Statistical machine learning of sleep and physical activity phenotypes from sensor data in 96,220 UK Biobank participants.
Ontology highlight
ABSTRACT:
SUBMITTER: Willetts M
PROVIDER: S-EPMC5962537 | biostudies-literature | 2018 May
REPOSITORIES: biostudies-literature
Publications
Statistical machine learning of sleep and physical activity phenotypes from sensor data in 96,220 UK Biobank participants.
Willetts Matthew M Hollowell Sven S Aslett Louis L Holmes Chris C Doherty Aiden A
Scientific reports 20180521 1
Current public health guidelines on physical activity and sleep duration are limited by a reliance on subjective self-reported evidence. Using data from simple wrist-worn activity monitors, we developed a tailored machine learning model, using balanced random forests with Hidden Markov Models, to reliably detect a number of activity modes. We show that physical activity and sleep behaviours can be classified with 87% accuracy in 159,504 minutes of recorded free-living behaviours from 132 adults. ...[more]
Similar Datasets
Project description:Mood disorders, including depression and anxiety, affect almost one-fifth of the world's adult population and are becoming increasingly prevalent. Mutations in circadian clock genes have previously been associated with mood disorders both directly and indirectly through alterations in circadian phase, suggesting that the circadian clock influences multiple molecular pathways involved in mood. By targeting previously identified single nucleotide polymorphisms (SNPs) that have been implicated in anxiety and depressive disorders, we use a combination of statistical and machine learning techniques to investigate associations with the generalized anxiety disorder assessment (GAD-7) scores in a UK Biobank sample of 90,882 individuals. As in previous studies, we observed that females exhibited higher GAD-7 scores than males regardless of genotype. Interestingly, we found no significant effects on anxiety from individual circadian gene variants; only circadian genotypes with multiple SNP variants showed significant associations with anxiety. For both sexes, severe anxiety is associated with a 120-fold increase in odds for individuals with CRY2_AG(rs1083852)/ZBTB20_TT(rs1394593) genotypes and is associated with a near 40-fold reduction in odds for individuals with PER3-A_CG(rs228697)/ZBTB20_TT(rs1394593) genotypes. We also report several sex-specific associations with anxiety. In females, the CRY2/ZBTB20 genotype combination showed a > 200-fold increase in odds of anxiety and PER3/ZBTB20 and CRY1 /PER3-A genotype combinations also appeared as female risk factors. In males, CRY1/PER3-A and PER3-B/ZBTB20 genotype combinations were associated with anxiety risk. Mediation analysis revealed direct associations of CRY2/ZBTB20 variant genotypes with moderate anxiety in females and CRY1/PER3-A variant genotypes with severe anxiety in males. The association of CRY1/PER3-A variant genotypes with severe anxiety in females was partially mediated by extreme evening chronotype. Our results reinforce existing findings that females exhibit stronger anxiety outcomes than males, and provide evidence for circadian gene associations with anxiety, particularly in females. Our analyses only identified significant associations using two-gene combinations, underscoring the importance of combined gene effects on anxiety risk. We describe novel, robust associations between gene combinations involving the ZBTB20 SNP (rs1394593) and risk of anxiety symptoms in a large population sample. Our findings also support previous findings that the ZBTB20 SNP is an important factor in mood disorders, including seasonal affective disorder. Our results suggest that reduced expression of this gene significantly modulates the risk of anxiety symptoms through direct influences on mood-related pathways. Together, these observations provide novel links between the circadian clockwork and anxiety symptoms and identify potential molecular pathways through which clock genes may influence anxiety risk.
| S-EPMC10724169 | biostudies-literature
Project description:BackgroundIdentifying people at risk of cardiovascular diseases (CVD) is a cornerstone of preventative cardiology. Risk prediction models currently recommended by clinical guidelines are typically based on a limited number of predictors with sub-optimal performance across all patient groups. Data-driven techniques based on machine learning (ML) might improve the performance of risk predictions by agnostically discovering novel risk predictors and learning the complex interactions between them. We tested (1) whether ML techniques based on a state-of-the-art automated ML framework (AutoPrognosis) could improve CVD risk prediction compared to traditional approaches, and (2) whether considering non-traditional variables could increase the accuracy of CVD risk predictions.Methods and findingsUsing data on 423,604 participants without CVD at baseline in UK Biobank, we developed a ML-based model for predicting CVD risk based on 473 available variables. Our ML-based model was derived using AutoPrognosis, an algorithmic tool that automatically selects and tunes ensembles of ML modeling pipelines (comprising data imputation, feature processing, classification and calibration algorithms). We compared our model with a well-established risk prediction algorithm based on conventional CVD risk factors (Framingham score), a Cox proportional hazards (PH) model based on familiar risk factors (i.e, age, gender, smoking status, systolic blood pressure, history of diabetes, reception of treatments for hypertension and body mass index), and a Cox PH model based on all of the 473 available variables. Predictive performances were assessed using area under the receiver operating characteristic curve (AUC-ROC). Overall, our AutoPrognosis model improved risk prediction (AUC-ROC: 0.774, 95% CI: 0.768-0.780) compared to Framingham score (AUC-ROC: 0.724, 95% CI: 0.720-0.728, p < 0.001), Cox PH model with conventional risk factors (AUC-ROC: 0.734, 95% CI: 0.729-0.739, p < 0.001), and Cox PH model with all UK Biobank variables (AUC-ROC: 0.758, 95% CI: 0.753-0.763, p < 0.001). Out of 4,801 CVD cases recorded within 5 years of baseline, AutoPrognosis was able to correctly predict 368 more cases compared to the Framingham score. Our AutoPrognosis model included predictors that are not usually considered in existing risk prediction models, such as the individuals' usual walking pace and their self-reported overall health rating. Furthermore, our model improved risk prediction in potentially relevant sub-populations, such as in individuals with history of diabetes. We also highlight the relative benefits accrued from including more information into a predictive model (information gain) as compared to the benefits of using more complex models (modeling gain).ConclusionsOur AutoPrognosis model improves the accuracy of CVD risk prediction in the UK Biobank population. This approach performs well in traditionally poorly served patient subgroups. Additionally, AutoPrognosis uncovered novel predictors for CVD disease that may now be tested in prospective studies. We found that the "information gain" achieved by considering more risk factors in the predictive model was significantly higher than the "modeling gain" achieved by adopting complex predictive models.
| S-EPMC6519796 | biostudies-literature
Project description:BackgroundThe prevalence of sleep disturbance is high and increasing. The study investigated whether active, former and passive smoking were associated with sleep disturbance.MethodsThis cross-sectional study used data from the UK Biobank: a cohort study of 502 655 participants, of whom 498 208 provided self-reported data on smoking and sleep characteristics. Multivariable multinomial and logistic regression models were used to examine the associations between smoking and sleep disturbance.ResultsLong-sleep duration (>9 h) was more common among current smokers [odds ratio (OR): 1.47; 95% confidence interval (CI): 1.17-1.85; probability value (P) = 0.001] than never smokers, especially heavy (>20/day) smokers (OR: 2.85; 95% CI: 1.66-4.89; P < 0.001). Former heavy (>20/day) smokers were also more likely to report short (<6 h) sleep duration (OR: 1.41; 95% CI: 1.25-1.60; P < 0.001), long-sleep duration (OR: 1.99; 95% CI: 1.47-2.71; P < 0.001) and sleeplessness (OR: 1.47; 95% CI: 1.38-1.57; P < 0.001) than never smokers. Among never smokers, those who lived with more than one smoker had higher odds of long-sleep duration than those not cohabitating with a smoker (OR: 2.71; 95% CI: 1.26-5.82; P = 0.011).ConclusionsActive and passive exposure to high levels of tobacco smoke are associated with sleep disturbance. Existing global tobacco control interventions need to be enforced.
| S-EPMC6166587 | biostudies-literature
Project description:BackgroundCopy number variants (CNVs) have been shown to increase risk for physical anomalies, developmental, psychiatric and medical disorders. Some of them have been associated with changes in weight, height, and other physical traits. As most studies have been performed on children and young people, these effects of CNVs in middle-aged and older people are not well established. The UK Biobank recruited half a million adults who provided a variety of physical measurements. We called all CNVs from the Affymetrix microarrays and selected a set of 54 CNVs implicated as pathogenic (including their reciprocal deletions/duplications) and that were found in five or more persons. Linear regression analysis was used to establish their association with 16 physical traits relevant to human health.Results396,725 participants of white British or Irish descent (excluding first-degree relatives) passed our quality control filters. Out of the 864 CNV/trait associations, 214 were significant at a false discovery rate of 0.1, most of them novel. Many of these traits increase risk for adverse health outcomes: e.g. increases in weight, waist-to-hip ratio, pulse rate and body fat composition. Deletions at 16p11.2, 16p12.1, NRXN1 and duplications at 16p13.11 and 22q11.2 produced the highest numbers of significant associations. Five CNVs produced average changes of over one standard deviation for the 16 traits, compared to controls: deletions at 16p11.2 and 22q11.2, and duplications at 3q29, the Williams-Beuren and Potocki-Lupski regions. CNVs at 1q21.1, 2q13, 16p11.2 and 16p11.2 distal, 16p12.1, 17p12 and 17q12 demonstrated one or more mirror image effects of deletions versus duplications.ConclusionsCarriers of many CNVs should be monitored for physical traits that increase morbidity and mortality. Genes within these CNVs can give insights into biological processes and therapeutic interventions.
| S-EPMC6278042 | biostudies-literature
Project description:BackgroundFrailty indices (FIs) measure variation in health between aging individuals. Researching FIs in resources with large-scale genetic and phenotypic data will provide insights into the causes and consequences of frailty. Thus, we aimed to develop an FI using UK Biobank data, a cohort study of 500,000 middle-aged and older adults.MethodsAn FI was calculated using 49 self-reported questionnaire items on traits covering health, presence of diseases and disabilities, and mental well-being, according to standard protocol. We used multiple imputation to derive FI values for the entire eligible sample in the presence of missing item data (N = 500,336). To validate the measure, we assessed associations of the FI with age, sex, and risk of all-cause mortality (follow-up ≤ 9.7 years) using linear and Cox proportional hazards regression models.ResultsMean FI in the cohort was 0.125 (SD = 0.075), and there was a curvilinear trend toward higher values in older participants. FI values were also marginally higher on average in women than in men. In survival models, 10% higher baseline frailty (ie, a 0.1 FI increment) was associated with higher risk of death (hazard ratio = 1.65; 95% confidence interval: 1.62-1.68). Associations were stronger in younger participants than in older participants, and in men than in women (hazard ratios: 1.72 vs. 1.56, respectively).ConclusionsThe FI is a valid measure of frailty in UK Biobank. The cohort's data are open access for researchers to use, and we provide script for deriving this tool to facilitate future studies on frailty.
| S-EPMC6417451 | biostudies-literature
Project description:BACKGROUND:Physical inactivity is associated with an increased risk of major chronic diseases, although uncertainty exists about which chronic diseases, themselves, might contribute to physical inactivity. The objective of this study was to compare the physical activity of those with chronic diseases to healthy individuals using an objective measure of physical activity. METHODS:We conducted a cross-sectional analysis of data from 96 706 participants aged 40 years or older from the UK Biobank prospective cohort study (2006-10). Diagnoses were identified through ICD 9 and 10 coding within hospital admission records and a cancer registry linked to UK Biobank participants. We extracted summary physical activity information from participants who wore a wrist-worn triaxial accelerometer for 7 days. Statistical analyses included computation of adjusted geometric means and means using general linear models. RESULTS:Participants with chronic disease undertook 9% or 61 minutes (95% confidence interval: 57.8-64.8) less moderate activity and 11% or 3 minutes (95% confidence interval: 2.7-3.3) less vigorous activity per week than individuals without chronic disease. Participants in every chronic-disease subgroup undertook less physical activity than those without chronic disease. Sixty-seven diagnoses within these subgroups were associated with lower moderate activity. CONCLUSIONS:The cross-sectional association of physical activity with chronic disease is broad. Given the substantial health benefits of being physically active, clinicians and policymakers should be aware that their patients with any chronic disease are at greater health risk from other diseases than anticipated because of their physical inactivity.
| S-EPMC6693885 | biostudies-literature
Project description:Pharmacogenetics studies how genetic variation leads to variability in drug response. Guidelines for selecting the right drug and right dose for patients based on their genetics are clinically effective, but are widely unused. For some drugs, the normal clinical decision making process may lead to the optimal dose of a drug that minimizes side effects and maximizes effectiveness. Without measurements of genotype, physicians and patients may adjust dosage in a manner that reflects the underlying genetics. The emergence of genetic data linked to longitudinal clinical data in large biobanks offers an opportunity to confirm known pharmacogenetic interactions as well as discover novel associations by investigating outcomes from normal clinical practice. Here we use the UK Biobank to search for pharmacogenetic interactions among 200 drugs and 9 genes among 200,000 participants. We identify associations between pharmacogene phenotypes and drug maintenance dose as well as differential drug response phenotypes. We find support for several known drug-gene associations as well as novel pharmacogenetic interactions.
| S-EPMC7951365 | biostudies-literature
Project description:(1) Background: The association between metabolic obesity phenotypes and incident lung cancer (LC) remains unclear. (2) Methods: Based on the combination of baseline BMI categories and metabolic health status, participants were categorized into eight groups: metabolically healthy underweight (MHUW), metabolically unhealthy underweight (MUUW), metabolically healthy normal (MHN), metabolically unhealthy normal (MUN), metabolically healthy overweight (MHOW), metabolically unhealthy overweight (MUOW), metabolically healthy obesity (MHO), and metabolically unhealthy obesity (MUO). The Cox proportional hazards model and Mendelian randomization (MR) were applied to assess the association between metabolic obesity phenotypes with LC risk. (3) Results: During a median follow-up of 9.1 years, 3654 incident LC patients were confirmed among 450,482 individuals. Compared with participants with MHN, those with MUUW had higher rates of incident LC (hazard ratio (HR) = 3.24, 95% confidence interval (CI) = 1.33-7.87, p = 0.009). MHO and MHOW individuals had a 24% and 18% lower risk of developing LC, respectively (MHO: HR = 0.76, 95% CI = 0.61-0.95, p = 0.02; MHO: HR = 0.82, 95% CI = 0.70-0.96, p = 0.02). No genetic association of metabolic obesity phenotypes and LC risk was observed in MR analysis. (4) Conclusions: In this prospective cohort study, individuals with MHOW and MHO phenotypes were at a lower risk and MUUW were at a higher risk of LC. However, MR failed to reveal any evidence that metabolic obesity phenotypes would be associated with a higher risk of LC.
| S-EPMC9414360 | biostudies-literature
Project description:ObjectiveAtrial fibrillation (AF) is the most common cardiac arrythmia, and it is associated with increased risk for ischemic stroke, which is underestimated, as AF can be asymptomatic. The aim of this study was to develop optimal ML models for prediction of AF in the population, and secondly for ischemic stroke in AF patients.MethodsTo develop ML models for prediction of 1) AF in the general population and 2) ischemic stroke in patients with AF we constructed XGBoost, LightGBM, Random Forest, Deep Neural Network, Support Vector Machine and Lasso penalised logistic regression models using UK-Biobank's extensive real-world clinical data, questionnaires, as well as biochemical and genetic data, and their predictive performances were compared. Ranking and contribution of the different features was assessed by SHapley Additive exPlanations (SHAP) analysis. The clinical tool CHA2DS2-VASc for prediction of ischemic stroke among AF patients, was used for comparison to the best performing ML model.FindingsThe best performing model for AF prediction was LightGBM, with an area-under-the-roc-curve (AUROC) of 0.729 (95% confidence intervals (CI): 0.719, 0.738). The best performing model for ischemic stroke prediction in AF patients was XGBoost with AUROC of 0.631 (95% CI: 0.604, 0.657). The improved AUROC in the XGBoost model compared to CHA2DS2-VASc was statistically significant based on DeLong's test (p-value = 2.20E-06). In addition, the SHAP analysis showed that several peripheral blood biomarkers (e.g. creatinine, glycated haemoglobin, monocytes) were associated with ischemic stroke, which are not considered by CHA2DS2-VASc.ImplicationsThe best performing ML models presented have the potential for clinical use, but further validation in independent studies is required. Our results endorse the incorporation of some routinely measured blood biomarkers for ischemic stroke prediction in AF patients.
| S-EPMC10987914 | biostudies-literature
Project description:AimsTo quantify the association of combined sleep behaviours and genetic susceptibility with the incidence of cardiovascular disease (CVD).Methods and resultsThis study included 385 292 participants initially free of CVD from UK Biobank. We newly created a healthy sleep score according to five sleep factors and defined the low-risk groups as follows: early chronotype, sleep 7-8?h per day, never/rarely insomnia, no snoring, and no frequent excessive daytime sleepiness. Weighted genetic risk scores of coronary heart disease (CHD) or stroke were calculated. During a median of 8.5 years of follow-up, we documented 7280 incident CVD cases including 4667 CHD and 2650 stroke cases. Compared to those with a sleep score of 0-1, participants with a score of 5 had a 35% (19-48%), 34% (22-44%), and 34% (25-42%) reduced risk of CVD, CHD, and stroke, respectively. Nearly 10% of cardiovascular events in this cohort could be attributed to poor sleep pattern. Participants with poor sleep pattern and high genetic risk showed the highest risk of CHD and stroke.ConclusionIn this large prospective study, a healthy sleep pattern was associated with reduced risks of CVD, CHD, and stroke among participants with low, intermediate, or high genetic risk.
| S-EPMC7071844 | biostudies-literature