OpenBioLink: a benchmarking framework for large-scale biomedical link prediction.

SUMMARY: Recently, novel machine-learning algorithms have shown potential for predicting undiscovered links in biomedical knowledge networks. However, dedicated benchmarks for measuring algorithmic progress have not yet emerged. With OpenBioLink, we introduce a large-scale, high-quality and highly challenging biomedical link prediction benchmark to transparently and reproducibly evaluate such algorithms. Furthermore, we present preliminary baseline evaluation results. AVAILABILITY AND IMPLEMENTATION: Source code and data are openly available at https://github.com/OpenBioLink/OpenBioLink. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

k-Space-based coil combination via geometric deep learning for reconstruction of non-Cartesian MRSI data.

PURPOSE: State-of-the-art whole-brain MRSI with spatial-spectral encoding and multichannel acquisition generates huge amounts of data, which must be efficiently processed to stay within reasonable reconstruction times. Although coil combination significantly reduces the amount of data, currently it is performed in image space at the end of the reconstruction. This prolongs reconstruction times and increases RAM requirements. We propose an alternative k-space-based coil combination that uses geometric deep learning to combine MRSI data already in native non-Cartesian k-space. METHODS: Twelve volunteers were scanned at a 3T MR scanner with a 20-channel head coil at 10 different positions with water-unsuppressed MRSI. At the eleventh position, water-suppressed MRSI data were acquired. Data of 7 volunteers were used to estimate sensitivity maps and form a base for simulating training data. A neural network was designed and trained to remove the effect of sensitivity profiles of the coil elements from the MRSI data. The water-suppressed MRSI data of the remaining volunteers were used to evaluate the performance of the new k-space-based coil combination relative to that of a conventional image-based alternative. RESULTS: For both approaches, the resulting metabolic ratio maps were similar. The SNR of the k-space-based approach was comparable to the conventional approach in low SNR regions, but underperformed for high SNR. The Cramér-Rao lower bounds show the same trend. The analysis of the FWHM showed no difference between the two methods. CONCLUSION: k-Space-based coil combination of MRSI data is feasible and reduces the amount of raw data immediately after their sampling.

The roles of predictors in cardiovascular risk models - a question of modeling culture?

BACKGROUND: While machine learning (ML) algorithms may predict cardiovascular outcomes more accurately than statistical models, their result is usually not representable by a transparent formula. Hence, it is often unclear how specific values of predictors lead to the predictions. We aimed to demonstrate with graphical tools how predictor-risk relations in cardiovascular risk prediction models fitted by ML algorithms and by statistical approaches may differ, and how sample size affects the stability of the estimated relations. METHODS: We reanalyzed data from a large registry of 1.5 million participants in a national health screening program. Three data analysts developed analytical strategies to predict cardiovascular events within 1 year from health screening. This was done for the full data set and with gradually reduced sample sizes, and each data analyst followed their favorite modeling approach. Predictor-risk relations were visualized by partial dependence and individual conditional expectation plots. RESULTS: When comparing the modeling algorithms, we found some similarities between these visualizations but also occasional divergence. The smaller the sample size, the more the predictor-risk relation depended on the modeling algorithm used, and also sampling variability played an increased role. Predictive performance was similar if the models were derived on the full data set, whereas smaller sample sizes favored simpler models. CONCLUSION: Predictor-risk relations from ML models may differ from those obtained by statistical models, even with large sample sizes. Hence, predictors may assume different roles in risk prediction models. As long as sample size is sufficient, predictive accuracy is not largely affected by the choice of algorithm.

Neural sentence embedding models for semantic similarity estimation in the biomedical domain.

BACKGROUND: Neural network based embedding models are receiving significant attention in the field of natural language processing due to their capability to effectively capture semantic information representing words, sentences or even larger text elements in low-dimensional vector space. While current state-of-the-art models for assessing the semantic similarity of textual statements from biomedical publications depend on the availability of laboriously curated ontologies, unsupervised neural embedding models only require large text corpora as input and do not need manual curation. In this study, we investigated the efficacy of current state-of-the-art neural sentence embedding models for semantic similarity estimation of sentences from biomedical literature. We trained different neural embedding models on 1.7 million articles from the PubMed Open Access dataset, and evaluated them based on a biomedical benchmark set containing 100 sentence pairs annotated by human experts and a smaller contradiction subset derived from the original benchmark set. RESULTS: Experimental results showed that, with a Pearson correlation of 0.819, our best unsupervised model based on the Paragraph Vector Distributed Memory algorithm outperforms previous state-of-the-art results achieved on the BIOSSES biomedical benchmark set. Moreover, our proposed supervised model that combines different string-based similarity metrics with a neural embedding model surpasses previous ontology-dependent supervised state-of-the-art approaches in terms of Pearson's r (r = 0.871) on the biomedical benchmark set. In contrast to the promising results for the original benchmark, we found our best models' performance on the smaller contradiction subset to be poor. CONCLUSIONS: In this study, we have highlighted the value of neural network-based models for semantic similarity estimation in the biomedical domain by showing that they can keep up with and even surpass previous state-of-the-art approaches for semantic similarity estimation that depend on the availability of laboriously curated ontologies, when evaluated on a biomedical benchmark set. Capturing contradictions and negations in biomedical sentences, however, emerged as an essential area for further work.

Fast and scalable neural embedding models for biomedical sentence classification.

BACKGROUND: Biomedical literature is expanding rapidly, and tools that help locate information of interest are needed. To this end, a multitude of different approaches for classifying sentences in biomedical publications according to their coarse semantic and rhetoric categories (e.g., Background, Methods, Results, Conclusions) have been devised, with recent state-of-the-art results reported for a complex deep learning model. Recent evidence showed that shallow and wide neural models such as fastText can provide results that are competitive or superior to complex deep learning models while requiring drastically lower training times and having better scalability. We analyze the efficacy of the fastText model in the classification of biomedical sentences in the PubMed 200k RCT benchmark, and introduce a simple pre-processing step that enables the application of fastText on sentence sequences. Furthermore, we explore the utility of two unsupervised pre-training approaches in scenarios where labeled training data are limited. RESULTS: Our fastText-based methodology yields a state-of-the-art F1 score of.917 on the PubMed 200k benchmark when sentence ordering is taken into account, with a training time of only 73 s on standard hardware. Applying fastText on single sentences, without taking sentence ordering into account, yielded an F1 score of.852 (training time 13 s). Unsupervised pre-training of N-gram vectors greatly improved the results for small training set sizes, with an increase of F1 score of.21 to.74 when trained on only 1000 randomly picked sentences without taking sentence ordering into account. CONCLUSIONS: Because of it's ease of use and performance, fastText should be among the first choices of tools when tackling biomedical text classification problems with large corpora. Unsupervised pre-training of N-gram vectors on domain-specific corpora also makes it possible to apply fastText when labeled training data are limited.

Predicting Intraoperative Hypothermia Burden during Non-Cardiac Surgery: A Retrospective Study Comparing Regression to Six Machine Learning Algorithms.

BACKGROUND: Inadvertent intraoperative hypothermia is a common complication that affects patient comfort and morbidity. As the development of hypothermia is a complex phenomenon, predicting it using machine learning (ML) algorithms may be superior to logistic regression. METHODS: We performed a single-center retrospective study and assembled a feature set comprised of 71 variables. The primary outcome was hypothermia burden, defined as the area under the intraoperative temperature curve below 37 °C over time. We built seven prediction models (logistic regression, extreme gradient boosting (XGBoost), random forest (RF), multi-layer perceptron neural network (MLP), linear discriminant analysis (LDA), k-nearest neighbor (KNN), and Gaussian naïve Bayes (GNB)) to predict whether patients would not develop hypothermia or would develop mild, moderate, or severe hypothermia. For each model, we assessed discrimination (F1 score, area under the receiver operating curve, precision, recall) and calibration (calibration-in-the-large, calibration intercept, calibration slope). RESULTS: We included data from 87,116 anesthesia cases. Predicting the hypothermia burden group using logistic regression yielded a weighted F1 score of 0.397. Ranked from highest to lowest weighted F1 score, the ML algorithms performed as follows: XGBoost (0.44), RF (0.418), LDA (0.406), LDA (0.4), KNN (0.362), and GNB (0.32). CONCLUSIONS: ML is suitable for predicting intraoperative hypothermia and could be applied in clinical practice.

Development of a Reinforcement Learning Algorithm to Optimize Corticosteroid Therapy in Critically Ill Patients with Sepsis.

BACKGROUND: The optimal indication, dose, and timing of corticosteroids in sepsis is controversial. Here, we used reinforcement learning to derive the optimal steroid policy in septic patients based on data on 3051 ICU admissions from the AmsterdamUMCdb intensive care database. METHODS: We identified septic patients according to the 2016 consensus definition. An actor-critic RL algorithm using ICU mortality as a reward signal was developed to determine the optimal treatment policy from time-series data on 277 clinical parameters. We performed off-policy evaluation and testing in independent subsets to assess the algorithm's performance. RESULTS: Agreement between the RL agent's policy and the actual documented treatment reached 59%. Our RL agent's treatment policy was more restrictive compared to the actual clinician behavior: our algorithm suggested withholding corticosteroids in 62% of the patient states, versus 52% according to the physicians' policy. The 95% lower bound of the expected reward was higher for the RL agent than clinicians' historical decisions. ICU mortality after concordant action in the testing dataset was lower both when corticosteroids had been withheld and when corticosteroids had been prescribed by the virtual agent. The most relevant variables were vital parameters and laboratory values, such as blood pressure, heart rate, leucocyte count, and glycemia. CONCLUSIONS: Individualized use of corticosteroids in sepsis may result in a mortality benefit, but optimal treatment policy may be more restrictive than the routine clinical practice. Whilst external validation is needed, our study motivates a 'precision-medicine' approach to future prospective controlled trials and practice.

Machine learning-derived electrocardiographic algorithm for the detection of cardiac amyloidosis.

BACKGROUND: Diagnosis of cardiac amyloidosis (CA) requires advanced imaging techniques. Typical surface ECG patterns have been described, but their diagnostic abilities are limited. OBJECTIVE: The aim was to perform a thorough electrophysiological characterisation of patients with CA and derive an easy-to-use tool for diagnosis. METHODS: We applied electrocardiographic imaging (ECGI) to acquire electroanatomical maps in patients with CA and controls. A machine learning approach was then used to decipher the complex data sets obtained and generate a surface ECG-based diagnostic tool. FINDINGS: Areas of low voltage were localised in the basal inferior regions of both ventricles and the remaining right ventricular segments in CA. The earliest epicardial breakthrough of myocardial activation was visualised on the right ventricle. Potential maps revealed an accelerated and diffuse propagation pattern. We correlated the results from ECGI with 12-lead ECG recordings. Ventricular activation correlated best with R-peak timing in leads V1-V3. Epicardial voltage showed a strong positive correlation with R-peak amplitude in the inferior leads II, III and aVF. Respective surface ECG leads showed two characteristic patterns. Ten blinded cardiologists were asked to identify patients with CA by analysing 12-lead ECGs before and after training on the defined ECG patterns. Training led to significant improvements in the detection rate of CA, with an area under the curve of 0.69 before and 0.97 after training. INTERPRETATION: Using a machine learning approach, an ECG-based tool was developed from detailed electroanatomical mapping of patients with CA. The ECG algorithm is simple and has proven helpful to suspect CA without the aid of advanced imaging modalities.

Convolutional Neural Networks for Fully Automated Diagnosis of Cardiac Amyloidosis by Cardiac Magnetic Resonance Imaging.

AIMS: We tested the hypothesis that artificial intelligence (AI)-powered algorithms applied to cardiac magnetic resonance (CMR) images could be able to detect the potential patterns of cardiac amyloidosis (CA). Readers in CMR centers with a low volume of referrals for the detection of myocardial storage diseases or a low volume of CMRs, in general, may overlook CA. In light of the growing prevalence of the disease and emerging therapeutic options, there is an urgent need to avoid misdiagnoses. METHODS AND RESULTS: Using CMR data from 502 patients (CA: n = 82), we trained convolutional neural networks (CNNs) to automatically diagnose patients with CA. We compared the diagnostic accuracy of different state-of-the-art deep learning techniques on common CMR imaging protocols in detecting imaging patterns associated with CA. As a result of a 10-fold cross-validated evaluation, the best-performing fine-tuned CNN achieved an average ROC AUC score of 0.96, resulting in a diagnostic accuracy of 94% sensitivity and 90% specificity. CONCLUSIONS: Applying AI to CMR to diagnose CA may set a remarkable milestone in an attempt to establish a fully computational diagnostic path for the diagnosis of CA, in order to support the complex diagnostic work-up requiring a profound knowledge of experts from different disciplines.

Magnetic Resonance Imaging for Evaluation of Interstitial Fibrosis in Kidney Allografts.

Interstitial fibrosis (IF) is the common pathway of chronic kidney injury in various conditions. Magnetic resonance imaging (MRI) may be a promising tool for the noninvasive assessment of IF in renal allografts. METHODS: This prospective trial was primarily designed to investigate whether the results of T1-weighted MRI associate with the degree of IF. Thirty-two kidney transplant recipients were subjected to 1.5-Tesla MRI scans shortly before or after routine allograft biopsies. MRI parameters [T1 and T2 relaxation times; apparent diffusion coefficient (ADC)] were assessed for cortical and medullary sections. RESULTS: Advanced IF (Banff ci score >1) was associated with higher cortical T1 (but not T2) values [1451 (median; interquartile range: 1331-1506) versus 1306 (1197-1321) ms in subjects with ci scores ≤1; P = 0.011; receiver operating characteristic area under the curve for prediction of ci > 1: 0.76]. In parallel, T1 values were associated with kidney function and proteinuria. There was also a relationship between IF and corticomedullary differences on ADC maps (receiver operating characteristic area under the curve for prediction of ci ≤ 1: 0.79). CONCLUSIONS: Our results support the use of MRI for noninvasive assessment of allograft scarring. Future studies will have to clarify the role of T1 (and ADC) mapping as a surrogate endpoint reflecting the progression of chronic graft damage.

Machine Learning Enables Prediction of Cardiac Amyloidosis by Routine Laboratory Parameters: A Proof-of-Concept Study.

(1) Background: Cardiac amyloidosis (CA) is a rare and complex condition with poor prognosis. While novel therapies improve outcomes, many affected individuals remain undiagnosed due to a lack of awareness among clinicians. This study was undertaken to develop an expert-independent machine learning (ML) prediction model for CA relying on routinely determined laboratory parameters. (2) Methods: In a first step, we developed baseline linear models based on logistic regression. In a second step, we used an ML algorithm based on gradient tree boosting to improve our linear prediction model, and to perform non-linear prediction. Then, we compared the performance of all diagnostic algorithms. All prediction models were developed on a training cohort, consisting of patients with proven CA (positive cases, n = 121) and amyloidosis-unrelated heart failure (HF) patients (negative cases, n = 415). Performances of all prediction models were evaluated on a separate prognostic validation cohort with 37 CA-positive and 124 CA-negative patients. (3) Results: Our best model, based on gradient-boosted ensembles of decision trees, achieved an area under the receiver operating characteristic curve (ROC AUC) score of 0.86, with sensitivity and specificity of 89.2% and 78.2%, respectively. The best linear model had an ROC AUC score of 0.75, with sensitivity and specificity of 84.6 and 71.7, respectively. (4) Conclusions: Our work demonstrates that ML makes it possible to utilize basic laboratory parameters to generate a distinct CA-related HF profile compared with CA-unrelated HF patients. This proof-of-concept study opens a potential new avenue in the diagnostic workup of CA and may assist physicians in clinical reasoning.

Patients with Heart Failure and Preserved Ejection Fraction Are at Risk of Gastrointestinal Bleeding.

AIMS: Two thirds of patients with heart failure and preserved ejection fraction (HFpEF) have an indication for oral anticoagulation (OAC) to prevent thromboembolic events. However, evidence regarding the safety of OAC in HFpEF is limited. Therefore, our aim was to describe bleeding events and to find predictors of bleeding in a large HFpEF cohort. METHODS AND RESULTS: We recorded bleeding events in a prospective HFpEF cohort. Out of 328 patients (median age 71 years (interquartile range (IQR) 67-77)), 64.6% (n = 212) were treated with OAC. Of those, 65.1% (n = 138) received vitamin-K-antagonists (VKA) and 34.9% (n = 72) non-vitamin K oral anticoagulants (NOACs). During a median follow-up time of 42 (IQR 17-63) months, a total of 54 bleeding events occurred. Patients on OAC experienced more bleeding events (n = 49 (23.1%) versus n = 5 (4.3%), p < 0.001). Major drivers of events were gastrointestinal (GI) bleeding (n = 18 (36.7%)]. HAS-BLED (Hypertension, Abnormal Renal/Liver Function, Stroke, Bleeding History or Predisposition, Labile INR, Elderly, Drugs/Alcohol Concomitantly) score (hazard ratios (HR) of 2.15 (95% confidence interval (CI) 1.65-2.79, p < 0.001)) was the strongest independent predictor for overall bleeding. In the subgroup of GI bleeding, mean right atrial pressure (mRAP: HR of 1.13 (95% CI 1.03-1.25, p = 0.013)) and HAS-BLED score (HR of 1.74 (95% CI 1.15-2.64, p = 0.009)] remained significantly associatiated with bleeding events after adjustment. mRAP provided additional prognostic value beyond the HAS-BLED score with an improvement from 0.63 to 0.71 (95% CI 0.58-0.84, p for comparison 0.032), by C-statistic. This additional prognostic value was confirmed by significant improvements in net reclassification index (61.3%, p = 0.019) and integrated discrimination improvement (3.4%, p = 0.015). CONCLUSION: OAC-treated HFpEF patients are at high risk of GI bleeding. High mRAP as an indicator of advanced stage of disease was predictive for GI bleeding events and provided additional risk stratification information beyond that obtained by HAS-BLED score.

Angs (Angiotensins) of the Alternative Renin-Angiotensin System Predict Outcome in Patients With Heart Failure and Preserved Ejection Fraction.

The renin-angiotensin system plays an important role in the development and progression of heart failure (HF). In addition to the classical renin-angiotensin pathway, an alternative pathway produces Angs (angiotensins), which counteract the negative effects of Ang II. We hypothesized that Ang profiling could provide insights into the pathogenesis and prognosis of HF with preserved ejection fraction. We aimed to investigate the effects of Angs on outcome in HF with preserved ejection fraction. Consecutive patients were included into a prospective single-center registry. Clinical, laboratory, and imaging parameters were assessed and serum samples were taken at baseline and measured by mass spectroscopy. Serum equilibrium levels were analyzed in regard to the combined clinical end point of cardiovascular death or HF hospitalization. In total, 155 patients were included during a median follow-up time of 22.5 (interquartile range, 4.0-61.0) months, 52 individuals (34%) reached the combined end point. We identified higher levels of Ang 1-7 and Ang 1-5 as predictors for poor outcome. After adjusting for potential confounding factors, Ang 1-5 remained predictive for poor outcome. In addition to Ang 1-7 and Ang 1-5, the novel ACE (angiotensin-converting enzyme) independent Ang composite marker [Ang 1-7+Ang 1-5] was shown to predict adverse events. We conclude that Angs of the alternative renin-angiotensin system seem to play a role in HF with preserved ejection fraction and are linked to outcome in patients with HF and preserved ejection fraction. Ang 1-5 and the alternative renin-angiotensin system composite marker [Ang 1-7+Ang 1-5] are independent predictors of outcome.

Supporting shared hypothesis testing in the biomedical domain.

BACKGROUND: Pathogenesis of inflammatory diseases can be tracked by studying the causality relationships among the factors contributing to its development. We could, for instance, hypothesize on the connections of the pathogenesis outcomes to the observed conditions. And to prove such causal hypotheses we would need to have the full understanding of the causal relationships, and we would have to provide all the necessary evidences to support our claims. In practice, however, we might not possess all the background knowledge on the causality relationships, and we might be unable to collect all the evidence to prove our hypotheses. RESULTS: In this work we propose a methodology for the translation of biological knowledge on causality relationships of biological processes and their effects on conditions to a computational framework for hypothesis testing. The methodology consists of two main points: hypothesis graph construction from the formalization of the background knowledge on causality relationships, and confidence measurement in a causality hypothesis as a normalized weighted path computation in the hypothesis graph. In this framework, we can simulate collection of evidences and assess confidence in a causality hypothesis by measuring it proportionally to the amount of available knowledge and collected evidences. CONCLUSIONS: We evaluate our methodology on a hypothesis graph that represents both contributing factors which may cause cartilage degradation and the factors which might be caused by the cartilage degradation during osteoarthritis. Hypothesis graph construction has proven to be robust to the addition of potentially contradictory information on the simultaneously positive and negative effects. The obtained confidence measures for the specific causality hypotheses have been validated by our domain experts, and, correspond closely to their subjective assessments of confidences in investigated hypotheses. Overall, our methodology for a shared hypothesis testing framework exhibits important properties that researchers will find useful in literature review for their experimental studies, planning and prioritizing evidence collection acquisition procedures, and testing their hypotheses with different depths of knowledge on causal dependencies of biological processes and their effects on the observed conditions.