Search

Background: Women have been underrepresented in clinical trials for many years. Machine-learning models trained on clinical trial abstracts may capture and amplify biases in the data. Specifically, word embeddings are models that enable representing words as vectors and are the building block of most natural language processing systems. If word embeddings are trained on clinical trial abstracts, predictive models that use the embeddings will exhibit gender performance gaps., Objective: We aim to capture temporal trends in clinical trials through temporal distribution matching on contextual word embeddings (specifically, BERT) and explore its effect on the bias manifested in downstream tasks., Methods: We present TeDi-BERT, a method to harness the temporal trend of increasing women's inclusion in clinical trials to train contextual word embeddings. We implement temporal distribution matching through an adversarial classifier, trying to distinguish old from new clinical trial abstracts based on their embeddings. The temporal distribution matching acts as a form of domain adaptation from older to more recent clinical trials. We evaluate our model on 2 clinical tasks: prediction of unplanned readmission to the intensive care unit and hospital length of stay prediction. We also conduct an algorithmic analysis of the proposed method., Results: In readmission prediction, TeDi-BERT achieved area under the receiver operating characteristic curve of 0.64 for female patients versus the baseline of 0.62 (P<.001), and 0.66 for male patients versus the baseline of 0.64 (P<.001). In the length of stay regression, TeDi-BERT achieved a mean absolute error of 4.56 (95% CI 4.44-4.68) for female patients versus 4.62 (95% CI 4.50-4.74, P<.001) and 4.54 (95% CI 4.44-4.65) for male patients versus 4.6 (95% CI 4.50-4.71, P<.001)., Conclusions: In both clinical tasks, TeDi-BERT improved performance for female patients, as expected; but it also improved performance for male patients. Our results show that accuracy for one gender does not need to be exchanged for bias reduction, but rather that good science improves clinical results for all. Contextual word embedding models trained to capture temporal trends can help mitigate the effects of bias that changes over time in the training data., (©Shunit Agmon, Uriel Singer, Kira Radinsky. Originally published in JMIR AI (https://ai.jmir.org), 02.10.2024.)

Early detection of left ventricular systolic dysfunction (LVSD) may prompt early care and improve outcomes for asymptomatic patients. Standard 12-lead ECG may be used to predict LVSD. We aimed to compare the performance of Machine Learning Algorithms (MLA) and physicians in predicting LVSD from a standard 12-lead ECG. By utilizing a dataset of 13,820 pairs of ECGs and echocardiography, a deep residual convolutional neural network was trained for predicting LVSD (ejection fraction (EF) < 50%) from ECG. The ECGs of the test set (n = 850) were assessed for LVSD by the MLA and six physicians. The performance was compared using sensitivity, specificity, and C-statistics. The interobserver agreement between the physicians for the prediction of LVSD was moderate (κ = 0.50), with average sensitivity and specificity of 70%. The C-statistic of the MLA was 0.85. Repeating this analysis with LVSD defined as EF < 35% resulted in an improvement in physicians’ average sensitivity to 84% but their specificity decreased to 57%. The MLA C-statistic was 0.88 with this threshold. We conclude that although MLA outperformed physicians in predicting LVSD from standard ECG, prior to robust implementation of MLA in ECG machines, physicians should be encouraged to use this approach as a simple and readily available aid for LVSD screening.

Objective: To analyze gender bias in clinical trials, to design an algorithm that mitigates the effects of biases of gender representation on natural-language (NLP) systems trained on text drawn from clinical trials, and to evaluate its performance., Materials and Methods: We analyze gender bias in clinical trials described by 16 772 PubMed abstracts (2008-2018). We present a method to augment word embeddings, the core building block of NLP-centric representations, by weighting abstracts by the number of women participants in the trial. We evaluate the resulting gender-sensitive embeddings performance on several clinical prediction tasks: comorbidity classification, hospital length of stay prediction, and intensive care unit (ICU) readmission prediction., Results: For female patients, the gender-sensitive model area under the receiver-operator characteristic (AUROC) is 0.86 versus the baseline of 0.81 for comorbidity classification, mean absolute error 4.59 versus the baseline of 4.66 for length of stay prediction, and AUROC 0.69 versus 0.67 for ICU readmission. All results are statistically significant., Discussion: Women have been underrepresented in clinical trials. Thus, using the broad clinical trials literature as training data for statistical language models could result in biased models, with deficits in knowledge about women. The method presented enables gender-sensitive use of publications as training data for word embeddings. In experiments, the gender-sensitive embeddings show better performance than baseline embeddings for the clinical tasks studied. The results highlight opportunities for recognizing and addressing gender and other representational biases in the clinical trials literature., Conclusion: Addressing representational biases in data for training NLP embeddings can lead to better results on downstream tasks for underrepresented populations., (© The Author(s) 2021. Published by Oxford University Press on behalf of the American Medical Informatics Association. All rights reserved. For permissions, please email: journals.permissions@oup.com.)

Summary: How do nuances of scientists' attention influence what they discover? We pursue an understanding of the influences of patterns of attention on discovery with a case study about confirmations of protein-protein interactions over time. We find that modeling and accounting for attention can help us to recognize and interpret biases in large-scale and widely used databases of confirmed interactions and to better understand missing data and unknowns. Additionally, we present an analysis of how awareness of patterns of attention and use of debiasing techniques can foster earlier discoveries., Availability and Implementation: The data is freely available at https://github.com/urielsinger/PPI-unbias., (© The Author(s) 2020. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.)

Recent health reforms have created incentives for cardiologists and accountable care organizations to participate in value-based care models for heart failure (HF). Accurate risk stratification of HF patients is critical to efficiently deploy interventions aimed at reducing preventable utilization. The goal of this paper was to compare deep learning approaches with traditional logistic regression (LR) to predict preventable utilization among HF patients. We conducted a prognostic study using data on 93,260 HF patients continuously enrolled for 2-years in a large U.S. commercial insurer to develop and validate prediction models for three outcomes of interest: preventable hospitalizations, preventable emergency department (ED) visits, and preventable costs. Patients were split into training, validation, and testing samples. Outcomes were modeled using traditional and enhanced LR and compared to gradient boosting model and deep learning models using sequential and non-sequential inputs. Evaluation metrics included precision (positive predictive value) at k, cost capture, and Area Under the Receiver operating characteristic (AUROC). Deep learning models consistently outperformed LR for all three outcomes with respect to the chosen evaluation metrics. Precision at 1% for preventable hospitalizations was 43% for deep learning compared to 30% for enhanced LR. Precision at 1% for preventable ED visits was 39% for deep learning compared to 33% for enhanced LR. For preventable cost, cost capture at 1% was 30% for sequential deep learning, compared to 18% for enhanced LR. The highest AUROCs for deep learning were 0.778, 0.681 and 0.727, respectively. These results offer a promising approach to identify patients for targeted interventions.

Background: Opioid use disorder (OUD) affects an estimated 16 million people worldwide. The diagnosis of OUD is commonly delayed or missed altogether. We aimed to test the utility of machine learning in creating a prediction model and algorithm for early diagnosis of OUD., Subjects and Methods: We analyzed data gathered in a commercial claim database from January 1, 2006, to December 31, 2018 of 10 million medical insurance claims from 550 000 patient records. We compiled 436 predictor candidates, divided to six feature groups - demographics, chronic conditions, diagnosis and procedures features, medication features, medical costs, and episode counts. We employed the Word2Vec algorithm and the Gradient Boosting trees algorithm for the analysis., Results: The c-statistic for the model was 0.959, with a sensitivity of 0.85 and specificity of 0.882. Positive Predictive Value (PPV) was 0.362 and Negative Predictive Value (NPV) was 0.998. Significant differences between positive OUD- and negative OUD- controls were in the mean annual amount of opioid use days, number of overlaps in opioid prescriptions per year, mean annual opioid prescriptions, and annual benzodiazepine and muscle relaxant prescriptions. Notable differences were the count of intervertebral disc disorder-related complaints per year, post laminectomy syndrome diagnosed per year, and pain disorders diagnosis per year. Significant differences were also found in the episodes and costs categories., Conclusions: The new algorithm offers a mean 14.4 months reduction in time to diagnosis of OUD, at potential saving in further morbidity, medical cost, addictions and mortality., (© 2020 The Authors. Pharmacology Research & Perspectives published by John Wiley & Sons Ltd, British Pharmacological Society and American Society for Pharmacology and Experimental Therapeutics.)

Background: End stage renal disease (ESRD) describes the most severe stage of chronic kidney disease (CKD), when patients need dialysis or renal transplant. There is often a delay in recognizing, diagnosing, and treating the various etiologies of CKD. The objective of the present study was to employ machine learning algorithms to develop a prediction model for progression to ESRD based on a large-scale multidimensional database., Methods: This study analyzed 10,000,000 medical insurance claims from 550,000 patient records using a commercial health insurance database. Inclusion criteria were patients over the age of 18 diagnosed with CKD Stages 1-4. We compiled 240 predictor candidates, divided into six feature groups: demographics, chronic conditions, diagnosis and procedure features, medication features, medical costs, and episode counts. We used a feature embedding method based on implementation of the Word2Vec algorithm to further capture temporal information for the three main components of the data: diagnosis, procedures, and medications. For the analysis, we used the gradient boosting tree algorithm (XGBoost implementation)., Results: The C-statistic for the model was 0.93 [(0.916-0.943) 95% confidence interval], with a sensitivity of 0.715 and specificity of 0.958. Positive Predictive Value (PPV) was 0.517, and Negative Predictive Value (NPV) was 0.981. For the top 1 percentile of patients identified by our model, the PPV was 1.0. In addition, for the top 5 percentile of patients identified by our model, the PPV was 0.71. All the results above were tested on the test data only, and the threshold used to obtain these results was 0.1. Notable features contributing to the model were chronic heart and ischemic heart disease as a comorbidity, patient age, and number of hypertensive crisis events., Conclusions: When a patient is approaching the threshold of ESRD risk, a warning message can be sent electronically to the physician, who will initiate a referral for a nephrology consultation to ensure an investigation to hasten the establishment of a diagnosis and initiate management and therapy when appropriate.

Despite effective medications, rates of uncontrolled glucose levels in type 2 diabetes remain high. We aimed to test the utility of machine learning applied to big data in identifying the potential role of concomitant drugs not taken for diabetes which may contribute to lowering blood glucose. Success in controlling blood glucose was defined as achieving HgA1c levels < 6.5% after 90-365 days following diagnosis and initiating treatment. Among numerous concomitant drugs taken by type 2 diabetic patients, alpha 1 (α1)-adrenoceptor antagonist drugs were the only group of medications that significantly improved the success rate of glucose control. Searching the published literature, this effect of α1-adrenoceptor antagonists has been shown in animal models, where this class of medications appears to induce insulin secretion. In conclusion, machine learning of big data is a novel method to identify effective antidiabetic effects for potential repurposable medications already on the market for other indications. Because these α1-adrenoceptor antagonists are widely used in men for treating benign prostate hyperplasia (BPH) at age groups exhibiting increased rates of type 2 diabetes, this finding is of potential clinical significance., Competing Interests: The authors declare no conflict of interest associated with this study. No financial support was received to conduct this study., (© 2019 The Authors. Pharmacology Research & Perspectives published by John Wiley & Sons Ltd, British Pharmacological Society and American Society for Pharmacology and Experimental Therapeutics.)

Introduction : With the majority of elderly persons consuming multiple drugs, inappropriate drug use is a major issue in geriatric medicine. Areas covered : We reviewed PubMed, Embase, and Cochrane from inception to 1 May 2019 for potentially inappropriate use of medications, polypharmacy, and age-dependent changes in pharmacokinetics and pharmacodynamics. We selected to highlight new aspects that have emerged in recent years: appropriate monitoring of drug adherence and the introduction of Big Data analysis in advancing geriatric pharmacology. Expert opinion : There are major gaps in the pharmacological treatment of the elderly. Most drugs were designed and tested in adults, with no pharmacokinetic and pharmacodynamic data on changes in old age. This void must be corrected through systematic and well-designed research programs. Potentially inappropriate use of medications (PIM) in the elderly is a serious issue in advanced age. Analysis of PIM shows relatively low predictive value in real life medicine. Most physicians continue to prescribe to the elderly medicines which should not be given at all, or not combined. Polypharmacy is a complex issue in old age, and in many cases treating physicians are not conducting critical assessment of the need for numerous medications.

Background: Most patients with Parkinson's disease exhibit intracellular accumulation of the α-synuclein protein encoded by the α-synuclein gene. It was recently shown that β 2 -adrenoreceptor agonists downregulate this gene, decreasing the apparent risk of Parkinson's disease by up to 40%. In contrast, exposure to β-blocking drugs increases production of the α-synuclein protein., Objective: The aim of this study was to examine whether chronic exposure to β-blockers is associated with an increased risk for Parkinson's disease., Patients and Methods: From the electronic charts of Maccabi Health Services, we identified all patients receiving their first β-blocker treatment between 1998 and 2004, and followed them up, for a diagnosis of Parkinson's disease, between 2005 and 2016. We calculated the morbidity hazard of Parkinson's disease diagnosis in users of β-blockers compared with non-users, as well as users of angiotensin-converting enzyme (ACE) inhibitors for hypertension, after adjusting for sex, age, weight, smoking status, cholesterol levels and use of statins, employing the Cox proportional hazard model. We also conducted a Kaplan-Meier survival analysis., Results: Overall, 145,098 patients received β-blockers, and 1,187,151 patients did not. The adjusted hazard ratio for Parkinson's disease among β-blocker users was 1.51 (95% confidence interval 1.28-1.77; p < 0.0001). In contrast, the Parkinson's disease morbidity hazard for patients receiving ACE inhibitors was no different than for the general population. The morbidity risk showed the effect of cumulative dose response with low threshold levels., Conclusions: Chronic use of β-blockers confers a time- and dose-dependent increased risk for Parkinson's disease. In view of the available alternatives for β-blockers, their chronic use should be carefully reconsidered.

Designing a new drug is a lengthy and expensive process. As the space of potential molecules is very large ( Polishchuk , P. G. ; Madzhidov , T. I. ; Varnek , A. Estimation of the size of drug-like chemical space based on GDB-17 data . J. Comput.-Aided Mol. Des. 2013 , 27 , 675 -679 10.1007/s10822-013-9672-4 ), a common technique during drug discovery is to start from a molecule which already has some of the desired properties. An interdisciplinary team of scientists generates hypothesis about the required changes to the prototype. In this work, we develop a deep-learning unsupervised-approach that automatically generates potential drug molecules given a prototype drug. We show that the molecules generated by the system are valid molecules and significantly different from the prototype drug. Out of the compounds generated by the system, we identified 35 known FDA-approved drugs. As an example, our system generated isoniazid, one of the main drugs for tuberculosis. We suggest several ranking functions for the generated molecules and present results that the top ten generated molecules per prototype drug contained in our retrospective experiments 23 known FDA-approved drugs.

Despite effective medications, rates of uncontrolled hypertension remain high. Treatment protocols are largely based on randomized trials and meta-analyses of these studies. The objective of this study was to test the utility of machine learning of big data in gaining insight into the treatment of hypertension. We applied machine learning techniques such as decision trees and neural networks, to identify determinants that contribute to the success of hypertension drug treatment on a large set of patients. We also identified concomitant drugs not considered to have antihypertensive activity, which may contribute to lowering blood pressure (BP) control. Higher initial BP predicts lower success rates. Among the medication options and their combinations, treatment with beta blockers appears to be more commonly effective, which is not reflected in contemporary guidelines. Among numerous concomitant drugs taken by hypertensive patients, proton pump inhibitors (PPIs), and HMG CO-A reductase inhibitors (statins) significantly improved the success rate of hypertension. In conclusions, machine learning of big data is a novel method to identify effective antihypertensive therapy and for repurposing medications already on the market for new indications. Our results related to beta blockers, stemming from machine learning of a large and diverse set of big data, in contrast to the much narrower criteria for randomized clinic trials (RCTs), should be corroborated and affirmed by other methods, as they hold potential promise for an old class of drugs which may be presently underutilized. These previously unrecognized effects of PPIs and statins have been very recently identified as effective in lowering BP in preliminary clinical observations, lending credibility to our big data results.