scholarly journals Machine learning approach to dynamic risk modeling of mortality in COVID-19: a UK Biobank study

2021 ◽  
Author(s):  
Mohammad A. Dabbah ◽  
Angus B. Reed ◽  
Adam T.C. Booth ◽  
Arrash Yassaee ◽  
Alex Despotovic ◽  
...  
Keyword(s):  
High Risk ◽  
Mortality Risk ◽  
Prediction Models ◽  
Vital Signs ◽  
Classification Model ◽  
Risk Modeling ◽  
Uk Biobank ◽  
Individual Level ◽  
Random Forest Classification ◽  
The Uk

Abstract The COVID-19 pandemic has created an urgent need for robust, scalable monitoring tools supporting stratification of high-risk patients. This research aims to develop and validate prediction models, using the UK Biobank, to estimate COVID-19 mortality risk in confirmed cases. From the 11,245 participants testing positive for COVID-19, we develop a data-driven random forest classification model with excellent performance (AUC: 0.91), using baseline characteristics, pre-existing conditions, symptoms, and vital signs, such that the score could dynamically assess mortality risk with disease deterioration. We also identify several significant novel predictors of COVID-19 mortality with equivalent or greater predictive value than established high-risk comorbidities, such as detailed anthropometrics and prior acute kidney failure, urinary tract infection, and pneumonias. The model design and feature selection enables utility in outpatient settings. Possible applications include supporting individual-level risk profiling and monitoring disease progression across patients with COVID-19 at-scale, especially in hospital-at-home settings.

scholarly journals Machine learning approach to dynamic risk modeling of mortality in COVID-19: a UK Biobank study

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Mohammad A. Dabbah ◽  
Angus B. Reed ◽  
Adam T. C. Booth ◽  
Arrash Yassaee ◽  
Aleksa Despotovic ◽  
...  
Keyword(s):  
High Risk ◽  
Mortality Risk ◽  
Prediction Models ◽  
Vital Signs ◽  
Classification Model ◽  
Risk Modeling ◽  
Uk Biobank ◽  
Individual Level ◽  
Random Forest Classification ◽  
The Uk

AbstractThe COVID-19 pandemic has created an urgent need for robust, scalable monitoring tools supporting stratification of high-risk patients. This research aims to develop and validate prediction models, using the UK Biobank, to estimate COVID-19 mortality risk in confirmed cases. From the 11,245 participants testing positive for COVID-19, we develop a data-driven random forest classification model with excellent performance (AUC: 0.91), using baseline characteristics, pre-existing conditions, symptoms, and vital signs, such that the score could dynamically assess mortality risk with disease deterioration. We also identify several significant novel predictors of COVID-19 mortality with equivalent or greater predictive value than established high-risk comorbidities, such as detailed anthropometrics and prior acute kidney failure, urinary tract infection, and pneumonias. The model design and feature selection enables utility in outpatient settings. Possible applications include supporting individual-level risk profiling and monitoring disease progression across patients with COVID-19 at-scale, especially in hospital-at-home settings.

scholarly journals 699 Sleep Health Traits and COVID-19: Mortality Risk from the UK Biobank

SLEEP ◽  
2021 ◽  
Vol 44 (Supplement_2) ◽  
pp. A273-A273
Author(s):  
Xi Zheng ◽  
Ma Cherrysse Ulsa ◽  
Peng Li ◽  
Lei Gao ◽  
Kun Hu
Keyword(s):  
Risk Factors ◽  
Sleep Apnea ◽  
Sleep Duration ◽  
Sex Education ◽  
Mortality Risk ◽  
Self Report ◽  
Severe Illness ◽  
Uk Biobank ◽  
Sleep Health ◽  
The Uk

Abstract Introduction While there is emerging evidence for acute sleep disruption in the aftermath of coronavirus disease 2019 (COVID-19), it is unknown whether sleep traits contribute to mortality risk. In this study, we tested whether earlier-life sleep duration, chronotype, insomnia, napping or sleep apnea were associated with increased 30-day COVID-19 mortality. Methods We included 34,711 participants from the UK Biobank, who presented for COVID-19 testing between March and October 2020 (mean age at diagnosis: 69.4±8.3; range 50.2–84.6). Self-reported sleep duration (less than 6h/6-9h/more than 9h), chronotype (“morning”/”intermediate”/”evening”), daytime dozing (often/rarely), insomnia (often/rarely), napping (often/rarely) and presence of sleep apnea (ICD-10 or self-report) were obtained between 2006 and 2010. Multivariate logistic regression models were used to adjust for age, sex, education, socioeconomic status, and relevant risk factors (BMI, hypertension, diabetes, respiratory diseases, smoking, and alcohol). Results The mean time between sleep measures and COVID-19 testing was 11.6±0.9 years. Overall, 5,066 (14.6%) were positive. In those who were positive, 355 (7.0%) died within 30 days (median = 8) after diagnosis. Long sleepers (>9h vs. 6-9h) [20/103 (19.4%) vs. 300/4,573 (6.6%); OR 2.09, 95% 1.19–3.64, p=0.009), often daytime dozers (OR 1.68, 95% 1.04–2.72, p=0.03), and nappers (OR 1.52, 95% 1.04–2.23, p=0.03) were at greater odds of mortality. Prior diagnosis of sleep apnea also saw a two-fold increased odds (OR 2.07, 95% CI: 1.25–3.44 p=0.005). No associations were seen for short sleepers, chronotype or insomnia with COVID-19 mortality. Conclusion Data across all current waves of infection show that prior sleep traits/disturbances, in particular long sleep duration, daytime dozing, napping and sleep apnea, are associated with increased 30-day mortality after COVID-19, independent of health-related risk factors. While sleep health traits may reflect unmeasured poor health, further work is warranted to examine the exact underlying mechanisms, and to test whether sleep health optimization offers resilience to severe illness from COVID-19. Support (if any) NIH [T32GM007592 and R03AG067985 to L.G. RF1AG059867, RF1AG064312, to K.H.], the BrightFocus Foundation A2020886S to P.L. and the Foundation of Anesthesia Education and Research MRTG-02-15-2020 to L.G.

scholarly journals Association of Antioxidants Use with All-Cause and Cause-Specific Mortality: A Prospective Study of the UK Biobank

Antioxidants ◽  
2020 ◽  
Vol 9 (12) ◽  
pp. 1287
Author(s):  
Inken Behrendt ◽  
Gerrit Eichner ◽  
Mathias Fasshauer
Keyword(s):  
Respiratory Disease ◽  
Mortality Risk ◽  
Cancer Mortality ◽  
Primary Objective ◽  
Inverse Association ◽  
Uk Biobank ◽  
Cancer Subtypes ◽  
Specific Mortality ◽  
The Uk ◽  
The Impact

Prospective studies and randomized controlled trials elucidating the impact of antioxidants supplementation on mortality risk are inconclusive. The present analysis determined association between regular antioxidants use and all-cause (primary objective), as well as cause-specific, mortality in 345,626 participants of the UK Biobank cohort using Cox proportional hazard models. All models were adjusted for confounders and multiple testing. Antioxidants users were defined as participants who indicated to regularly use at least one of the following: multivitamins, vitamin C, vitamin E, selenium, and zinc. Median age of antioxidants users (n = 101,159) and non-users (n = 244,467) at baseline was 57 years. During 3.9 million person-years and a median follow-up of 11.5 years, 19,491 deaths occurred. Antioxidants use was not significantly associated with all-cause, cancer, and non-cancer mortality including several cancer and non-cancer subtypes. Interestingly, mortality risk from respiratory disease was significantly 21% lower among antioxidants users as compared to non-users (hazard ratio: 0.79; 95% confidence interval: 0.67, 0.92). In conclusion, the present study findings do not support recommendations for antioxidants supplementation to prevent all-cause, cancer, or non-cancer mortality on a population level. The significant inverse association between antioxidants use and respiratory disease mortality needs further study.

scholarly journals Comparison of prognostic models to predict the occurrence of colorectal cancer in asymptomatic individuals: a systematic literature review and external validation in the EPIC and UK Biobank prospective cohort studies

Gut ◽  
2018 ◽  
Vol 68 (4) ◽  
pp. 672-683 ◽  
Author(s):  
Todd Smith ◽  
David C Muller ◽  
Karel G M Moons ◽  
Amanda J Cross ◽  
Mattias Johansson ◽  
...  
Keyword(s):  
Colorectal Cancer ◽  
Systematic Review ◽  
Prediction Models ◽  
Large Population ◽  
External Validation ◽  
Population Based ◽  
Prognostic Models ◽  
Uk Biobank ◽  
Asymptomatic Individuals ◽  
The Uk

ObjectiveTo systematically identify and validate published colorectal cancer risk prediction models that do not require invasive testing in two large population-based prospective cohorts.DesignModels were identified through an update of a published systematic review and validated in the European Prospective Investigation into Cancer and Nutrition (EPIC) and the UK Biobank. The performance of the models to predict the occurrence of colorectal cancer within 5 or 10 years after study enrolment was assessed by discrimination (C-statistic) and calibration (plots of observed vs predicted probability).ResultsThe systematic review and its update identified 16 models from 8 publications (8 colorectal, 5 colon and 3 rectal). The number of participants included in each model validation ranged from 41 587 to 396 515, and the number of cases ranged from 115 to 1781. Eligible and ineligible participants across the models were largely comparable. Calibration of the models, where assessable, was very good and further improved by recalibration. The C-statistics of the models were largely similar between validation cohorts with the highest values achieved being 0.70 (95% CI 0.68 to 0.72) in the UK Biobank and 0.71 (95% CI 0.67 to 0.74) in EPIC.ConclusionSeveral of these non-invasive models exhibited good calibration and discrimination within both external validation populations and are therefore potentially suitable candidates for the facilitation of risk stratification in population-based colorectal screening programmes. Future work should both evaluate this potential, through modelling and impact studies, and ascertain if further enhancement in their performance can be obtained.

scholarly journals Polynomial Mendelian Randomization reveals widespread non-linear causal effects in the UK Biobank

2021 ◽  
Author(s):  
Jonathan Sulc ◽  
Jenny Sjaarda ◽  
Zoltan Kutalik
Keyword(s):  
Causal Inference ◽  
Large Scale ◽  
Causal Effect ◽  
Causal Effects ◽  
Mendelian Randomisation ◽  
Uk Biobank ◽  
Individual Level ◽  
Glucose Levels ◽  
Causal Function ◽  
The Uk

Causal inference is a critical step in improving our understanding of biological processes and Mendelian randomisation (MR) has emerged as one of the foremost methods to efficiently interrogate diverse hypotheses using large-scale, observational data from biobanks. Although many extensions have been developed to address the three core assumptions of MR-based causal inference (relevance, exclusion restriction, and exchangeability), most approaches implicitly assume that any putative causal effect is linear. Here we propose PolyMR, an MR-based method which provides a polynomial approximation of an (arbitrary) causal function between an exposure and an outcome. We show that this method provides accurate inference of the shape and magnitude of causal functions with greater accuracy than existing methods. We applied this method to data from the UK Biobank, testing for effects between anthropometric traits and continuous health-related phenotypes and found most of these (84%) to have causal effects which deviate significantly from linear. These deviations ranged from slight attenuation at the extremes of the exposure distribution, to large changes in the magnitude of the effect across the range of the exposure (e.g. a 1 kg/m2 change in BMI having stronger effects on glucose levels if the initial BMI was higher), to non-monotonic causal relationships (e.g. the effects of BMI on cholesterol forming an inverted U shape). Finally, we show that the linearity assumption of the causal effect may lead to the misinterpretation of health risks at the individual level or heterogeneous effect estimates when using cohorts with differing average exposure levels.

scholarly journals ApoE e4 Genotype Predicts Severe COVID-19 in the UK Biobank Community Cohort

2020 ◽  
Author(s):  
Chia-Ling Kuo ◽  
Luke C Pilling ◽  
Janice L Atkins ◽  
Jane AH Masoli ◽  
João Delgado ◽  
...  
Keyword(s):  
Older Adults ◽  
Risk Factor ◽  
High Risk ◽  
The Novel ◽  
Shortness Of Breath ◽  
Uk Biobank ◽  
Impaired Consciousness ◽  
Homozygous Genotype ◽  
The Common ◽  
The Uk

The novel respiratory disease COVID-19 produces varying symptoms, with fever, cough, and shortness of breath being common. In older adults, we found that pre-existing dementia is a major risk factor (OR = 3.07, 95% CI: 1.71 to 5.50) for COVID-19 hospitalization in the UK Biobank (UKB). In another UK study of 16,749 patients hospitalized for COVID-19, dementia was among the common comorbidities and was associated with higher mortality. Additionally, impaired consciousness, including delirium, is common in severe cases. The ApoE e4 genotype is associated with both dementia and delirium, with the e4e4 (homozygous) genotype associated with high risk of dementia. We therefore aimed to test associations between ApoE e4 alleles and COVID-19 severity, using the UKB data.

Abstract 231: Predicting Readmission Risk Following Coronary Revascularization at the Time of Admission

2014 ◽  
Vol 7 (suppl_1) ◽  
Author(s):  
Daniel J Elliott ◽  
Paul Kolm ◽  
William Weintraub
Keyword(s):  
High Risk ◽  
Prediction Models ◽  
Discharge Planning ◽  
Strong Predictor ◽  
Coronary Revascularization ◽  
Bypass Graft ◽  
Vital Signs ◽  
Laboratory Values ◽  
Cabg Patients ◽  
Time Of Admission

Objective: Reducing readmissions following hospitalization for coronary revascularization is a national priority. Identifying patients at high risk for readmission early in a hospitalization would enable hospitals to target these individuals for enhanced discharge planning. Traditional risk models identify patients based on characteristics that may not be available until after discharge. We sought to compare the model performance for models based on data that is available earlier in the hospitalization. Methods: We developed models to predict 30-day inpatient readmission to our instititution for a cohort of patients who were revascularized at our institution between January 2010 and December 2013. We developed separate models for patients with percutaneous coronary revascularization (PCI) and coronary artery bypass graft surgery (CABG). We developed three models using data available at three different time points in the hospitalization: 1) at admission, 2) early in the hospitalization and 3) at discharge. Candidate variables for the admission model included demographics, comorbidities, and previous utilization within our system. The second model added initial vital signs and laboratory values. The discharge model added discharge vital signs and laboratory values, discharge medications, new comorbidities, and length of stay. We assessed each model using the c-index. Results: Our cohort included 4,941 PCI patients and 1,633 CABG patients. The readmission rate was 8.4% in the PCI group and 14.2% in the CABG group. For both populations, the discriminative ability of the admission model was high (0.805 for PCI, 0.824 for CABG). The addition of laboratory values and vital signs was associated with slight improvement in discrimination (0.812 for PCI, 0.829 for CABG). The addition of data available at the time of discharge further increased model discrimination (0.833 for PCI, 0.848 for CABG). Except for one model, the strongest predictor of readmission was carrying a diagnosis of hypertension at baseline (OR > 2 for all)). Previous hospitalization within six months was a strong predictor among PCI patients (OR 1.18, p<0.001), and having a previous acute myocardial infarction was the strongest predictor among CABG patients (OR 2.46, p = 0.002). Discussion: Risk prediction models based on data available only at discharge minimally improved the performance of models based solely on demographic and utilization data available at the time of admission. These simplified models may be sufficient to identify patients at highest risk of readmission following coronary revascularization early in the hospitlization. This would allow providers and health systems to target high-risk patients with enhanced discharge planning during the course of the hospitalization, and this may improve the ability to avoid readmissions.

scholarly journals Phenome-wide Heritability Analysis of the UK Biobank

2016 ◽  
Author(s):  
Tian Ge ◽  
Chia-Yen Chen ◽  
Benjamin M. Neale ◽  
Mert R. Sabuncu ◽  
Jordan W. Smoller
Keyword(s):  
Complex Traits ◽  
Large Scale ◽  
Prediction Models ◽  
Population Characteristics ◽  
Uk Biobank ◽  
Multiple Traits ◽  
Common Genetic Variants ◽  
Heritability Estimation ◽  
Heritability Analysis ◽  
The Uk

Heritability estimation provides important information about the relative contribution of genetic and environmental factors to phenotypic variation, and provides an upper bound for the utility of genetic risk prediction models. Recent technological and statistical advances have enabled the estimation of additive heritability attributable to common genetic variants (SNP heritability) across a broad phenotypic spectrum. However, assessing the comparative heritability of multiple traits estimated in different cohorts may be misleading due to the population-specific nature of heritability. Here we report the SNP heritability for 551 complex traits derived from the large-scale, population-based UK Biobank, comprising both quantitative phenotypes and disease codes, and examine the moderating effect of three major demographic variables (age, sex and socioeconomic status) on the heritability estimates. Our study represents the first comprehensive phenome-wide heritability analysis in the UK Biobank, and underscores the importance of considering population characteristics in comparing and interpreting heritability.

scholarly journals Combined Use of Serum Uromodulin and eGFR to Estimate Mortality Risk

2021 ◽  
Vol 8 ◽  
Author(s):  
Babak Yazdani ◽  
Graciela E. Delgado ◽  
Hubert Scharnagl ◽  
Bernhard K. Krämer ◽  
Heinz Drexel ◽  
...  
Keyword(s):  
Mortality Risk ◽  
Cardiovascular Health ◽  
Prediction Models ◽  
Driving Forces ◽  
Health Study ◽  
Cardiovascular Health Study ◽  
Inverse Association ◽  
Risk Modeling ◽  
Metabolic Disturbances ◽  
Improve Risk Stratification

Serum uromodulin (sUmod) shows a strong direct correlation with eGFR in patients with impaired kidney function and an inverse association with mortality. However, there are patients in whom only one of both markers is decreased. Therefore, we aimed to investigate the effect of marker discordance on mortality risk. sUmod and eGFR were available in 3,057 participants of the Ludwigshafen Risk and Cardiovascular Health study and 529 participants of the VIVIT study. Both studies are monocentric prospective studies of patients that had been referred for coronary angiography. Participants were categorized into four groups according to the median values of sUmod (LURIC: 146 ng/ml, VIVIT: 156) and eGFR (LURIC: 84 ml/min/1.73 m2, VIVIT: 87). In 945 LURIC participants both markers were high (UHGH), in 935 both were low (ULGL), in 589 only eGFR (UHGL), and in 582 only sUmod (ULGH) was low. After balancing the groups for cardiovascular risk factors, hazard ratios (95%CI) for all-cause mortality as compared to UHGH were 2.03 (1.63–2.52), 1.43 (1.13–1.81), and 1.32 (1.03–1.69) for ULGL, UHGL, and ULGH, respectively. In VIVIT, HRs were 3.12 (1.38–7.08), 2.38 (1.01–5.61), and 2.06 (0.81–5.22). Adding uromodulin to risk prediction models that already included eGFR as a covariate slightly increased the Harrell's C and significantly improved the AUC in LURIC. In UHGL patients, hypertension, heart failure and upregulation of the renin-angiotensin-aldosterone-system seem to be the driving forces of disease development, whereas in ULGH patients metabolic disturbances might be key drivers of increased mortality. In conclusion, SUmod/eGFR subgroups mirror distinct metabolic and clinical patterns. Assessing sUmod additionally to creatinine or cystatin C has the potential to allow a more precise risk modeling and might improve risk stratification.

Sign in / Sign up

Export Citation Format

Share Document