective study

Only around 20-30% of patients with non-small-cell lung cancer (NCSLC) have durable benefit from immune-checkpoint inhibitors. Although tissue-based biomarkers (eg, PD-L1) are limited by suboptimal performance, tissue availability, and tumour heterogeneity, radiographic images might holistically capture the underlying cancer biology. We aimed to investigate the application of deep learning on chest CT scans to derive an imaging signature of response to immune checkpoint inhibitors and evaluate its added value in the clinical context. In this retrospective modelling study, 976 patients with metastatic, EGFR/ALK negative NSCLC treated with immune checkpoint inhibitors at MD Anderson and Stanford were enrolled from Jan 1, 2014, to Feb 29, 2020. We built and tested an ensemble deep learning model on pretreatment CTs (Deep-CT) to predict overall survival and progression-free survival after treatment with immune checkpoint inhibitors. We also evaluated the added predictive value of the Deep-CT model in the context of existing clinicopathological and radiological metrics. Our Deep-CT model demonstrated robust stratification of patient survival of the MD Anderson testing set, which was validated in the external Stanford set. The performance of the Deep-CT model remained significant on subgroup analyses stratified by PD-L1, histology, age, sex, and race. In univariate analysis, Deep-CT outperformed the conventional risk factors, including histology, smoking status, and PD-L1 expression, and remained an independent predictor after multivariate adjustment. Integrating the Deep-CT model with conventional risk factors demonstrated significantly improved prediction performance, with overall survival C-index increases from 0·70 (clinical model) to 0·75 (composite model) during testing. On the other hand, the deep learning risk scores correlated with some radiomics features, but radiomics alone could not reach the performance level of deep learning, indicating that the deep learning model effectively captured additional imaging patterns beyond known radiomics features. This proof-of-concept study shows that automated profiling of radiographic scans through deep learning can provide orthogonal information independent of existing clinicopathological biomarkers, bringing the goal of precision immunotherapy for patients with NSCLC closer. National Institutes of Health, Mark Foundation Damon Runyon Foundation Physician Scientist Award, MD Anderson Strategic Initiative Development Program, MD Anderson Lung Moon Shot Program, Andrea Mugnaini, and Edward L C Smith.

STL-decomposition ensemble deep learning models for daily reservoir inflow forecast for hydroelectricity production

Accurate reservoir inflow forecasting is crucial for efficient water management. In this study, different deep learning models, including Dense, Long short-term memory (LSTM), and one-dimensional convolutional neural networks (Conv1D), were used to build ensembles. Seasonal-trend decomposition using loess (STL) was applied to decompose reservoir inflows and precipitations into random, seasonal, and trend components. Seven ensemble models, namely STL-Dense, STL-Conv1D, STL-LSTM, STL-Dense-LSTM-Conv1D, STL-Dense multivariate, STL-LSTM multivariate, and STL-Conv1D multivariate, were proposed and evaluated using daily inflows and precipitation decomposed data from the Lom Pangar reservoir from 2015 to 2020. Evaluation metrics, such as Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and Nash Sutcliff Efficiency (NSE), were applied to assess model performance. Results showed that the STL-Dense multivariate model was the best ensemble among the thirteen models with MAE of 14.636 m3/s, RMSE of 20.841 m3/s, MAPE of 6.622%, and NSE of 0.988. These findings stress the importance of considering multiple inputs and models for accurate reservoir inflow forecasting and optimal water management. Not all ensemble models were good for Lom pangar inflow forecast as the Dense, Conv1D, and LSTM models performed better than their proposed STL monovariate ensemble models.

EDLM: Ensemble Deep Learning Model to Detect Mutation for the Early Detection of Cholangiocarcinoma.

The most common cause of mortality and disability globally right now is cholangiocarcinoma, one of the worst forms of cancer that may affect people. When cholangiocarcinoma develops, the DNA of the bile duct cells is altered. Cholangiocarcinoma claims the lives of about 7000 individuals annually. Women pass away less often than men. Asians have the greatest fatality rate. Following Whites (20%) and Asians (22%), African Americans (45%) saw the greatest increase in cholangiocarcinoma mortality between 2021 and 2022. For instance, 60-70% of cholangiocarcinoma patients have local infiltration or distant metastases, which makes them unable to receive a curative surgical procedure. Across the board, the median survival time is less than a year. Many researchers work hard to detect cholangiocarcinoma, but this is after the appearance of symptoms, which is late detection. If cholangiocarcinoma progression is detected at an earlier stage, then it will help doctors and patients in treatment. Therefore, an ensemble deep learning model (EDLM), which consists of three deep learning algorithms-long short-term model (LSTM), gated recurrent units (GRUs), and bi-directional LSTM (BLSTM)-is developed for the early identification of cholangiocarcinoma. Several tests are presented, such as a 10-fold cross-validation test (10-FCVT), an independent set test (IST), and a self-consistency test (SCT). Several statistical techniques are used to evaluate the proposed model, such as accuracy (Acc), sensitivity (Sn), specificity (Sp), and Matthew's correlation coefficient (MCC). There are 672 mutations in 45 distinct cholangiocarcinoma genes among the 516 human samples included in the proposed study. The IST has the highest Acc at 98%, outperforming all other validation approaches.

Swarm intelligence empowered three-stage ensemble deep learning for arm volume measurement in patients with lymphedema

Arm lymphedema is mostly a side effect of the breast cancer surgery and radiation therapy, which may happen after the cancer treatment. Arm volume in patients with lymphedema must be accurately measured at the time of diagnosis to assess its severity and remeasured during the treatment to evaluate response to treatment. In this paper, a swarm intelligence three-stage hetero/homogeneous ensemble deep learning model (named THENDEL), embedded in a camera scanning tool, is presented as an automatic, noninvasive, inexpensive, reliable, and high-speed tool for lymphedema arm volume measurement, suitable for daily use in clinics. THENDEL is a three-stage ensemble deep learning model comprising different deep feature extractors (first heterogeneous stage), different regressors (second heterogeneous stage), and a homogeneous stage for multiple training of each feature extractor-regressor. Hyperparameters of THENDEL (i.e., weights of heterogeneous models) are automatically tuned in an offline procedure via a swarm intelligence algorithm based on grey wolf optimizer. The THENDEL model has been successfully developed to measure the volume of 730 arms from 365 women (age, 62.3 ± 9.6 years; body mass index, 25.4 ± 3.4 kg/m2; arm volume, 2187 ± 373 mL), including healthy and patients with lymphedema. To measure the volume of an arm, it is scanned by taking two images at the same time from orthogonal directions, and then, the arm volume is estimated by the THENDEL model. Experimental results demonstrate the reliability of the proposed method by obtaining 36.65 mL mean absolute error, 1.69 % mean percent error, and 0.992 correlation between the estimated and actual arm volumes.

Ensemble learning‐based channel estimation and hybrid precoding for millimeter‐wave massive multiple input multiple output system

AbstractMillimeter‐Wave (mmWave) massive Multi‐User Multiple Input Multiple Output (MU‐MIMO) utilizes the hybrid analog precoding design helps to minimize the amount of Radio‐Frequency (RF) chains without causing loss of sum‐rate performance. The Channel Estimation (CE) is designed to demand the conditions of channels with high time‐varying features. The existing CE frameworks are extremely complex. The accurate CE is still challenging in mmWave massive MIMO system based on the hardware constraints, a maximum number of antennas, and so on. Various proficient algorithms were combined to yield reliable channel estimates. The mmWave massive MIMO system utilizes hybrid precoding to minimize energy consumption loss and reduce hardware complexity. The fully connected layer in hybrid precoding results in high energy efficiency loss. This proposed work aims to develop two ensemble deep learning models for CE and hybrid precoding of the “mmWave massive MIMO system.” Here, the estimation of the channel is performed by a “deep Convolutional Neural Network (CNN) with a Recurrent Neural Network (RNN).” With the reconstructed channel, the hybrid precoding phase is performed by deep CNN with AdaBoost. The performance of CE and hybrid precoding is improved with Self Adaptive Searched Galactic Swarm Optimization (SAS‐GSO) by attaining the maximum spectral efficiency and also minimizing the Normalized Mean‐Squared Error (NMSE). Hence, the simulation result is shown that the proposed model achieves enhanced performance. Throughout the analysis, at the 20th iteration, the SAS‐GSO‐ensemble approach gives a lower NMSE rate, which is 31%, 25.9%, 20%, 19.6%, 19.3%, 18.6%, and 18.3% superior to OMP‐Ensemble, AMP‐Ensemble, DLCS‐Ensemble, DGMP‐Ensemble, QALS‐Ensemble, DLQP‐Ensemble, and CSO‐Ensemble, respectively, for hybrid precoding. Therefore, the offered mmWave massive MIMO is achieved better performance regarding diverse error metrics.

Coupling Progressive Deep Learning with the AdaBoost Framework for Landslide Displacement Rate Prediction in the Baihetan Dam Reservoir, China

Disasters caused by landslides pose a considerable threat to people’s lives and property, resulting in substantial losses each year. Landslide displacement rate prediction (LDRP) provides a useful fundamental tool for mitigating landslide disasters. However, more accurately predicting LDRP remains a challenge in the study of landslides. Lately, ensemble deep learning algorithms have shown promise in delivering a more precise and effective spatial modeling solution. The core aims of this research are to explore and evaluate the prediction capability of three progressive evolutionary deep learning (DL) techniques, i.e., a recurrent neural network (RNN), long short-term memory (LSTM), and a gated recurrent unit (GRU) ensemble AdaBoost algorithm for modeling rainfall-induced and reservoir-induced landslides in the Baihetan reservoir area in China. The outcomes show that the ensemble DL model could predict the Wangjiashan landslide in the Baihetan reservoir area with improved accuracy. The highest accuracy was achieved in the testing set when the window length equaled 30. However, assembling two predictors outperformed the accuracy of assembling three predictors, with the mean absolute error and root mean square error reaching 1.019 and 1.300, respectively. These findings suggest that the combination of strong learners and DL can yield satisfactory prediction results.

MSENet: Mean and standard deviation based ensemble network for cervical cancer detection

Cervical cancer is one of the most concerning carcinogenic diseases among women worldwide. The condition is especially bad in low- or middle-income countries due to the lack of medical facilities. In such situations, computer-aided diagnosis (CAD) systems can alleviate the need to a large extent. However, sometimes a single learning model may not be effective enough to capture relevant information for accurate prediction of diseases from complex data. To this end, we propose an ensemble of deep learning models, called Mean and Standard Deviation-based Ensemble Network (MSENet), for detecting cervical cancer from Pap smear images. Our ensemble model consists of three standard base classifiers, namely Xception, Inception V3, and VGG-16. We further improve the classification abilities of these models by implementing a novel probability enhancement scheme that takes into account the mean and standard deviation of the confidence scores. This technique enables the overall framework to capture complementary information offered by the base classifiers and is tailored to the characteristics of deep learners. Finally, we use the product rule to aggregate the obtained outcomes and get final predictions. The proposed MSENet has been evaluated on a standard public benchmark dataset, called SIPaKMeD. With a classification accuracy of 97.21% using a 5-fold cross-validation scheme, the MSENet outperforms many state-of-the-art methods. The source codes are made public in the following Github repository https://github.com/rishavpramanik/msenet.

Correction of sub-seasonal predictions of summer precipitation in Southwest China based on the Transformer-Seq2Seq-DNN ensemble deep learning model

Correction of sub-seasonal predictions of summer precipitation in Southwest China based on the Transformer-Seq2Seq-DNN ensemble deep learning model

Automated Triage of Screening Breast MRI Examinations in High-Risk Women Using an Ensemble Deep Learning Model.

The aim of the study is to develop and evaluate the performance of a deep learning (DL) model to triage breast magnetic resonance imaging (MRI) findings in high-risk patients without missing any cancers. In this retrospective study, 16,535 consecutive contrast-enhanced MRIs performed in 8354 women from January 2013 to January 2019 were collected. From 3 New York imaging sites, 14,768 MRIs were used for the training and validation data set, and 80 randomly selected MRIs were used for a reader study test data set. From 3 New Jersey imaging sites, 1687 MRIs (1441 screening MRIs and 246 MRIs performed in recently diagnosed breast cancer patients) were used for an external validation data set. The DL model was trained to classify maximum intensity projection images as "extremely low suspicion" or "possibly suspicious." Deep learning model evaluation (workload reduction, sensitivity, specificity) was performed on the external validation data set, using a histopathology reference standard. A reader study was performed to compare DL model performance to fellowship-trained breast imaging radiologists. In the external validation data set, the DL model triaged 159/1441 of screening MRIs as "extremely low suspicion" without missing a single cancer, yielding a workload reduction of 11%, a specificity of 11.5%, and a sensitivity of 100%. The model correctly triaged 246/246 (100% sensitivity) of MRIs in recently diagnosed patients as "possibly suspicious." In the reader study, 2 readers classified MRIs with a specificity of 93.62% and 91.49%, respectively, and missed 0 and 1 cancer, respectively. On the other hand, the DL model classified MRIs with a specificity of 19.15% and missed 0 cancers, highlighting its potential use not as an independent reader but as a triage tool. Our automated DL model triages a subset of screening breast MRIs as "extremely low suspicion" without misclassifying any cancer cases. This tool may be used to reduce workload in standalone mode, to shunt low suspicion cases to designated radiologists or to the end of the workday, or to serve as base model for other downstream AI tools.

Risk factor refinement and ensemble deep learning methods on prediction of heart failure using real healthcare records

The prediction of heart failure (HF) is crucial in preventing disease progression by implementing lifestyle changes and pharmacological interventions before the onset of heart diseases. While there have been numerous attempts to predict HF, many have failed to consider the coexisting risk factors and their complex relationships with one another. In this research paper, we present an early warning and prediction method for HF using deep learning approaches. Our proposed method involves a risk factor selection method to identify significant risk factors that contain relevant and valuable information for HF prediction. Additionally, we present an anomaly detection method to eliminate abnormal data that may be caused by mood changes or environmental factors. Finally, we propose an ensemble deep learning model for HF prediction based on scalable conjugate-gradient concept and back propagation learning algorithm that aims to predict and provide early warning of HF in massive medical data. We evaluate our proposed method based on our real research project, HeartCarer, and achieve an accuracy of 98.5%, which surpasses other state-of-the-art methods and our prior work (90%).