Publications
126 results found
Stowell C, Howard J, Cole G, et al., 2021, Automated left ventricular dimension assessment using artificial intelligence, Publisher: OXFORD UNIV PRESS, Pages: 1-1, ISSN: 0195-668X
Naderi H, Robinson S, Swaans MJ, et al., 2021, Adapting the role of handheld echocardiography during the COVID-19 pandemic: A practical guide, PERFUSION-UK, Vol: 36, Pages: 547-558, ISSN: 0267-6591
- Author Web Link
- Cite
- Citations: 1
Banerjee M, Chiew D, Patel K, et al., 2021, The impact of artificial intelligence on clinical education: Perceptions of postgraduate trainee doctors in London (UK) and recommendations for trainers., BMC Medical Education, Vol: 21, Pages: 1-10, ISSN: 1472-6920
BackgroundArtificial intelligence (AI) technologies are increasingly used in clinical practice. Although there is robust evidence that AI innovations can improve patient care, reduce clinicians’ workload and increase efficiency, their impact on medical training and education remains unclear.MethodsA survey of trainee doctors’ perceived impact of AI technologies on clinical training and education was conducted at UK NHS postgraduate centers in London between October and December 2020. Impact assessment mirrored domains in training curricula such as ‘clinical judgement’, ‘practical skills’ and ‘research and quality improvement skills’. Significance between Likert-type data was analysed using Fisher’s exact test. Response variations between clinical specialities were analysed using k-modes clustering. Free-text responses were analysed by thematic analysis.Results210 doctors responded to the survey (response rate 72%). The majority (58%) perceived an overall positive impact of AI technologies on their training and education. Respondents agreed that AI would reduce clinical workload (62%) and improve research and audit training (68%). Trainees were skeptical that it would improve clinical judgement (46% agree, p=0.12) and practical skills training (32% agree, p<0.01). The majority reported insufficient AI training in their current curricula (92%), and supported having more formal AI training (81%).ConclusionsTrainee doctors have an overall positive perception of AI technologies’ impact on clinical training. There is optimism that it will improve ‘research and quality improvement’ skills and facilitate ‘curriculum mapping’. There is skepticism that it may reduce educational opportunities to develop ‘clinical judgement’ and ‘practical skills’. Medical educators should be mindful that these domains are protected as AI develops. We recommend that ‘Applied AI&r
Kotecha T, Knight DS, Moon JC, et al., 2021, The evolution of cardiovascular COVID-19 research, EUROPEAN HEART JOURNAL, Vol: 42, Pages: 2953-2954, ISSN: 0195-668X
Rajkumar C, Shun-Shin M, Seligman H, et al., 2021, Placebo-controlled efficacy of percutaneous coronary intervention for focal and diffuse patterns of stable coronary artery disease, Circulation: Cardiovascular Interventions, Vol: 14, Pages: 809-818, ISSN: 1941-7640
Background Physiological assessment with pressure wire pullback can characterize coronary artery disease (CAD) with a focal or diffuse pattern. However, the clinical relevance of this distinction is unknown. We use data from ORBITA to test if the pattern of CAD predicts the placebo-controlled efficacy of percutaneous coronary intervention (PCI) on stress echocardiography ischemia and symptom endpoints.Methods164 patients in ORBITA underwent blinded instantaneous wave-free ratio (iFR) pullback assessment prior to randomization. Focal disease was defined as 0.03 iFR unit drop within 15mm, rather than over a longer distance. Analyses were performed using regression modelling. ResultsIn the PCI arm (n=85), 48 were focal and 37 were diffuse. In the placebo arm (n=79), 35 were focal and 44 were diffuse. Focal stenoses were associated with significantly lower fractional flow reserve (FFR) and iFR values than diffusely diseased vessels (focal mean FFR and iFR 0.600.15 and 0.650.24, diffuse 0.780.10 and 0.880.08 respectively, p<0.0001). With adjustment for this difference, PCI for focal stenoses resulted in significantly greater reduction in stress echo ischemia than PCI for diffuse disease (p<0.05). The effect of PCI on between-arm pre-randomization-adjusted exercise time was 9.32 seconds (95% CI, -17.1 to 35.7s; p=0.487). When stratified for pattern of disease, there was no detectable difference between focal and diffuse CAD (Pinteraction=0.700). PCI improved Seattle Angina Questionnaire angina frequency score and freedom from angina more than placebo (p=0.034; p=0.0035). However, there was no evidence of interaction between the physiological pattern of CAD and these effects (Pinteraction=0.436; Pinteraction=0.908).ConclusionPCI achieved significantly greater reduction of stress echocardiography ischemia in focal compared to diffuse CAD. However, for symptom endpoints, no such difference was observed.
Levy S, Cole G, Pabari P, et al., 2021, New horizons in cardiogeriatrics: geriatricians and heart failure care-the custard in the tart, not the icing on the cake, AGE AND AGEING, Vol: 50, Pages: 1064-1068, ISSN: 0002-0729
Gorecka M, McCann GP, Berry C, et al., 2021, Demographic, multi-morbidity and genetic impact on myocardial involvement and its recovery from COVID-19: protocol design of COVID-HEART-a UK, multicentre, observational study, JOURNAL OF CARDIOVASCULAR MAGNETIC RESONANCE, Vol: 23, ISSN: 1097-6647
- Author Web Link
- Cite
- Citations: 13
Lane ES, Azarmehr N, Jevsikov J, et al., 2021, Multibeat echocardiographic phase detection using deep neural networks, COMPUTERS IN BIOLOGY AND MEDICINE, Vol: 133, ISSN: 0010-4825
- Author Web Link
- Cite
- Citations: 7
Howard J, Stowell C, Cole G, et al., 2021, Automated left ventricular dimension assessment using artificial intelligence developed and validated by a UK-wide collaborative, Circulation: Cardiovascular Imaging, Vol: 14, Pages: 405-415, ISSN: 1941-9651
Background: Echocardiography artificial intelligence (AI) requires training and validation to standards expected of humans. We developed an online platform and established the Unity Collaborative to build a dataset of expertise from 17 hospitals for training, validation, and standardisation of such techniques. Methods: The training dataset were 2056individual frames drawn at random from 1265parasternal long-axis video-loops of patients undergoing clinical echocardiography in 2015-2016. Nine experts labelled these images using our online platform. From this, we trained a convolutional neural network to identify key points. Subsequently, 13 experts labelled a validation dataset of the end-systolic and end-diastolic frame from100new video-loops, twice each. The 26-opinionconsensus was used as the reference standard. The primary outcome was “precision SD”, the standard deviation of difference between AI measurement and expert consensus. Results: In the validation dataset, the AI’s precision SD for left ventricular internal dimension was 3.5mm. For context, precision SD of individual expert measurements against the expert consensus was 4.4mm. Intraclass correlation coefficient (ICC) between AI and expert consensus was 0.926 (95% CI 0.904–0.944), compared with 0.817 (0.778–0.954) between individual experts and expert consensus. For interventricular septum thickness, precision SD was 1.8mm for AI (ICC 0.809; 0.729–0.967), versus 2.0 for individuals (ICC 0.641; 0.568–0.716). For posterior wall thickness, precision SD was 1.4mm for AI (ICC 0.535; 95% CI 0.379–0.661), versus 2.2mm for individuals(0.366; 0.288 to 0.462).We present all images and annotations. This highlights challenging cases, including poor image quality, tapered ventricles, and indistinct boundaries. Conclusions: Experts at multiple institutions successfully cooperated to build a collaborative AI. This performed as well as individual experts. Future echocardiogr
Kotecha T, Knight DS, Razvi Y, et al., 2021, Patterns of myocardial injury in recovered troponin-positive COVID-19 patients assessed by cardiovascular magnetic resonance, EUROPEAN HEART JOURNAL, Vol: 42, Pages: 1866-1878, ISSN: 0195-668X
- Author Web Link
- Cite
- Citations: 1
Zaman S, Seligman H, Lloyd FH, et al., 2021, Aerosolised fluorescein can quantify FFP mask faceseal leakage: a cost-effective adaptation to the existing point of care fit test, CLINICAL MEDICINE, Vol: 21, Pages: E263-E268, ISSN: 1470-2118
- Author Web Link
- Cite
- Citations: 1
Azarmehr N, Ye X, Howard JP, et al., 2021, Neural architecture search of echocardiography view classifiers, JOURNAL OF MEDICAL IMAGING, Vol: 8, ISSN: 2329-4302
- Author Web Link
- Cite
- Citations: 4
Mikhail G, Khawaja SA, Mohan P, et al., 2021, COVID-19 and its impact on the cardiovascular system, Open Heart, Vol: 8, Pages: 1-9, ISSN: 2053-3624
Objectives: The clinical impact of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has varied across countries with varying cardiovascular manifestations. We review the cardiac presentations, in-hospital outcomes and development of cardiovascular complications in the initial cohort of SARS-CoV-2 positive patients at Imperial College Healthcare NHS Trust, United Kingdom.Methods: We retrospectively analysed 498 COVID-19 positive adult admissions to our institute from 7th March to 7th April 2020. Patient data was collected for baseline demographics, co-morbidities and in-hospital outcomes, especially relating to cardiovascular intervention.Results:Mean age was 67.4±16.1 years and 62.2%(n=310) were male. 64.1%(n=319) of our cohort had underlying cardiovascular disease (CVD) with 53.4%(n=266) having hypertension. 43.2%(n=215) developed acute myocardial injury. Mortality was significantly increased in those patients with myocardial injury (47.4% vs 18.4%,p<0.001). Only 4 COVID-19 patients had invasive coronary angiography,2 underwent percutaneous coronary intervention and 1 required a permanent pacemaker implantation. 7.0%(n=35) of patients had an inpatient echocardiogram. Acute myocardial injury (OR 2.39,1.31-4.40,p=0.005) and history of hypertension (OR 1.88 ,1.01-3.55,p=0.049) approximately doubled the odds of in-hospital mortality in patients admitted with COVID-19 after other variables had been controlled for.Conclusion:Hypertension, pre-existing CVD and acute myocardial injury were associated with increased in-hospital mortality in our cohort of COVID-19 patients. However, only a low number of patients required invasive cardiac intervention.
Sweeney M, Cole GD, Pabari P, et al., 2021, Urinary drug metabolite testing in chronic heart failure patients indicates high levels of adherence with life-prolonging therapies, ESC Heart Failure, Vol: 8, Pages: 2334-2337, ISSN: 2055-5822
AimsDespite medical therapy for heart failure (HF) having proven benefits of improving quality of life and survival, many patients remain under-treated. This may be due to a combination of under-prescription by medical professionals and poor adherence from patients. In HF, as with many other chronic diseases, adherence to medication can deteriorate over time particularly when symptoms are well controlled. Therefore, detecting and addressing non-adherence has a crucial role in the management of HF. Significant flaws and inaccuracies exist in the methods currently used to assess adherence such as patient reporting, pill counts, and pharmacy fill records. We aim to use high-performance liquid chromatography–tandem mass spectrometry (HPLC-MS) to detect metabolites of HF medications in the urine samples of chronic HF patients.Methods and resultsUrine samples were collected from 35 patients in a specialist HF clinic. Patients were included if they had an ejection fraction <45% and were taking at least two disease-modifying HF medications. They were excluded if they had been admitted to hospital for HF in the 3 months preceding clinic attendance. These samples were sent for HPLC-MS and tested for all HF medications prescribed for that patient. A high rate of complete adherence of 89% was detected in these patients, with 94% being partially adherent (at least one HF medication detected) to therapy (at least one HF medication detected). This analysis also highlighted that mineralocorticoid antagonists represent both the most under-prescribed (67%) and poorly adhered (75%) medication class.ConclusionsThis analysis revealed a surprisingly high level of adherence to disease-modifying therapy in chronic HF patients and highlights that most of our ‘total’ under-treatment is likely to be from a failure to prescribe rather than a failure to adhere. Testing for metabolites of disease-modifying HF drugs in urine using HPLC-MS is feasible and is a useful adjunct t
Cox-Smith A, Cooper T, Punjabi P, et al., 2021, LACK OF EVIDENCE FOR REDUCED EFFICACY OF MEDICAL THERAPY FOR HEART FAILURE IN OLDER ADULTS, Publisher: OXFORD UNIV PRESS, ISSN: 0002-0729
Sivalokanathan S, Foley M, Cole G, et al., 2021, Gastroenteritis and cardiogenic shock in a healthcare worker: a case report of COVID-19 myocarditis confirmed with serology, European Heart Journal: Case Reports, Vol: 5, Pages: 1-5, ISSN: 2514-2119
BackgroundCoronavirus disease 2019 (COVID-19) myocarditis is emerging as a component of the hyperactive inflammatory response secondary to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Isolated gastrointestinal symptoms are uncommon presenting features in adults with COVID-19 myocarditis. The availability of antibody testing is a valuable addition to the confirmation of COVID-19, when repeated reverse transcriptase–polymerase chain reaction of nasopharyngeal swabs are negative.Case summaryA young healthcare worker presented with dizziness and pre-syncope, 4 weeks after his original symptoms that included fever, lethargy, and diarrhoea. Despite 2 weeks of isolation, followed by a quiescent spell, his symptoms had returned. Shortly after, he presented in cardiogenic shock (left ventricular ejection fraction 25%), that required vasopressor support, at the height of the COVID-19 pandemic. Cardiac magnetic resonance imaging suggested florid myocarditis. Three nasopharyngeal swabs (Days 1, 3, and 5) were negative for SARS-CoV-2, but subsequent serology (Day 13) confirmed the presence of SARS-CoV-2 IgG. Treatment with intravenous immunoglobulin and glucocorticoids led to full recovery.DiscussionOur case study highlights the significance of the use of the available serological assays for diagnosis of patients presenting late with SARS-CoV-2. Importantly, it supports further research in the use of immunomodulatory drugs for the hyperinflammatory microenvironment induced by COVID-19.
Chatrath N, Kaza N, Pabari PA, et al., 2020, The effect of concomitant COVID-19 infection on outcomes in patients hospitalized with heart failure, ESC HEART FAILURE, Vol: 7, Pages: 4443-4447, ISSN: 2055-5822
- Author Web Link
- Cite
- Citations: 19
Howard JP, Zaman S, Ragavan A, et al., 2020, Automated analysis and detection of abnormalities in transaxial anatomical cardiovascular magnetic resonance images: a proof of concept study with potential to optimize image acquisition, International Journal of Cardiovascular Imaging, Vol: 37, Pages: 1033-1042, ISSN: 1569-5794
The large number of available MRI sequences means patients cannot realistically undergo them all, so the range of sequences to be acquired during a scan are protocolled based on clinical details. Adapting this to unexpected findings identified early on in the scan requires experience and vigilance. We investigated whether deep learning of the images acquired in the first few minutes of a scan could provide an automated early alert of abnormal features. Anatomy sequences from 375 CMR scans were used as a training set. From these, we annotated 1500 individual slices and used these to train a convolutional neural network to perform automatic segmentation of the cardiac chambers, great vessels and any pleural effusions. 200 scans were used as a testing set. The system then assembled a 3D model of the thorax from which it made clinical measurements to identify important abnormalities. The system was successful in segmenting the anatomy slices (Dice 0.910) and identified multiple features which may guide further image acquisition. Diagnostic accuracy was 90.5% and 85.5% for left and right ventricular dilatation, 85% for left ventricular hypertrophy and 94.4% for ascending aorta dilatation. The area under ROC curve for diagnosing pleural effusions was 0.91. We present proof-of-concept that a neural network can segment and derive accurate clinical measurements from a 3D model of the thorax made from transaxial anatomy images acquired in the first few minutes of a scan. This early information could lead to dynamic adaptive scanning protocols, and by focusing scanner time appropriately and prioritizing cases for supervision and early reporting, improve patient experience and efficiency.
Rajkumar C, Shun-Shin M, Seligman H, et al., 2020, Placebo-Controlled Efficacy of Percutaneous Coronary Intervention for Focal and Diffuse Patterns of Stable Coronary Artery Disease: A Secondary Analysis From ORBITA, 32nd Annual Transcatheter Cardiovascular Therapeutics Symposium (TCT CONNECT), Publisher: ELSEVIER SCIENCE INC, Pages: B165-B165, ISSN: 0735-1097
Sweeney M, Cole GD, Hadjiphilippou S, et al., 2020, It's us, not them: first use of urinary metabolite testing in heart failure patients indicates high level of adherence with life-prolonging therapies, Publisher: WILEY, Pages: 240-240, ISSN: 1388-9842
Bampouri T, Gardezi SKM, Hadjiphilippou S, et al., 2020, Hypochloraemia is associated with mortality in patients admitted with heart failure, Publisher: WILEY, Pages: 32-33, ISSN: 1388-9842
Howard JP, Tan J, Shun-Shin MJ, et al., 2020, Improving ultrasound video classification: an evaluation of novel deep learning methods in echocardiography., J Med Artif Intell, Vol: 3
Echocardiography is the commonest medical ultrasound examination, but automated interpretation is challenging and hinges on correct recognition of the 'view' (imaging plane and orientation). Current state-of-the-art methods for identifying the view computationally involve 2-dimensional convolutional neural networks (CNNs), but these merely classify individual frames of a video in isolation, and ignore information describing the movement of structures throughout the cardiac cycle. Here we explore the efficacy of novel CNN architectures, including time-distributed networks and two-stream networks, which are inspired by advances in human action recognition. We demonstrate that these new architectures more than halve the error rate of traditional CNNs from 8.1% to 3.9%. These advances in accuracy may be due to these networks' ability to track the movement of specific structures such as heart valves throughout the cardiac cycle. Finally, we show the accuracies of these new state-of-the-art networks are approaching expert agreement (3.6% discordance), with a similar pattern of discordance between views.
Chacko L, P Howard J, Rajkumar C, et al., 2020, Effects of percutaneous coronary intervention on death and myocardial infarction stratified by stable and unstable coronary artery disease: a meta-analysis of randomized controlled trials, Circulation: Cardiovascular Quality and Outcomes, Vol: 13, ISSN: 1941-7705
Background:In patients presenting with ST-segment–elevation myocardial infarction, percutaneous coronary intervention (PCI) reduces mortality when compared with fibrinolysis. In other forms of coronary artery disease (CAD), however, it has been controversial whether PCI reduces mortality. In this meta-analysis, we examine the benefits of PCI in (1) patients post–myocardial infarction (MI) who did not receive immediate revascularization; (2) patients who have undergone primary PCI for ST-segment–elevation myocardial infarction but have residual coronary lesions; (3) patients who have suffered a non–ST-segment–elevation acute coronary syndrome; and (4) patients with truly stable CAD with no recent infarct. This analysis includes data from the recently presented International Study of Comparative Health Effectiveness with Medical and Invasive Approaches (ISCHEMIA) and Complete versus Culprit-Only Revascularization Strategies to Treat Multivessel Disease after Early PCI for STEMI (COMPLETE) trials.Methods and Results:We systematically identified all randomized trials of PCI on a background of medical therapy for the treatment of CAD. The ISCHEMIA trial, presented in November 2019, was eligible for inclusion. Data were combined using a random-effects meta-analysis. The primary end point was all-cause mortality. Forty-six trials, including 37 757 patients, were eligible. In the 3 unstable scenarios, PCI had the following effects on mortality: unrevascularized post-MI relative risk (RR) 0.68 (95% CI, 0.45–1.03); P=0.07; multivessel disease following ST-segment–elevation myocardial infarction (RR, 0.84 [95% CI, 0.69–1.04]; P=0.11); non–ST-segment–elevation acute coronary syndrome (RR, 0.84 [95% CI, 0.72–0.97]; P=0.02). Overall, in these unstable scenarios PCI was associated with a significant reduction in mortality (RR, 0.84 [95% CI, 0.75–0.93]; P=0.02). In unstable CAD, PCI also reduced cardiac
Tarkin JM, Cole GD, Gopalan D, et al., 2020, Multimodal imaging of granulomatosis with polyangiitis aortitis complicated by severe aortic regurgitation and complete heart block, Circulation: Cardiovascular Imaging, Vol: 13, Pages: 1-3, ISSN: 1941-9651
Al-lamee RK, Shun-Shin M, Howard J, et al., 2019, Dobutamine Stress Echocardiography Ischaemia as a Predictor of the Placebo-Controlled Efficacy of Percutaneous Coronary Intervention in Stable Coronary Artery Disease: The Stress Echo-Stratified Analysis of ORBITA, Resuscitation Science Symposium (ReSS), Publisher: LIPPINCOTT WILLIAMS & WILKINS, Pages: E985-E985, ISSN: 0009-7322
Patel RK, Moore AM, Piper S, et al., 2019, Clozapine and cardiotoxicity - A guide for psychiatrists written by cardiologists, Psychiatry Research, Vol: 282, Pages: 1-8, ISSN: 0165-1781
This review discusses the rare but potentially life-threatening cardiovascular side-effects of myocarditis and dilated cardiomyopathy associated with the use of Clozapine. The clinical presentation of these conditions is non-specific, making it difficult to both risk-stratify and identify patients who develop these consequences. This review aims to examine the proposed aetiologies, diagnostic approaches and subsequent management strategies of cardiotoxicity associated with clozapine use; offering guidance to psychiatrists and general physicians. Current evidence highlights the importance of accurate diagnosis to prevent premature and unnecessary cessation of clozapine. Guidance on monitoring and reintroduction of the drug is emerging and current practice recommends a combination of regular monitoring of biomarkers and imaging to make a diagnosis of cardiotoxicity although further work is needed to establish evidence-based guidelines.
Al-Lamee R, Shun-Shin M, Howard J, et al., 2019, Dobutamine stress echocardiography ischemia as a predictor of the placebo-controlled efficacy of percutaneous coronary intervention in stable coronary artery disease: the stress echo-stratified analysis of ORBITA, Circulation, Vol: 140, Pages: 1971-1980, ISSN: 0009-7322
BackgroundDobutamine stress echocardiography (DSE) is widely used to test for ischemia in patients with stable coronary artery disease (CAD). In this analysis we studied the ability of pre-randomization stress echo score to predict the placebo-controlled efficacy of percutaneous coronary intervention (PCI) within the ORBITA trial. MethodsOne hundred and eighty-three patients underwent DSE before randomization. The stress echo score is broadly the number of segments abnormal at peak stress, with akinetic segments counting double and dyskinetic segments counting triple. The ability of pre-randomization stress echo to predict the placebo-controlled effect of PCI on response variables was tested using regression modelling.ResultsAt pre-randomization, the stress echo score was 1.561.77 in the PCI arm (n=98) and 1.611.73 in the placebo arm (n=85). There was a detectable interaction between pre-randomization stress echo score and the effect of PCI on angina frequency score with a larger placebo-controlled effect in patients with the highest stress echo score (pinteraction=0.031). With our sample size we were unable to detect an interaction between stress echo score and any other patient-reported response variables: freedom from angina (pinteraction=0.116), physical limitation (pinteraction=0.461), quality of life (pinteraction=0.689), EQ-5D-5L quality of life score (pinteraction=0.789) or between stress echo score and physician-assessed Canadian Cardiovascular Society angina class (pinteraction=0.693), and treadmill exercise time (pinteraction=0.426). ConclusionsThe degree of ischemia assessed by DSE predicts the placebo-controlled efficacy of PCI on patient-reported angina frequency. The greater the downstream stress echo abnormality caused by a stenosis, the greater the reduction in symptoms from PCI.
Arnold A, Howard J, Chiew K, et al., 2019, Right ventricular pacing for hypertrophic obstructive cardiomyopathy: meta-analysis and meta-regression of clinical trials, European Heart Journal - Quality of Care and Clinical Outcomes, Vol: 5, Pages: 321-333, ISSN: 2058-5225
AimsRight ventricular pacing for left ventricular outflow tract gradient reduction in hypertrophic obstructive cardiomyopathy remains controversial. We undertook a meta-analysis for echocardiographic and functional outcomes.Methods and resultsThirty-four studies comprising 1135 patients met eligibility criteria. In the four blinded randomized controlled trials (RCTs), pacing reduced gradient by 35% [95% confidence interval (CI) 23.2–46.9, P < 0.0001], but there was only a trend towards improved New York Heart Association (NYHA) class [odds ratio (OR) 1.82, CI 0.96–3.44; P = 0.066]. The unblinded observational studies reported a 54.3% (CI 44.1–64.6, P < 0.0001) reduction in gradient, which was a 18.6% greater reduction than the RCTs (P = 0.0351 for difference between study designs). Observational studies reported an effect on unblinded NYHA class at an OR of 8.39 (CI 4.39–16.04, P < 0.0001), 450% larger than the OR in RCTs (P = 0.0042 for difference between study designs). Across all studies, the gradient progressively decreased at longer follow durations, by 5.2% per month (CI 2.5–7.9, P = 0.0001).ConclusionRight ventricular pacing reduces gradient in blinded RCTs. There is a non-significant trend to reduction in NYHA class. The bias in assessment of NYHA class in observational studies appears to be more than twice as large as any genuine treatment effect.
Bhuva AN, Bai W, Lau C, et al., 2019, A multicenter, scan-rescan, human and machine learning CMR study to test generalizability and precision in imaging biomarker analysis, Circulation: Cardiovascular Imaging, Vol: 12, Pages: 1-11, ISSN: 1941-9651
Background:Automated analysis of cardiac structure and function using machine learning (ML) has great potential, but is currently hindered by poor generalizability. Comparison is traditionally against clinicians as a reference, ignoring inherent human inter- and intraobserver error, and ensuring that ML cannot demonstrate superiority. Measuring precision (scan:rescan reproducibility) addresses this. We compared precision of ML and humans using a multicenter, multi-disease, scan:rescan cardiovascular magnetic resonance data set.Methods:One hundred ten patients (5 disease categories, 5 institutions, 2 scanner manufacturers, and 2 field strengths) underwent scan:rescan cardiovascular magnetic resonance (96% within one week). After identification of the most precise human technique, left ventricular chamber volumes, mass, and ejection fraction were measured by an expert, a trained junior clinician, and a fully automated convolutional neural network trained on 599 independent multicenter disease cases. Scan:rescan coefficient of variation and 1000 bootstrapped 95% CIs were calculated and compared using mixed linear effects models.Results:Clinicians can be confident in detecting a 9% change in left ventricular ejection fraction, with greater than half of coefficient of variation attributable to intraobserver variation. Expert, trained junior, and automated scan:rescan precision were similar (for left ventricular ejection fraction, coefficient of variation 6.1 [5.2%–7.1%], P=0.2581; 8.3 [5.6%–10.3%], P=0.3653; 8.8 [6.1%–11.1%], P=0.8620). Automated analysis was 186× faster than humans (0.07 versus 13 minutes).Conclusions:Automated ML analysis is faster with similar precision to the most precise human techniques, even when challenged with real-world scan:rescan data. Assessment of multicenter, multi-vendor, multi-field strength scan:rescan data (available at www.thevolumesresource.com) permits a generalizable assessment of ML precision and may facili
Bhuva A, Bai W, Lau C, et al., 2019, Fully automated left ventricular analysis matches clinician precision: a multi-centre, multi-vendor, multi-field strength, multi-disease scan:rescan CMR study, Publisher: OXFORD UNIV PRESS, Pages: 255-256, ISSN: 2047-2404
This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.