Deep learning to detect left ventricular structural abnormalities in chest X-rays.

BACKGROUND AND AIMS: Early identification of cardiac structural abnormalities indicative of heart failure is crucial to improving patient outcomes. Chest X-rays (CXRs) are routinely conducted on a broad population of patients, presenting an opportunity to build scalable screening tools for structural abnormalities indicative of Stage B or worse heart failure with deep learning methods. In this study, a model was developed to identify severe left ventricular hypertrophy (SLVH) and dilated left ventricle (DLV) using CXRs. METHODS: A total of 71 589 unique CXRs from 24 689 different patients completed within 1 year of echocardiograms were identified. Labels for SLVH, DLV, and a composite label indicating the presence of either were extracted from echocardiograms. A deep learning model was developed and evaluated using area under the receiver operating characteristic curve (AUROC). Performance was additionally validated on 8003 CXRs from an external site and compared against visual assessment by 15 board-certified radiologists. RESULTS: The model yielded an AUROC of 0.79 (0.76-0.81) for SLVH, 0.80 (0.77-0.84) for DLV, and 0.80 (0.78-0.83) for the composite label, with similar performance on an external data set. The model outperformed all 15 individual radiologists for predicting the composite label and achieved a sensitivity of 71% vs. 66% against the consensus vote across all radiologists at a fixed specificity of 73%. CONCLUSIONS: Deep learning analysis of CXRs can accurately detect the presence of certain structural abnormalities and may be useful in early identification of patients with LV hypertrophy and dilation. As a resource to promote further innovation, 71 589 CXRs with adjoining echocardiographic labels have been made publicly available.

Deep Learning Electrocardiographic Analysis for Detection of Left-Sided Valvular Heart Disease.

BACKGROUND: Valvular heart disease is an important contributor to cardiovascular morbidity and mortality and remains underdiagnosed. Deep learning analysis of electrocardiography (ECG) may be useful in detecting aortic stenosis (AS), aortic regurgitation (AR), and mitral regurgitation (MR). OBJECTIVES: This study aimed to develop ECG deep learning algorithms to identify moderate or severe AS, AR, and MR alone and in combination. METHODS: A total of 77,163 patients undergoing ECG within 1 year before echocardiography from 2005-2021 were identified and split into train (n = 43,165), validation (n = 12,950), and test sets (n = 21,048; 7.8% with any of AS, AR, or MR). Model performance was assessed using area under the receiver-operating characteristic (AU-ROC) and precision-recall curves. Outside validation was conducted on an independent data set. Test accuracy was modeled using different disease prevalence levels to simulate screening efficacy using the deep learning model. RESULTS: The deep learning algorithm model accuracy was as follows: AS (AU-ROC: 0.88), AR (AU-ROC: 0.77), MR (AU-ROC: 0.83), and any of AS, AR, or MR (AU-ROC: 0.84; sensitivity 78%, specificity 73%) with similar accuracy in external validation. In screening program modeling, test characteristics were dependent on underlying prevalence and selected sensitivity levels. At a prevalence of 7.8%, the positive and negative predictive values were 20% and 97.6%, respectively. CONCLUSIONS: Deep learning analysis of the ECG can accurately detect AS, AR, and MR in this multicenter cohort and may serve as the basis for the development of a valvular heart disease screening program.

Inverse-Weighted Survival Games.

Deep models trained through maximum likelihood have achieved state-of-the-art results for survival analysis. Despite this training scheme, practitioners evaluate models under other criteria, such as binary classification losses at a chosen set of time horizons, e.g. Brier score (BS) and Bernoulli log likelihood (BLL). Models trained with maximum likelihood may have poor BS or BLL since maximum likelihood does not directly optimize these criteria. Directly optimizing criteria like BS requires inverse-weighting by the censoring distribution. However, estimating the censoring model under these metrics requires inverse-weighting by the failure distribution. The objective for each model requires the other, but neither are known. To resolve this dilemma, we introduce Inverse-Weighted Survival Games. In these games, objectives for each model are built from re-weighted estimates featuring the other model, where the latter is held fixed during training. When the loss is proper, we show that the games always have the true failure and censoring distributions as a stationary point. This means models in the game do not leave the correct distributions once reached. We construct one case where this stationary point is unique. We show that these games optimize BS on simulations and then apply these principles on real world cancer and critically-ill patient data.

The Counterfactual χ-GAN: Finding comparable cohorts in observational health data.

Causal inference often relies on the counterfactual framework, which requires that treatment assignment is independent of the outcome, known as strong ignorability. Approaches to enforcing strong ignorability in causal analyses of observational data include weighting and matching methods. Effect estimates, such as the average treatment effect (ATE), are then estimated as expectations under the re-weighted or matched distribution, P. The choice of P is important and can impact the interpretation of the effect estimate and the variance of effect estimates. In this work, instead of specifying P, we learn a distribution that simultaneously maximizes coverage and minimizes variance of ATE estimates. In order to learn this distribution, this research proposes a generative adversarial network (GAN)-based model called the Counterfactual χ-GAN (cGAN), which also learns feature-balancing weights and supports unbiased causal estimation in the absence of unobserved confounding. Our model minimizes the Pearson χ2-divergence, which we show simultaneously maximizes coverage and minimizes the variance of importance sampling estimates. To our knowledge, this is the first such application of the Pearson χ2-divergence. We demonstrate the effectiveness of cGAN in achieving feature balance relative to established weighting methods in simulation and with real-world medical data.

Characterizing non-heroin opioid overdoses using electronic health records.

INTRODUCTION: The opioid epidemic is a modern public health emergency. Common interventions to alleviate the opioid epidemic aim to discourage excessive prescription of opioids. However, these methods often take place over large municipal areas (state-level) and may fail to address the diversity that exists within each opioid case (individual-level). An intervention to combat the opioid epidemic that takes place at the individual-level would be preferable. METHODS: This research leverages computational tools and methods to characterize the opioid epidemic at the individual-level using the electronic health record data from a large, academic medical center. To better understand the characteristics of patients with opioid use disorder (OUD) we leveraged a self-controlled analysis to compare the healthcare encounters before and after an individual's first overdose event recorded within the data. We further contrast these patients with matched, non-OUD controls to demonstrate the unique qualities of the OUD cohort. RESULTS: Our research confirms that the rate of opioid overdoses in our hospital significantly increased between 2006 and 2015 (P < 0.001), at an average rate of 9% per year. We further found that the period just prior to the first overdose is marked by conditions of pain or malignancy, which may suggest that overdose stems from pharmaceutical opioids prescribed for these conditions. CONCLUSIONS: Informatics-based methodologies, like those presented here, may play a role in better understanding those individuals who suffer from opioid dependency and overdose, and may lead to future research and interventions that could successfully prevent morbidity and mortality associated with this epidemic.

X-CAL: Explicit Calibration for Survival Analysis.

Survival analysis models the distribution of time until an event of interest, such as discharge from the hospital or admission to the ICU. When a model's predicted number of events within any time interval is similar to the observed number, it is called well-calibrated. A survival model's calibration can be measured using, for instance, distributional calibration (D-CALIBRATION) [Haider et al., 2020] which computes the squared difference between the observed and predicted number of events within different time intervals. Classically, calibration is addressed in post-training analysis. We develop explicit calibration (X-CAL), which turns D-CALIBRATION into a differentiable objective that can be used in survival modeling alongside maximum likelihood estimation and other objectives. X-CAL allows practitioners to directly optimize calibration and strike a desired balance between predictive power and calibration. In our experiments, we fit a variety of shallow and deep models on simulated data, a survival dataset based on MNIST, on length-of-stay prediction using MIMIC-III data, and on brain cancer data from The Cancer Genome Atlas. We show that the models we study can be miscalibrated. We give experimental evidence on these datasets that X-CAL improves D-CALIBRATION without a large decrease in concordance or likelihood.

Causal Estimation with Functional Confounders.

Causal inference relies on two fundamental assumptions: ignorability and positivity. We study causal inference when the true confounder value can be expressed as a function of the observed data; we call this setting estimation with functional confounders (EFC). In this setting ignorability is satisfied, however positivity is violated, and causal inference is impossible in general. We consider two scenarios where causal effects are estimable. First, we discuss interventions on a part of the treatment called functional interventions and a sufficient condition for effect estimation of these interventions called functional positivity. Second, we develop conditions for nonparametric effect estimation based on the gradient fields of the functional confounder and the true outcome function. To estimate effects under these conditions, we develop Level-set Orthogonal Descent Estimation (LODE). Further, we prove error bounds on LODE's effect estimates, evaluate our methods on simulated and real data, and empirically demonstrate the value of EFC.

Increase in the incidence of severe hypoglycaemia in people with Type 2 diabetes in spite of new drugs: analysis based on health insurance data from Germany.

AIM: To evaluate the use of new anti-hyperglycaemic agents that offer effective glycaemic control while reducing risk of hypoglycaemia, by analysing the incidence rates of severe hypoglycaemia in 2006 vs 2011 in relation to the medication. METHODS: This cross-sectional, population-based study used German health insurance data. All adults diagnosed with Type 2 diabetes mellitus (extrapolated to the German population: 6.35 million in 2006 and 7.52 million in 2011) were screened for severe hypoglycaemia. Anti-hyperglycaemic agents were identified by their Anatomical Therapeutic Chemical code, and defined daily doses of each medication were calculated. RESULTS: The severe hypoglycaemic event rate was 460 per 100,000 people/year in 2006 and 490 per 100,000 people/year in 2011. In 2006 and 2011, 10.9% and 7.3%, respectively, of all people with severe hypoglycaemia were on sulfonylureas, while 12.7% and 9.3%, respectively, were on a combination therapy of metformin and sulfonylureas. Among those with severe hypoglycaemia, there were no prescriptions of dipeptidyl peptidase-4 inhibitors or glucagon-like peptide-1 receptor agonists in 2006, but in 2011, 1.55% and 0.17%, of those with severe hypoglycaemia were receiving the respective treatments. In 2006 vs 2011, human insulin was prescribed for 11.3% vs 10.3% of people with severe hypoglycaemia, while insulin analogues were prescribed for 5.4% vs 8.1%, and mixed human insulins for 19.7% vs 14.0% of patients with severe hypoglycaemia. People receiving insulin analogue therapy had a higher risk of severe hypoglycaemia than those receiving metformin, after adjusting for age, gender, nephropathy diagnosis and year of survey (odds ratio 14.6; CI 13.3-15.9). CONCLUSION: The incidence of severe hypoglycaemic events in Germany increased between 2006 and 2011, despite increased use of newer anti-hyperglycaemic agents and decreased use of insulins.

Elevated GM3 plasma concentration in idiopathic Parkinson's disease: A lipidomic analysis.

Parkinson's disease (PD) is a common neurodegenerative disease whose pathological hallmark is the accumulation of intracellular α-synuclein aggregates in Lewy bodies. Lipid metabolism dysregulation may play a significant role in PD pathogenesis; however, large plasma lipidomic studies in PD are lacking. In the current study, we analyzed the lipidomic profile of plasma obtained from 150 idiopathic PD patients and 100 controls, taken from the 'Spot' study at Columbia University Medical Center in New York. Our mass spectrometry based analytical panel consisted of 520 lipid species from 39 lipid subclasses including all major classes of glycerophospholipids, sphingolipids, glycerolipids and sterols. Each lipid species was analyzed using a logistic regression model. The plasma concentrations of two lipid subclasses, triglycerides and monosialodihexosylganglioside (GM3), were different between PD and control participants. GM3 ganglioside concentration had the most significant difference between PD and controls (1.531±0.037 pmol/µl versus 1.337±0.040 pmol/µl respectively; p-value = 5.96E-04; q-value = 0.048; when normalized to total lipid: p-value = 2.890E-05; q-value = 2.933E-03). Next, we used a collection of 20 GM3 and glucosylceramide (GlcCer) species concentrations normalized to total lipid to perform a ROC curve analysis, and found that these lipids compare favorably with biomarkers reported in previous studies (AUC = 0.742 for males, AUC = 0.644 for females). Our results suggest that higher plasma GM3 levels are associated with PD. GM3 lies in the same glycosphingolipid metabolic pathway as GlcCer, a substrate of the enzyme glucocerebrosidase, which has been associated with PD. These findings are consistent with previous reports implicating lower glucocerebrosidase activity with PD risk.

Developing research priorities for palliative care of people with intellectual disabilities in Europe: a consultation process using nominal group technique.

BACKGROUND: Empirical knowledge around palliative care provision and needs of people with intellectual disabilities is extremely limited, as is the availability of research resources, including expertise and funding. This paper describes a consultation process that sought to develop an agenda for research priorities for palliative care of people with intellectual disabilities in Europe. METHODS: A two-day workshop was convened, attended by 16 academics and clinicians in the field of palliative care and intellectual disability from six European countries. The first day consisted of round-table presentations and discussions about the current state of the art, research challenges and knowledge gaps. The second day was focused on developing consensus research priorities with 12 of the workshop participants using nominal group technique, a structured method which involved generating a list of research priorities and ranking them in order of importance. RESULTS: A total of 40 research priorities were proposed and collapsed into eleven research themes. The four most important research themes were: investigating issues around end of life decision making; mapping the scale and scope of the issue; investigating the quality of palliative care for people with intellectual disabilities, including the challenges in achieving best practice; and developing outcome measures and instruments for palliative care of people with intellectual disabilities. CONCLUSIONS: The proposal of four major priority areas and a range of minor themes for future research in intellectual disability, death, dying and palliative care will help researchers to focus limited resources and research expertise on areas where it is most needed and support the building of collaborations. The next steps are to cross-validate these research priorities with people with intellectual disabilities, carers, clinicians, researchers and other stakeholders across Europe; to validate them with local and national policy makers to determine how they could best be incorporated in policy and programmes; and to translate them into actual research studies by setting up European collaborations for specific studies that require such collaboration, develop research proposals and attract research funding.

Learning probabilistic phenotypes from heterogeneous EHR data.

We present the Unsupervised Phenome Model (UPhenome), a probabilistic graphical model for large-scale discovery of computational models of disease, or phenotypes. We tackle this challenge through the joint modeling of a large set of diseases and a large set of clinical observations. The observations are drawn directly from heterogeneous patient record data (notes, laboratory tests, medications, and diagnosis codes), and the diseases are modeled in an unsupervised fashion. We apply UPhenome to two qualitatively different mixtures of patients and diseases: records of extremely sick patients in the intensive care unit with constant monitoring, and records of outpatients regularly followed by care providers over multiple years. We demonstrate that the UPhenome model can learn from these different care settings, without any additional adaptation. Our experiments show that (i) the learned phenotypes combine the heterogeneous data types more coherently than baseline LDA-based phenotypes; (ii) they each represent single diseases rather than a mix of diseases more often than the baseline ones; and (iii) when applied to unseen patient records, they are correlated with the patients' ground-truth disorders. Code for training, inference, and quantitative evaluation is made available to the research community.

Role of Patient Factors and Practice Patterns in Determining Access to Liver Waitlist.

Geographic variability in access to care is a persistent challenge in transplantation. Little is known about how patients with end-stage liver disease are chosen for referral, evaluation and listing. Utilizing death certificate data from the Centers for Disease Control and Prevention from 2002 to 2009, estimated liver demand (ELD) was measured by aggregating annual deaths from liver disease and liver transplants performed in each donor service area (DSA). In DSAs with higher ELD, more patients per capita were listed for transplantation (p < 0.001). In addition, listing rates per ELD varied fivefold across DSAs, with more patients per ELD being transplanted in DSAs with higher listing rates (p < 0.001). After adjusting for liver donor risk index and MELD at transplant, there was no association between listing rate and posttransplant survival (HR 1.002, p = 0.77). In addition, DSAs with lower listing rates were more likely to export organs (p < 0.001) of lower liver donor risk index (p < 0.001). Listing sicker patients was associated with increased access to the waitlist and transplantation and more efficient organ utilization, but had minimal effect on posttransplant outcomes after adjusting for the resulting organ shortage.

Socioeconomic status and ethnicity of deceased donor kidney recipients compared to their donors.

Public perception and misperceptions of socioeconomic disparities affect the willingness to donate organs. To improve our understanding of the flow of deceased donor kidneys, we analyzed socioeconomic status (SES) and racial/ethnic gradients between donors and recipients. In a retrospective cohort study, traditional demographic and socioeconomic factors, as well as an SES index, were compared in 56,697 deceased kidney donor and recipient pairs transplanted between 2007 and 2012. Kidneys were more likely to be transplanted in recipients of the same racial/ethnic group as the donor (p < 0.001). Kidneys tended to go to recipients of lower SES index (50.5% of the time, p < 0.001), a relationship that remained after adjusting for other available markers of donor organ quality and SES (p < 0.001). Deceased donor kidneys do not appear to be transplanted from donors of lower SES to recipients of higher SES; this information may be useful in counseling potential donors and their families regarding the distribution of their organ gifts.

Randomised clinical trial: the 5-HT4 agonist revexepride in patients with gastro-oesophageal reflux disease who have persistent symptoms despite PPI therapy.

BACKGROUND: A substantial proportion of patients with gastro-oesophageal reflux disease (GERD) have only a partial response to proton pump inhibitor (PPI) therapy. Prokinetic drugs may improve reflux symptoms by enhancing oesophageal motility and gastric emptying. AIM: To evaluate the effect of revexepride, a novel prokinetic 5-hydroxytryptamine type 4 (5-HT4 ) receptor agonist, compared with placebo, in patients with GERD who have a partial response to PPIs. METHODS: A phase 2b, double-blind, parallel-group study was conducted, in which patients were randomised to one of three revexepride treatment groups (0.1, 0.5 and 2.0 mg three times daily) or placebo (1:1:1:1 ratio). Daily e-diary data captured patients' symptoms over an 8-week treatment period. The primary efficacy outcome was the weekly percentage of regurgitation-free days in the second half of the study (weeks 5-8). RESULTS: In total, 480 patients were randomised and 477 received treatment (mean age 47.9 years; 61% women). The mean percentage of regurgitation-free days increased from baseline (range, 15.0-18.8%) to week 8 (62.3-70.5%) in all four study arms; however, there were no statistically significant differences in this change between placebo and the three treatment arms. No dose-dependent relationship in treatment effect was observed for any of the study endpoints. The incidence of treatment-emergent adverse events (TEAEs) was revexepride dose-dependent. Only one serious TEAE occurred and none resulted in death. CONCLUSIONS: Revexepride was no more effective than placebo in controlling regurgitation in patients with GERD symptoms partially responsive to PPIs. Revexepride was well tolerated. ClinicalTrials.gov Identifier: NCT01472939.

c-Abl antagonizes the YAP oncogenic function.

YES-associated protein (YAP) is a central transcription coactivator that functions as an oncogene in a number of experimental systems. However, under DNA damage, YAP activates pro-apoptotic genes in conjunction with p73. This program switching is mediated by c-Abl (Abelson murine leukemia viral oncogene) via phosphorylation of YAP at the Y357 residue (pY357). YAP as an oncogene coactivates the TEAD (transcriptional enhancer activator domain) family transcription factors. Here we asked whether c-Abl regulates the YAP-TEAD functional module. We found that DNA damage, through c-Abl activation, specifically depressed YAP-TEAD-induced transcription. Remarkably, c-Abl counteracts YAP-induced transformation by interfering with the YAP-TEAD transcriptional program. c-Abl induced TEAD1 phosphorylation, but the YAP-TEAD complex remained unaffected. In contrast, TEAD coactivation was compromised by phosphomimetic YAP Y357E mutation but not Y357F, as demonstrated at the level of reporter genes and endogenous TEAD target genes. Furthermore, YAP Y357E also severely compromised the role of YAP in cell transformation, migration, anchorage-independent growth, and epithelial-to-mesenchymal transition (EMT) in human mammary MCF10A cells. These results suggest that YAP pY357 lost TEAD transcription activation function. Our results demonstrate that YAP pY357 inactivates YAP oncogenic function and establish a role for YAP Y357 phosphorylation in cell-fate decision.

[Improvement of emergency physician education through simulator training. Consideration on the basis of the model project "NASimSaar25"]. / Verbesserung der Notarztausbildung durch Simulatortraining. Betrachtung anhand des Modellprojekts "NASimSaar25"

BACKGROUND: Prehospital emergency medicine is a challenge for trainee emergency physicians. Rare injuries and diseases as well as patients in extreme age groups can unexpectedly face emergency physicians. In the regulations on medical education the German Medical Association requires participation in 50 emergency missions under the supervision of an experienced emergency physician. This needs to be improved because on-the-job training does not generally represent the whole spectrum of emergency medicine and a good and structured training under on call conditions is nearly impossible. AIM: The subject of the model project described was whether practical training for emergency physicians can be achieved by participation in simulation training instead of real emergency situations. MATERIAL AND METHODS: After modification of the Saarland regulations on medical education it was possible to replace up to 25 participations in emergency missions by simulation training. The concept of the course NASimSaar25 requires participants to complete 25 simulator cases in 3 days in small training groups. Emergency situations from all medical disciplines need to be treated. A special focus is on the treatment of life-threatening and rare diseases and injuries. Modern simulators and actors are used. The debriefings are conducted by experienced tutors based on approved principles. Medical contents, learning targets from the field of crew resource management (CRM) and soft skills are discussed in these debriefings. RESULTS: Education in the field of emergency medicine can be improved by simulator-based learning and training. However, practical work under a tutor in real and clinical experience cannot be completely replaced by simulation. Simulator training can only be successful if theoretical knowledge has already been acquired. CONCLUSION: A simulator-based course concept can result in an improvement of emergency medical education. The model project NASimSaar25 was well received by the target audience and mostly very well evaluated in terms of learning and reality. If this project becomes established the demand on simulation-based training will increase. The training should achieve a consistent standard of quality.