Hold-out Test Data Research Articles

Machine Learning based Pointing Corrections for Radio/Sub-millimeter Telescopes

Radio, sub-millimeter and millimeter ground-based telescopes are powerful instruments for studying the gas and dust-rich regions of the Universe that are invisible at optical wavelengths, but the pointing accuracy is crucial for obtaining high-quality data. Pointing errors are small deviations of the telescope’s orientation from its desired direction. The telescopes use linear regression pointing models to correct for these errors, taking into account various factors such as weather conditions, telescope mechanical structure, and the target’s position in the sky. However, residual pointing errors can still occur due to factors that are hard to model accurately, such as thermal and gravitational deformation and environmental conditions like humidity and wind. Here we present a proof-of-concept for reducing pointing error for the Atacama Pathfinder EXperiment (APEX) telescope in the high-altitude Atacama Desert in Chile based on machine learning. Using historic pointing data from 2022, we trained eXtreme Gradient Boosting (XGBoost) models that reduced the root-mean-square errors (RMSE) for azimuth and elevation (horizontal and vertical angle) pointing corrections by 4.3% and 9.5%, respectively, on hold-out test data. Our results will inform operations of current and future facilities such as the next-generation Atacama Large Aperture Submillimeter Telescope (AtLAST)

Computational staining of CD3/CD20 positive lymphocytes in human tissues with experimental confirmation in a genetically engineered mouse model.

Immune dysregulation plays a major role in cancer progression. The quantification of lymphocytic spatial inflammation may enable spatial system biology, improve understanding of therapeutic resistance, and contribute to prognostic imaging biomarkers. In this paper, we propose a knowledge-guided deep learning framework to measure the lymphocytic spatial architecture on human H&E tissue, where the fidelity of training labels is maximized through single-cell resolution image registration of H&E to IHC. We demonstrate that such an approach enables pixel-perfect ground-truth labeling of lymphocytes on H&E as measured by IHC. We then experimentally validate our technique in a genetically engineered, immune-compromised Rag2 mouse model, where Rag2 knockout mice lacking mature lymphocytes are used as a negative experimental control. Such experimental validation moves beyond the classical statistical testing of deep learning models and demonstrates feasibility of more rigorous validation strategies that integrate computational science and basic science. Using our developed approach, we automatically annotated more than 111,000 human nuclei (45,611 CD3/CD20 positive lymphocytes) on H&E images to develop our model, which achieved an AUC of 0.78 and 0.71 on internal hold-out testing data and external testing on an independent dataset, respectively. As a measure of the global spatial architecture of the lymphocytic microenvironment, the average structural similarity between predicted lymphocytic density maps and ground truth lymphocytic density maps was 0.86 ± 0.06 on testing data. On experimental mouse model validation, we measured a lymphocytic density of 96.5 ± %1% in a Rag2+/- control mouse, compared to an average of 16.2 ± %5% in Rag2-/- immune knockout mice (p<0.0001, ANOVA-test). These results demonstrate that CD3/CD20 positive lymphocytes can be accurately detected and characterized on H&E by deep learning and generalized across species. Collectively, these data suggest that our understanding of complex biological systems may benefit from computationally-derived spatial analysis, as well as integration of computational science and basic science.

Machine learning and explainable artificial intelligence to predict pathologic stage in men with localized prostate cancer.

Though several nomograms exist, machine learning (ML) approaches might improve prediction of pathologic stage in patients with prostate cancer. To develop ML models to predict pathologic stage that outperform existing nomograms that use readily available clinicopathologic variables. Patients with prostate adenocarcinoma who underwent surgery were identified in the National Cancer Database. Seven ML models were trained to predict organ-confined (OC) disease, extracapsular extension, seminal vesicle invasion (SVI), and lymph node involvement (LNI). Model performance was measured using area under the curve (AUC) on a holdout testing data set. Clinical utility was evaluated using decision curve analysis (DCA). Performance metrics were confirmed on an external validation data set. The ML-based extreme gradient boosted trees model achieved the best performance with an AUC of 0.744, 0.749, 0.816, 0.811 for the OC, ECE, SVI, and LNI models, respectively. The MSK nomograms achieved an AUC of 0.708, 0.742, 0.806, 0.802 for the OC, ECE, SVI, and LNI models, respectively. These models also performed the best on DCA. Findings were consistent on both a holdout internal validation data set as well as an external validation data set. Our ML models better predicted pathologic stage relative to existing nomograms at predicting pathologic stage. Accurate prediction of pathologic stage can help oncologists and patients determine optimal definitive treatment options for patients with prostate cancer.

Learning epistatic polygenic phenotypes with Boolean interactions.

Detecting epistatic drivers of human phenotypes is a considerable challenge. Traditional approaches use regression to sequentially test multiplicative interaction terms involving pairs of genetic variants. For higher-order interactions and genome-wide large-scale data, this strategy is computationally intractable. Moreover, multiplicative terms used in regression modeling may not capture the form of biological interactions. Building on the Predictability, Computability, Stability (PCS) framework, we introduce the epiTree pipeline to extract higher-order interactions from genomic data using tree-based models. The epiTree pipeline first selects a set of variants derived from tissue-specific estimates of gene expression. Next, it uses iterative random forests (iRF) to search training data for candidate Boolean interactions (pairwise and higher-order). We derive significance tests for interactions, based on a stabilized likelihood ratio test, by simulating Boolean tree-structured null (no epistasis) and alternative (epistasis) distributions on hold-out test data. Finally, our pipeline computes PCS epistasis p-values that probabilisticly quantify improvement in prediction accuracy via bootstrap sampling on the test set. We validate the epiTree pipeline in two case studies using data from the UK Biobank: predicting red hair and multiple sclerosis (MS). In the case of predicting red hair, epiTree recovers known epistatic interactions surrounding MC1R and novel interactions, representing non-linearities not captured by logistic regression models. In the case of predicting MS, a more complex phenotype than red hair, epiTree rankings prioritize novel interactions surrounding HLA-DRB1, a variant previously associated with MS in several populations. Taken together, these results highlight the potential for epiTree rankings to help reduce the design space for follow up experiments.

Machine learning models to evaluate mortality in pediatric patients with pneumonia in the intensive care unit.

This study aimed to predict mortality in children with pneumonia who were admitted to the intensive care unit (ICU) to aid decision-making. Retrospective cohort study conducted at a single tertiary hospital. This study included children who were admitted to the pediatric ICU at the National Taiwan University Hospital between 2010 and 2019 due to pneumonia. Two prediction models were developed using tree-structured machine learning algorithms. The primary outcomes were ICU mortality and 24-h ICU mortality. A total of 33 features, including demographics, underlying diseases, vital signs, and laboratory data, were collected from the electronic health records. The machine learning models were constructed using the development data set, and performance matrices were computed using the holdout test data set. A total of 1231 ICU admissions of children with pneumonia were included in the final cohort. The area under the receiver operating characteristic curves (AUROCs) of the ICU mortality model and 24-h ICU mortality models was 0.80 (95% confidence interval [CI], 0.69-0.91) and 0.92 (95% CI, 0.86-0.92), respectively. Based on feature importance, the model developed in this study tended to predict increased mortality for the subsequent 24 h if a reduction in the blood pressure, peripheral capillary oxygen saturation (SpO2), or higher partial pressure of carbon dioxide (PCO2) were observed. This study demonstrated that the machine learning models for predicting ICU mortality and 24-h ICU mortality in children with pneumonia have the potential to support decision-making, especially in resource-limited settings.

Automated inversion time selection for late gadolinium–enhanced cardiac magnetic resonance imaging

ObjectivesTo develop and share a deep learning method that can accurately identify optimal inversion time (TI) from multi-vendor, multi-institutional and multi-field strength inversion scout (TI scout) sequences for late gadolinium enhancement cardiac MRI.Materials and methodsRetrospective multicentre study conducted on 1136 1.5-T and 3-T cardiac MRI examinations from four centres and three scanner vendors. Deep learning models, comprising a convolutional neural network (CNN) that provides input to a long short-term memory (LSTM) network, were trained on TI scout pixel data from centres 1 to 3 to identify optimal TI, using ground truth annotations by two readers. Accuracy within 50 ms, mean absolute error (MAE), Lin’s concordance coefficient (LCCC) and reduced major axis regression (RMAR) were used to select the best model from validation results, and applied to holdout test data. Robustness of the best-performing model was also tested on imaging data from centre 4.ResultsThe best model (SE-ResNet18-LSTM) produced accuracy of 96.1%, MAE 22.9 ms and LCCC 0.47 compared to ground truth on the holdout test set and accuracy of 97.3%, MAE 15.2 ms and LCCC 0.64 when tested on unseen external (centre 4) data. Differences in vendor performance were observed, with greatest accuracy for the most commonly represented vendor in the training data.ConclusionA deep learning model was developed that can identify optimal inversion time from TI scout images on multi-vendor data with high accuracy, including on previously unseen external data. We make this model available to the scientific community for further assessment or development.Clinical relevance statementA robust automated inversion time selection tool for late gadolinium–enhanced imaging allows for reproducible and efficient cross-vendor inversion time selection.Key Points• A model comprising convolutional and recurrent neural networks was developed to extract optimal TI from TI scout images.• Model accuracy within 50 ms of ground truth on multi-vendor holdout and external data of 96.1% and 97.3% respectively was achieved.• This model could improve workflow efficiency and standardise optimal TI selection for consistent LGE imaging.

Defining a screening tool for post-traumatic stress disorder in East Africa: a penalized regression approach.

Scalable PTSD screening strategies must be brief, accurate and capable of administration by a non-specialized workforce. We used PTSD as determined by the structured clinical interview as our gold standard and considered predictors sets of (a) Posttraumatic Stress Checklist-5 (PCL-5), (b) Primary Care PTSD Screen for the DSM-5 (PC-PTSD) and, (c) PCL-5 and PC-PTSD questions to identify the optimal items for PTSD screening for public sector settings in Kenya. A logistic regression model using LASSO was fit by minimizing the average squared error in the validation data. Area under the receiver operating characteristic curve (AUROC) measured discrimination performance. Penalized regression analysis suggested a screening tool that sums the Likert scale values of two PCL-5 questions-intrusive thoughts of the stressful experience (#1) and insomnia (#21). This had an AUROC of 0.85 (using hold-out test data) for predicting PTSD as evaluated by the MINI, which outperformed the PC-PTSD. The AUROC was similar in subgroups defined by age, sex, and number of categories of trauma experienced (all AUROCs>0.83) except those with no trauma history- AUROC was 0.78. In some East African settings, a 2-item PTSD screening tool may outperform longer screeners and is easily scaled by a non-specialist workforce.

A Machine Learning Algorithm Predicting Acute Kidney Injury in Intensive Care Unit Patients (NAVOY Acute Kidney Injury): Proof-of-Concept Study.

Acute kidney injury (AKI) represents a significant global health challenge, leading to increased patient distress and financial health care burdens. The development of AKI in intensive care unit (ICU) settings is linked to prolonged ICU stays, a heightened risk of long-term renal dysfunction, and elevated short- and long-term mortality rates. The current diagnostic approach for AKI is based on late indicators, such as elevated serum creatinine and decreased urine output, which can only detect AKI after renal injury has transpired. There are no treatments to reverse or restore renal function once AKI has developed, other than supportive care. Early prediction of AKI enables proactive management and may improve patient outcomes. The primary aim was to develop a machine learning algorithm, NAVOY Acute Kidney Injury, capable of predicting the onset of AKI in ICU patients using data routinely collected in ICU electronic health records. The ultimate goal was to create a clinical decision support tool that empowers ICU clinicians to proactively manage AKI and, consequently, enhance patient outcomes. We developed the NAVOY Acute Kidney Injury algorithm using a hybrid ensemble model, which combines the strengths of both a Random Forest (Leo Breiman and Adele Cutler) and an XGBoost model (Tianqi Chen). To ensure the accuracy of predictions, the algorithm used 22 clinical variables for hourly predictions of AKI as defined by the Kidney Disease: Improving Global Outcomes guidelines. Data for algorithm development were sourced from the Massachusetts Institute of Technology Lab for Computational Physiology Medical Information Mart for Intensive Care IV clinical database, focusing on ICU patients aged 18 years or older. The developed algorithm, NAVOY Acute Kidney Injury, uses 4 hours of input and can, with high accuracy, predict patients with a high risk of developing AKI 12 hours before onset. The prediction performance compares well with previously published prediction algorithms designed to predict AKI onset in accordance with Kidney Disease: Improving Global Outcomes diagnosis criteria, with an impressive area under the receiver operating characteristics curve (AUROC) of 0.91 and an area under the precision-recall curve (AUPRC) of 0.75. The algorithm's predictive performance was externally validated on an independent hold-out test data set, confirming its ability to predict AKI with exceptional accuracy. NAVOY Acute Kidney Injury is an important development in the field of critical care medicine. It offers the ability to predict the onset of AKI with high accuracy using only 4 hours of data routinely collected in ICU electronic health records. This early detection capability has the potential to strengthen patient monitoring and management, ultimately leading to improved patient outcomes. Furthermore, NAVOY Acute Kidney Injury has been granted Conformite Europeenne (CE)-marking, marking a significant milestone as the first CE-marked AKI prediction algorithm for commercial use in European ICUs.

Deep learning uncertainty quantification for clinical text classification

Introduction:Machine learning algorithms are expected to work side-by-side with humans in decision-making pipelines. Thus, the ability of classifiers to make reliable decisions is of paramount importance. Deep neural networks (DNNs) represent the state-of-the-art models to address real-world classification. Although the strength of activation in DNNs is often correlated with the network’s confidence, in-depth analyses are needed to establish whether they are well calibrated. Method:In this paper, we demonstrate the use of DNN-based classification tools to benefit cancer registries by automating information extraction of disease at diagnosis and at surgery from electronic text pathology reports from the US National Cancer Institute (NCI) Surveillance, Epidemiology, and End Results (SEER) population-based cancer registries. In particular, we introduce multiple methods for selective classification to achieve a target level of accuracy on multiple classification tasks while minimizing the rejection amount—that is, the number of electronic pathology reports for which the model’s predictions are unreliable. We evaluate the proposed methods by comparing our approach with the current in-house deep learning-based abstaining classifier. Results:Overall, all the proposed selective classification methods effectively allow for achieving the targeted level of accuracy or higher in a trade-off analysis aimed to minimize the rejection rate. On in-distribution validation and holdout test data, with all the proposed methods, we achieve on all tasks the required target level of accuracy with a lower rejection rate than the deep abstaining classifier (DAC). Interpreting the results for the out-of-distribution test data is more complex; nevertheless, in this case as well, the rejection rate from the best among the proposed methods achieving 97% accuracy or higher is lower than the rejection rate based on the DAC. Conclusions:We show that although both approaches can flag those samples that should be manually reviewed and labeled by human annotators, the newly proposed methods retain a larger fraction and do so without retraining—thus offering a reduced computational cost compared with the in-house deep learning-based abstaining classifier.

Machine learning based readmission and mortality prediction in heart failure patients

This study intends to predict in-hospital and 6-month mortality, as well as 30-day and 90-day hospital readmission, using Machine Learning (ML) approach via conventional features. A total of 737 patients remained after applying the exclusion criteria to 1101 heart failure patients. Thirty-four conventional features were collected for each patient. First, the data were divided into train and test cohorts with a 70–30% ratio. Then train data were normalized using the Z-score method, and its mean and standard deviation were applied to the test data. Subsequently, Boruta, RFE, and MRMR feature selection methods were utilized to select more important features in the training set. In the next step, eight ML approaches were used for modeling. Next, hyperparameters were optimized using tenfold cross-validation and grid search in the train dataset. All model development steps (normalization, feature selection, and hyperparameter optimization) were performed on a train set without touching the hold-out test set. Then, bootstrapping was done 1000 times on the hold-out test data. Finally, the obtained results were evaluated using four metrics: area under the ROC curve (AUC), accuracy (ACC), specificity (SPE), and sensitivity (SEN). The RFE-LR (AUC: 0.91, ACC: 0.84, SPE: 0.84, SEN: 0.83) and Boruta-LR (AUC: 0.90, ACC: 0.85, SPE: 0.85, SEN: 0.83) models generated the best results in terms of in-hospital mortality. In terms of 30-day rehospitalization, Boruta-SVM (AUC: 0.73, ACC: 0.81, SPE: 0.85, SEN: 0.50) and MRMR-LR (AUC: 0.71, ACC: 0.68, SPE: 0.69, SEN: 0.63) models performed the best. The best model for 3-month rehospitalization was MRMR-KNN (AUC: 0.60, ACC: 0.63, SPE: 0.66, SEN: 0.53) and regarding 6-month mortality, the MRMR-LR (AUC: 0.61, ACC: 0.63, SPE: 0.44, SEN: 0.66) and MRMR-NB (AUC: 0.59, ACC: 0.61, SPE: 0.48, SEN: 0.63) models outperformed the others. Reliable models were developed in 30-day rehospitalization and in-hospital mortality using conventional features and ML techniques. Such models can effectively personalize treatment, decision-making, and wiser budget allocation. Obtained results in 3-month rehospitalization and 6-month mortality endpoints were not astonishing and further experiments with additional information are needed to fetch promising results in these endpoints.

Predicting habitual water intake from lifestyle questions.

Previous studies have used selective recall and descriptive dietary record methods, requiring considerable effort for assessing food and water intake. This study created a simplified lifestyle questionnaire to predict habitual water intake (SQW), accurately and quickly assessing the habitual water intake. We also evaluated the validity using descriptive dietary records as a cross-sectional study. First, we used crowdsourcing and machine learning to collect data, predict water intake records, and create questionnaires. We collected 305 lifestyle-related questions as predictor variables and selective recall methods for assessing water intake as an outcome variable. Random forests were used for the machine learning models because of their interpretability and accurate estimation. Random forest and single regression correlation analysis were augmented by the synthetic minority oversampling that trained the model. We separated the data by sex and evaluated our model using unseen hold-out testing data, predicting the individual and overall habitual water intake from various sources, including non-alcoholic beverages, alcohol, and food. We found a 0.60 Spearman's correlation coefficient for total water intake between the predicted and the selective recall method values, reflecting the target value to be achieved. This question set was then used for feasibility tests. The descriptive dietary record method helped to obtain a ground-truth value. We categorized the data by gender, season, and source: non-alcoholic beverages, alcohol, food, and total water intake, and the correlation was confirmed. Consequently, our results showed a Pearson's correlation coefficient of 0.50 for total water intake between the predicted and the selective recall method values. We hypothesize that dissemination of SQW can lead to better health management by easily determining the habitual water intake.

Fusion of Multiple Spectra for Investigating Chemical Bonding Properties via Machine Learning.

Chemical bonding properties are crucial to understanding the chemical behavior of molecules. Spectroscopy is a versatile technical tool to study various microscopic properties, but its interpretation suffers from human biases and the loss of high-dimensional information. Here, we present a machine learning approach to predict diverse bonding properties, including the bond dissociation energy, bond length, and α-C connectivity of hydroxyls in organic molecules, by fusing multiple spectra with different physical mechanisms. Combining nuclear magnetic resonance and vibrational spectroscopy exhibits higher prediction accuracy than what they did separately. On the hold-out test data set, the models achieve a mean absolute error of 1.243 kcal/mol and 1.041 × 10-4 Å for BDE and bond length and an accuracy of 95.09% for hydroxyl α-C connectivity. Our models demonstrate strong extrapolation capabilities when they are transferred to different molecules, external electric fields, and solvation environments. These end-to-end models pave the way to investigating chemical bonding properties by using spectroscopic observables.

SUVR quantification using attention-based 3D CNN with longitudinal Florbetapir PET images in Alzheimer’s disease

PurposeFlorbetapir PET images provide valuable information about the amount of amyloid deposition in the brain due to neurodegenerative diseases, which helps in the prognosis of patients. The purpose of this study is to develop a system that helps in the automatic amyloid quantification of the standard uptake value ratio so that drug treatments could be effectively determined. Methods2647 Florbetapir PET images obtained from multiple centres of Alzheimer Disease Neuroimaging Initiative (ADNI) and an external dataset of 1413 scans from the Anti-Amyloid Treatment in Asymptomatic Alzheimer’s study are used to design and test the proposed 3D CNN attention-based model to quantify the amyloid deposits. Only 80% of scans from the ADNI dataset are used to train the model. The remaining 20% of scans from ADNI and the external dataset are used for testing the trained model. ResultsThe proposed model achieves a root mean square error of 0.0362 and a mean absolute error of 0.026 on separate hold-out test data from ADNI and a root mean square error of 0.058 and a mean absolute error of 0.044 Anti-Amyloid Treatment in Asymptomatic Alzheimer’s study dataset. A graphical user interface is developed for the proposed model which will display a slice of the Florbetapir PET volume and its predicted standard uptake value ratio. Conclusion3D CNN architecture with both spatial and channel attention provided better results when compared to models without attention. The proposed model proves to be an efficient tool in the automatic amyloid standard uptake value ratio quantification.

Assessing wind field characteristics along the airport runway glide slope: an explainable boosting machine-assisted wind tunnel study

Aircraft landings are especially perilous when the wind is gusty near airport runways. For this reason, an aircraft may deviate from its glide slope, miss its approach, or even crash in the worst cases. In the study, we used the state-of-the-art glass-box model, the Explainable Boosting Machine (EBM), to estimate the variation in headwind speed and turbulence intensity along the airport runway glide slope and to interpret the various contributing factors. To begin, the wind field characteristics were examined by developing a scaled-down model of Hong Kong International Airport (HKIA) runway as well as and the surrounding buildings and complex terrain in the TJ-3 atmospheric boundary layer wind tunnel. The placement of probes along the glide slope of the model runway aided in the measurement of wind field characteristics at different locations in the presence and absence of surrounding buildings. Next, the experimental data was used to train the EBM model in conjunction with Bayesian optimization approach. The counterpart black box models (extreme gradient boosting, random forest, extra tree and adaptive boosting) as well as other glass box models (linear regression and decision tree) were compared with the outcomes of the EBM model. Based on the holdout testing data, the EBM model revealed superior performance for both variation in headwind speed and turbulence intensity in terms of mean absolute error, mean squared error, root mean squared error and R-square values. To further evaluate the impact of different factors on the wind field characteristics along the airport runway glide slope, the EBM model allows for a full interpretation of the contribution of individual and pairwise interactions of factors to the prediction results from both a global and a local perspective.

Preclinical evaluation of Raman spectroscopy for pedicular screw insertion surgical guidance in a porcine spine model.

Orthopedic surgery is frequently performed but currently lacks consensus and availability of ideal guidance methods, resulting in high variability of outcomes. Misdirected insertion of surgical instruments can lead to weak anchorage and unreliable fixation along with risk to critical structures including the spinal cord. Current methods for surgical guidance using conventional medical imaging are indirect and time-consuming with unclear advantages. The purpose of this study was to investigate the potential of intraoperative in situ near-infrared Raman spectroscopy (RS) combined with machine learning in guiding pedicular screw insertion in the spine. A portable system equipped with a hand-held RS probe was used to make fingerprint measurements on freshly excised porcine vertebrae, identifying six tissue types: bone, spinal cord, fat, cartilage, ligament, and muscle. Supervised machine learning techniques were used to train-and test on independent hold-out data subsets-a six-class model as well as two-class models engineered to distinguish bone from soft tissue. The two-class models were further tested using in vivo spectral fingerprint measurements made during intra-pedicular drilling in a porcine spine model. The five-class model achieved accuracy in distinguish all six tissue classes when applied onto a hold-out testing data subset. The binary classifier detecting bone versus soft tissue (all soft tissue or spinal cord only) yielded 100% accuracy. When applied onto in vivo measurements performed during interpedicular drilling, the soft tissue detection models correctly detected all spinal canal breaches. We provide a foundation for RS in the orthopedic surgical guidance field. It shows that RS combined with machine learning is a rapid and accurate modality capable of discriminating tissues that are typically encountered in orthopedic procedures, including pedicle screw placement. Future development of integrated RS probes and surgical instruments promises better guidance options for the orthopedic surgeon and better patient outcomes.

Evaluating the Ability of Open-Source Artificial Intelligence to Predict Accepting-Journal Impact Factor and Eigenfactor Score Using Academic Article Abstracts: Cross-sectional Machine Learning Analysis

Strategies to improve the selection of appropriate target journals may reduce delays in disseminating research results. Machine learning is increasingly used in content-based recommender algorithms to guide journal submissions for academic articles. We sought to evaluate the performance of open-source artificial intelligence to predict the impact factor or Eigenfactor score tertile using academic article abstracts. PubMed-indexed articles published between 2016 and 2021 were identified with the Medical Subject Headings (MeSH) terms "ophthalmology," "radiology," and "neurology." Journals, titles, abstracts, author lists, and MeSH terms were collected. Journal impact factor and Eigenfactor scores were sourced from the 2020 Clarivate Journal Citation Report. The journals included in the study were allocated percentile ranks based on impact factor and Eigenfactor scores, compared with other journals that released publications in the same year. All abstracts were preprocessed, which included the removal of the abstract structure, and combined with titles, authors, and MeSH terms as a single input. The input data underwent preprocessing with the inbuilt ktrain Bidirectional Encoder Representations from Transformers (BERT) preprocessing library before analysis with BERT. Before use for logistic regression and XGBoost models, the input data underwent punctuation removal, negation detection, stemming, and conversion into a term frequency-inverse document frequency array. Following this preprocessing, data were randomly split into training and testing data sets with a 3:1 train:test ratio. Models were developed to predict whether a given article would be published in a first, second, or third tertile journal (0-33rd centile, 34th-66th centile, or 67th-100th centile), as ranked either by impact factor or Eigenfactor score. BERT, XGBoost, and logistic regression models were developed on the training data set before evaluation on the hold-out test data set. The primary outcome was overall classification accuracy for the best-performing model in the prediction of accepting journal impact factor tertile. There were 10,813 articles from 382 unique journals. The median impact factor and Eigenfactor score were 2.117 (IQR 1.102-2.622) and 0.00247 (IQR 0.00105-0.03), respectively. The BERT model achieved the highest impact factor tertile classification accuracy of 75.0%, followed by an accuracy of 71.6% for XGBoost and 65.4% for logistic regression. Similarly, BERT achieved the highest Eigenfactor score tertile classification accuracy of 73.6%, followed by an accuracy of 71.8% for XGBoost and 65.3% for logistic regression. Open-source artificial intelligence can predict the impact factor and Eigenfactor score of accepting peer-reviewed journals. Further studies are required to examine the effect on publication success and the time-to-publication of such recommender systems.

GTAR: a new ensemble evolutionary autoregressive approach to model dissolved organic carbon

AbstractThis article explores the forecasting capabilities of three classic linear and nonlinear autoregressive modeling techniques and proposes a new ensemble evolutionary time series approach to model and forecast daily dynamics in stream dissolved organic carbon (DOC). The model used data from the Oulankajoki River basin, a boreal catchment in Northern Finland. The models that were evolved used both accuracy and parsimony measures including autoregressive (AR), vector autoregressive (VAR), and self-exciting threshold autoregressive (SETAR). The new method, called genetic-based SETAR (GTAR), evolved through the integration of state-of-the-art genetic programming with SETAR. To develop the models, high-resolution DOC concentration and daily streamflow (as the external input for VAR) were measured at the same gauging station throughout the ice free season. The results showed that all the models characterize the DOC dynamics with an acceptable 1-day-ahead forecasting accuracy. Use of the streamflow time series as an exogenous variable did not increase the predictive accuracy of AR models. Moreover, the hybrid GTAR provided the best accuracy for the holdout testing data and proved to be a suitable approach for predicting DOC in boreal conditions.

Predicting the prevalence of complex genetic diseases from individual genotype profiles using capsule networks

Diseases that have a complex genetic architecture tend to suffer from considerable amounts of genetic variants that, although playing a role in the disease, have not yet been revealed as such. Two major causes for this phenomenon are genetic variants that do not stack up effects, but interact in complex ways; in addition, as recently suggested, the omnigenic model postulates that variants interact in a holistic manner to establish disease phenotypes. Here we present DiseaseCapsule, as a capsule-network-based approach that explicitly addresses to capture the hierarchical structure of the underlying genome data, and has the potential to fully capture the non-linear relationships between variants and disease. DiseaseCapsule is the first such approach to operate in a whole-genome manner when predicting disease occurrence from individual genotype profiles. In experiments, we evaluated DiseaseCapsule on amyotrophic lateral sclerosis (ALS) and Parkinson’s disease, with a particular emphasis on ALS, which is known to have a complex genetic architecture and is affected by 40% missing heritability. On ALS, DiseaseCapsule achieves 86.9% accuracy on hold-out test data in predicting disease occurrence, thereby outperforming all other approaches by large margins. Also, DiseaseCapsule required sufficiently less training data for reaching optimal performance. Last but not least, the systematic exploitation of the network architecture yielded 922 genes of particular interest, and 644 ‘non-additive’ genes that are crucial factors in DiseaseCapsule, but remain masked within linear schemes.

Image based deep learning in 12-lead ECG diagnosis.

The electrocardiogram is an integral tool in the diagnosis of cardiovascular disease. Most studies on machine learning classification of electrocardiogram (ECG) diagnoses focus on processing raw signal data rather than ECG images. This presents a challenge for models in many areas of clinical practice where ECGs are printed on paper or only digital images are accessible, especially in remote and regional settings. This study aims to evaluate the accuracy of image based deep learning algorithms on 12-lead ECG diagnosis. Deep learning models using VGG architecture were trained on various 12-lead ECG datasets and evaluated for accuracy by testing on holdout test data as well as data from datasets not seen in training. Grad-CAM was utilized to depict heatmaps of diagnosis. The results demonstrated excellent AUROC, AUPRC, sensitivity and specificity on holdout test data from datasets used in training comparable to the best signal and image-based models. Detection of hidden characteristics such as gender were achieved at a high rate while Grad-CAM successfully highlight pertinent features on ECGs traditionally used by human interpreters. This study demonstrates feasibility of image based deep learning algorithms in ECG diagnosis and identifies directions for future research in order to develop clinically applicable image based deep-learning models in ECG diagnosis.

Nestedcv: an R package for fast implementation of nested cross-validation with embedded feature selection designed for transcriptomics and high-dimensional data.

Although machine learning models are commonly used in medical research, many analyses implement a simple partition into training data and hold-out test data, with cross-validation (CV) for tuning of model hyperparameters. Nested CV with embedded feature selection is especially suited to biomedical data where the sample size is frequently limited, but the number of predictors may be significantly larger (P ≫ n). The nestedcv R package implements fully nested k × l-fold CV for lasso and elastic-net regularized linear models via the glmnet package and supports a large array of other machine learning models via the caret framework. Inner CV is used to tune models and outer CV is used to determine model performance without bias. Fast filter functions for feature selection are provided and the package ensures that filters are nested within the outer CV loop to avoid information leakage from performance test sets. Measurement of performance by outer CV is also used to implement Bayesian linear and logistic regression models using the horseshoe prior over parameters to encourage a sparse model and determine unbiased model accuracy. The R package nestedcv is available from CRAN: https://CRAN.R-project.org/package=nestedcv.