Explore the vast range of titles available.

eXplainable artificial intelligence (XAI) methods have emerged to convert the black box of machine learning (ML) models into a more digestible form. These methods help to communicate how the model works with the aim of making ML models more transparent and increasing the trust of end‐users in their output. SHapley Additive exPlanations (SHAP) and Local Interpretable Model Agnostic Explanation (LIME) are two widely used XAI methods, particularly with tabular data. In this perspective piece, the way the explainability metrics of these two methods are generated is discussed and a framework for the interpretation of their outputs, highlighting their weaknesses and strengths is proposed. Specifically, their outcomes in terms of model‐dependency and in the presence of collinearity among the features, relying on a case study from the biomedical domain (classification of individuals with or without myocardial infarction) are discussed. The results indicate that SHAP and LIME are highly affected by the adopted ML model and feature collinearity, raising a note of caution on their usage and interpretation. SHapley Additive exPlanations and Local Interpretable Model Agnostic Explanation are two widely used eXplainable artificial intelligence methods. However, they have limitations related to model‐dependency and the presence of collinearity among the features which result in unrealistic explanations. This perspective discusses these two issues through two case studies and provides possible solutions to overcome and eliminate their impacts

ObjectiveTo examine association of COVID-19 with incident cardiovascular events in 17 871 UK Biobank cases between March 2020 and 2021.MethodsCOVID-19 cases were defined using health record linkage. Each case was propensity score-matched to two uninfected controls on age, sex, deprivation, body mass index, ethnicity, diabetes, prevalent ischaemic heart disease (IHD), smoking, hypertension and high cholesterol. We included the following incident outcomes: myocardial infarction, stroke, heart failure, atrial fibrillation, venous thromboembolism (VTE), pericarditis, all-cause death, cardiovascular death, IHD death. Cox proportional hazards regression was used to estimate associations of COVID-19 with each outcome over an average of 141 days (range 32–395) of prospective follow-up.ResultsNon-hospitalised cases (n=14 304) had increased risk of incident VTE (HR 2.74 (95% CI 1.38 to 5.45), p=0.004) and death (HR 10.23 (95% CI 7.63 to 13.70), p<0.0001). Individuals with primary COVID-19 hospitalisation (n=2701) had increased risk of all outcomes considered. The largest effect sizes were with VTE (HR 27.6 (95% CI 14.5 to 52.3); p<0.0001), heart failure (HR 21.6 (95% CI 10.9 to 42.9); p<0.0001) and stroke (HR 17.5 (95% CI 5.26 to 57.9); p<0.0001). Those hospitalised with COVID-19 as a secondary diagnosis (n=866) had similarly increased cardiovascular risk. The associated risks were greatest in the first 30 days after infection but remained higher than controls even after this period.ConclusionsIndividuals hospitalised with COVID-19 have increased risk of incident cardiovascular events across a range of disease and mortality outcomes. The risk of most events is highest in the early postinfection period. Individuals not requiring hospitalisation have increased risk of VTE, but not of other cardiovascular-specific outcomes.

Differences in cardiac and aortic structure and function are associated with cardiovascular diseases and a wide range of other types of disease. Here we analyzed cardiovascular magnetic resonance images from a population-based study, the UK Biobank, using an automated machine-learning-based analysis pipeline. We report a comprehensive range of structural and functional phenotypes for the heart and aorta across 26,893 participants, and explore how these phenotypes vary according to sex, age and major cardiovascular risk factors. We extended this analysis with a phenome-wide association study, in which we tested for correlations of a wide range of non-imaging phenotypes of the participants with imaging phenotypes. We further explored the associations of imaging phenotypes with early-life factors, mental health and cognitive function using both observational analysis and Mendelian randomization. Our study illustrates how population-based cardiac and aortic imaging phenotypes can be used to better define cardiovascular disease risks as well as heart–brain health interactions, highlighting new opportunities for studying disease mechanisms and developing image-based biomarkers. Using magnetic resonance images of the heart and aorta from 26,893 individuals in the UK Biobank, a phenome-wide association study associates cardiovascular imaging phenotypes with a wide range of demographic, lifestyle and clinical features.

Brain iron deposition has been linked to several neurodegenerative conditions and reported in alcohol dependence. Whether iron accumulation occurs in moderate drinkers is unknown. Our objectives were to investigate evidence in support of causal relationships between alcohol consumption and brain iron levels and to examine whether higher brain iron represents a potential pathway to alcohol-related cognitive deficits. Observational associations between brain iron markers and alcohol consumption (n = 20,729 UK Biobank participants) were compared with associations with genetically predicted alcohol intake and alcohol use disorder from 2-sample mendelian randomization (MR). Alcohol intake was self-reported via a touchscreen questionnaire at baseline (2006 to 2010). Participants with complete data were included. Multiorgan susceptibility-weighted magnetic resonance imaging (9.60 ± 1.10 years after baseline) was used to ascertain iron content of each brain region (quantitative susceptibility mapping (QSM) and T2*) and liver tissues (T2*), a marker of systemic iron. Main outcomes were susceptibility (χ) and T2*, measures used as indices of iron deposition. Brain regions of interest included putamen, caudate, hippocampi, thalami, and substantia nigra. Potential pathways to alcohol-related iron brain accumulation through elevated systemic iron stores (liver) were explored in causal mediation analysis. Cognition was assessed at the scan and in online follow-up (5.82 ± 0.86 years after baseline). Executive function was assessed with the trail-making test, fluid intelligence with puzzle tasks, and reaction time by a task based on the \"Snap\" card game. Mean age was 54.8 ± 7.4 years and 48.6% were female. Weekly alcohol consumption was 17.7 ± 15.9 units and never drinkers comprised 2.7% of the sample. Alcohol consumption was associated with markers of higher iron (χ) in putamen (β = 0.08 standard deviation (SD) [95% confidence interval (CI) 0.06 to 0.09], p < 0.001), caudate (β = 0.05 [0.04 to 0.07], p < 0.001), and substantia nigra (β = 0.03 [0.02 to 0.05], p < 0.001) and lower iron in the thalami (β = -0.06 [-0.07 to -0.04], p < 0.001). Quintile-based analyses found these associations in those consuming >7 units (56 g) alcohol weekly. MR analyses provided weak evidence these relationships are causal. Genetically predicted alcoholic drinks weekly positively associated with putamen and hippocampus susceptibility; however, these associations did not survive multiple testing corrections. Weak evidence for a causal relationship between genetically predicted alcohol use disorder and higher putamen susceptibility was observed; however, this was not robust to multiple comparisons correction. Genetically predicted alcohol use disorder was associated with serum iron and transferrin saturation. Elevated liver iron was observed at just >11 units (88 g) alcohol weekly c.f. <7 units (56 g). Systemic iron levels partially mediated associations of alcohol intake with brain iron. Markers of higher basal ganglia iron associated with slower executive function, lower fluid intelligence, and slower reaction times. The main limitations of the study include that χ and T2* can reflect changes in myelin as well as iron, alcohol use was self-reported, and MR estimates can be influenced by genetic pleiotropy. To the best of our knowledge, this study represents the largest investigation of moderate alcohol consumption and iron homeostasis to date. Alcohol consumption above 7 units weekly associated with higher brain iron. Iron accumulation represents a potential mechanism for alcohol-related cognitive decline.

Background Cardiovascular resonance (CMR) imaging is a standard imaging modality for assessing cardiovascular diseases (CVDs), the leading cause of death globally. CMR enables accurate quantification of the cardiac chamber volume, ejection fraction and myocardial mass, providing information for diagnosis and monitoring of CVDs. However, for years, clinicians have been relying on manual approaches for CMR image analysis, which is time consuming and prone to subjective errors. It is a major clinical challenge to automatically derive quantitative and clinically relevant information from CMR images. Methods Deep neural networks have shown a great potential in image pattern recognition and segmentation for a variety of tasks. Here we demonstrate an automated analysis method for CMR images, which is based on a fully convolutional network (FCN). The network is trained and evaluated on a large-scale dataset from the UK Biobank, consisting of 4,875 subjects with 93,500 pixelwise annotated images. The performance of the method has been evaluated using a number of technical metrics, including the Dice metric, mean contour distance and Hausdorff distance, as well as clinically relevant measures, including left ventricle (LV) end-diastolic volume (LVEDV) and end-systolic volume (LVESV), LV mass (LVM); right ventricle (RV) end-diastolic volume (RVEDV) and end-systolic volume (RVESV). Results By combining FCN with a large-scale annotated dataset, the proposed automated method achieves a high performance in segmenting the LV and RV on short-axis CMR images and the left atrium (LA) and right atrium (RA) on long-axis CMR images. On a short-axis image test set of 600 subjects, it achieves an average Dice metric of 0.94 for the LV cavity, 0.88 for the LV myocardium and 0.90 for the RV cavity. The mean absolute difference between automated measurement and manual measurement is 6.1 mL for LVEDV, 5.3 mL for LVESV, 6.9 gram for LVM, 8.5 mL for RVEDV and 7.2 mL for RVESV. On long-axis image test sets, the average Dice metric is 0.93 for the LA cavity (2-chamber view), 0.95 for the LA cavity (4-chamber view) and 0.96 for the RA cavity (4-chamber view). The performance is comparable to human inter-observer variability. Conclusions We show that an automated method achieves a performance on par with human experts in analysing CMR images and deriving clinically relevant measures.

Cardiovascular magnetic resonance (CMR) is the gold standard method for the assessment of cardiac structure and function. Reference ranges permit differentiation between normal and pathological states. To date, this study is the largest to provide CMR specific reference ranges for left ventricular, right ventricular, left atrial and right atrial structure and function derived from truly healthy Caucasian adults aged 45–74. Five thousand sixty-five UK Biobank participants underwent CMR using steady-state free precession imaging at 1.5 Tesla. Manual analysis was performed for all four cardiac chambers. Participants with non-Caucasian ethnicity, known cardiovascular disease and other conditions known to affect cardiac chamber size and function were excluded. Remaining participants formed the healthy reference cohort; reference ranges were calculated and were stratified by gender and age (45–54, 55–64, 65–74). After applying exclusion criteria, 804 (16.2%) participants were available for analysis. Left ventricular (LV) volumes were larger in males compared to females for absolute and indexed values. With advancing age, LV volumes were mostly smaller in both sexes. LV ejection fraction was significantly greater in females compared to males (mean ± standard deviation [SD] of 61 ± 5% vs 58 ± 5%) and remained static with age for both genders. In older age groups, LV mass was lower in men, but remained virtually unchanged in women. LV mass was significantly higher in males compared to females (mean ± SD of 53 ± 9 g/m2 vs 42 ± 7 g/m2). Right ventricular (RV) volumes were significantly larger in males compared to females for absolute and indexed values and were smaller with advancing age. RV ejection fraction was higher with increasing age in females only. Left atrial (LA) maximal volume and stroke volume were significantly larger in males compared to females for absolute values but not for indexed values. LA ejection fraction was similar for both sexes. Right atrial (RA) maximal volume was significantly larger in males for both absolute and indexed values, while RA ejection fraction was significantly higher in females. We describe age- and sex-specific reference ranges for the left ventricle, right ventricle and atria in the largest validated normal Caucasian population.

Medical imaging provides numerous insights into the subclinical changes that precede serious diseases such as heart disease and dementia. However, most imaging research either describes a single organ system or draws on clinical cohorts with small sample sizes. In this study, we use state-of-the-art multi-organ magnetic resonance imaging phenotypes to investigate cross-sectional relationships across the heart-brain-liver axis in 30,444 UK Biobank participants. Despite controlling for an extensive range of demographic and clinical covariates, we find significant associations between imaging-derived phenotypes of the heart (left ventricular structure, function and aortic distensibility), brain (brain volumes, white matter hyperintensities and white matter microstructure), and liver (liver fat, liver iron and fibroinflammation). Simultaneous three-organ modelling identifies differentially important pathways across the heart-brain-liver axis with evidence of both direct and indirect associations. This study describes a potentially cumulative burden of multiple-organ dysfunction and provides essential insight into multi-organ disease prevention. While heart disease, dementia and liver disease often co-occur, multi-organ imaging is needed for deeper elucidation of these cross-organ links. Here, the authors use image-derived phenotypes to describe underlying associations between heart, brain and liver health in a large population cohort.

To define the sex, age, and disease-specific associations of resting heart rate (RHR) with cardiovascular and mortality outcomes in 502,534 individuals from the UK Biobank over 7-12 years of prospective follow-up. The main outcomes were all-cause, cardiovascular, and ischaemic heart disease mortality. Additional outcomes included incident acute myocardial infarction (AMI), fatal AMI, and cancer mortality. We considered a wide range of confounders and the effects of competing hazards. Results are reported as hazard ratios (HR) for all-cause mortality and sub-distribution hazard ratios (SHR) for other outcomes with corresponding 95% confidence intervals (CI) per 10bpm increment of RHR. In men, for every 10bpm increase of RHR there was 22% (HR 1.22, CI 1.20 to 1.24, p = 3×10-123) greater hazard of all-cause and 17% (SHR 1.17, CI 1.13 to 1.21, p = 5.6×10-18) greater hazard of cardiovascular mortality; for women, corresponding figures were 19% (HR 1.19, CI 1.16 to 1.22, p = 8.9×10-45) and 14% (SHR 1.14, CI 1.07 to 1.22, p = 0.00008). Associations between RHR and ischaemic outcomes were of greater magnitude amongst men than women, but with similar magnitude of association for non-cardiovascular cancer mortality [men (SHR 1.18, CI 1.15-1.21, p = 5.2×10-46); women 15% (SHR 1.15, CI 1.11-1.18, p = 3.1×10-18)]. Associations with all-cause, incident AMI, and cancer mortality were of greater magnitude at younger than older ages. RHR is an independent predictor of mortality, with variation by sex, age, and disease. Ischaemic disease appeared a more important driver of this relationship in men, and associations were more pronounced at younger ages.

The UK Biobank is a large-scale population-based study utilising cardiovascular magnetic resonance (CMR) to generate measurements of atrial and ventricular structure and function. This study aimed to quantify the association between modifiable cardiovascular risk factors and cardiac morphology and function in individuals without known cardiovascular disease. Age, sex, ethnicity (non-modifiable) and systolic blood pressure, diastolic blood pressure, smoking status, exercise, body mass index (BMI), high cholesterol, diabetes, alcohol intake (modifiable) were considered important cardiovascular risk factors. Multivariable regression models were built to ascertain the association of risk factors on left ventricular (LV), right ventricular (RV), left atrial (LA) and right atrial (RA) CMR parameters. 4,651 participants were included in the analysis. All modifiable risk factors had significant effects on differing atrial and ventricular parameters. BMI was the modifiable risk factor most consistently associated with subclinical changes to CMR parameters, particularly in relation to higher LV mass (+8.3% per SD [4.3 kg/m2], 95% CI: 7.6 to 8.9%), LV (EDV: +4.8% per SD, 95% CI: 4.2 to 5.4%); ESV: +4.4% per SD, 95% CI: 3.5 to 5.3%), RV (EDV: +5.3% per SD, 95% CI: 4.7 to 5.9%; ESV: +5.4% per SD, 95% CI: 4.5 to 6.4%) and LA maximal (+8.6% per SD, 95% CI: 7.4 to 9.7%) volumes. Increases in SBP were associated with higher LV mass (+6.8% per SD, 95% CI: 5.9 to 7.7%), LV (EDV: +4.5% per SD, 95% CI: 3.6 to 5.4%; ESV: +2.0% per SD, 95% CI: 0.8 to 3.3%) volumes. The presence of diabetes or high cholesterol resulted in smaller volumes and lower ejection fractions. Modifiable risk factors are associated with subclinical alterations in structure and function in all four cardiac chambers. BMI and systolic blood pressure are the most important modifiable risk factors affecting CMR parameters known to be linked to adverse outcomes.

Recent evidence suggests that shorter telomere length (TL) is associated with neuro degenerative diseases and aging related outcomes. The causal association between TL and brain characteristics represented by image derived phenotypes (IDPs) from different magnetic resonance imaging (MRI) modalities remains unclear. Here, we use two-sample Mendelian randomization (MR) to systematically assess the causal relationships between TL and 3,935 brain IDPs. Overall, the MR results suggested that TL was causally associated with 193 IDPs with majority representing diffusion metrics in white matter tracts. 68 IDPs were negatively associated with TL indicating that longer TL causes decreasing in these IDPs, while the other 125 were associated positively (longer TL leads to increased IDPs measures). Among them, ten IDPs have been previously reported as informative biomarkers to estimate brain age. However, the effect direction between TL and IDPs did not reflect the observed direction between aging and IDPs: longer TL was associated with decreases in fractional anisotropy and increases in axial, radial and mean diffusivity. For instance, TL was positively associated with radial diffusivity in the left perihippocampal cingulum tract and with mean diffusivity in right perihippocampal cingulum tract. Our results revealed a causal role of TL on white matter integrity which makes it a valuable factor to be considered when brain age is estimated and investigated.