Self-supervised learning of accelerometer data provides new insights for sleep and its association with mortality.

Sleep is essential to life. Accurate measurement and classification of sleep/wake and sleep stages is important in clinical studies for sleep disorder diagnoses and in the interpretation of data from consumer devices for monitoring physical and mental well-being. Existing non-polysomnography sleep classification techniques mainly rely on heuristic methods developed in relatively small cohorts. Thus, we aimed to establish the accuracy of wrist-worn accelerometers for sleep stage classification and subsequently describe the association between sleep duration and efficiency (proportion of total time asleep when in bed) with mortality outcomes. We developed a self-supervised deep neural network for sleep stage classification using concurrent laboratory-based polysomnography and accelerometry. After exclusion, 1448 participant nights of data were used for training. The difference between polysomnography and the model classifications on the external validation was 34.7 min (95% limits of agreement (LoA): -37.8-107.2 min) for total sleep duration, 2.6 min for REM duration (95% LoA: -68.4-73.4 min) and 32.1 min (95% LoA: -54.4-118.5 min) for NREM duration. The sleep classifier was deployed in the UK Biobank with 100,000 participants to study the association of sleep duration and sleep efficiency with all-cause mortality. Among 66,214 UK Biobank participants, 1642 mortality events were observed. Short sleepers (<6 h) had a higher risk of mortality compared to participants with normal sleep duration of 6-7.9 h, regardless of whether they had low sleep efficiency (Hazard ratios (HRs): 1.58; 95% confidence intervals (CIs): 1.19-2.11) or high sleep efficiency (HRs: 1.45; 95% CIs: 1.16-1.81). Deep-learning-based sleep classification using accelerometers has a fair to moderate agreement with polysomnography. Our findings suggest that having short overnight sleep confers mortality risk irrespective of sleep continuity.

Data Interoperability for Ambulatory Monitoring of Cardiovascular Disease: A Scientific Statement From the American Heart Association.

Wearable devices are increasingly used by a growing portion of the population to track health and illnesses. The data emerging from these devices can potentially transform health care. This requires an interoperability framework that enables the deployment of platforms, sensors, devices, and software applications within diverse health systems, aiming to facilitate innovation in preventing and treating cardiovascular disease. However, the current data ecosystem includes several noninteroperable systems that inhibit such objectives. The design of clinically meaningful systems for accessing and incorporating these data into clinical workflows requires strategies to ensure the quality of data and clinical content and patient and caregiver accessibility. This scientific statement aims to address the best practices, gaps, and challenges pertaining to data interoperability in this area, with considerations for (1) data integration and the scope of measures, (2) application of these data into clinical approaches/strategies, and (3) regulatory/ethical/legal issues.

Self-Supervised Machine Learning to Characterise Step Counts from Wrist-Worn Accelerometers in the UK Biobank.

PURPOSE: Step count is an intuitive measure of physical activity frequently quantified in health-related studies; however, accurate step counting is difficult in the free-living environment, with error routinely above 20% in wrist-worn devices against camera-annotated ground truth. This study aims to describe the development and validation of step count derived from a wrist-worn accelerometer and assess its association with cardiovascular and all-cause mortality in a large prospective cohort. METHODS: We developed and externally validated a self-supervised machine learning step detection model, trained on an open-source and step-annotated free-living dataset. 39 individuals will free-living ground-truth annotated step counts were used for model development. An open-source dataset with 30 individuals was used for external validation. Epidemiological analysis was performed using 75,263 UK Biobank participants without prevalent cardiovascular disease (CVD) or cancer. Cox regression was used to test the association of daily step count with fatal CVD and all-cause mortality after adjustment for potential confounders. RESULTS: The algorithm substantially outperformed reference models (free-living mean absolute percent error of 12.5%, versus 65-231%). Our data indicate an inverse dose-response association, where taking 6,430-8,277 daily steps was associated with 37% [25-48%] and 28% [20-35%] lower risk of fatal CVD and all-cause mortality up to seven years later, compared to those taking fewer steps each day. CONCLUSIONS: We have developed an open and transparent method that markedly improves the measurement of steps in large-scale wrist-worn accelerometer datasets. The application of this method demonstrated expected associations with CVD and all-cause mortality, indicating excellent face validity. This reinforces public health messaging for increasing physical activity and can help lay the groundwork for the inclusion of target step counts in future public health guidelines.

Proteomic Analyses in Diverse Populations Improved Risk Prediction and Identified New Drug Targets for Type 2 Diabetes.

OBJECTIVE: Integrated analyses of plasma proteomics and genetic data in prospective studies can help assess the causal relevance of proteins, improve risk prediction, and discover novel protein drug targets for type 2 diabetes (T2D). RESEARCH DESIGN AND METHODS: We measured plasma levels of 2,923 proteins using Olink Explore among ∼2,000 randomly selected participants from China Kadoorie Biobank (CKB) without prior diabetes at baseline. Cox regression assessed associations of individual protein with incident T2D (n = 92 cases). Proteomic-based risk models were developed with discrimination, calibration, reclassification assessed using area under the curve (AUC), calibration plots, and net reclassification index (NRI), respectively. Two-sample Mendelian randomization (MR) analyses using cis-protein quantitative trait loci identified in a genome-wide association study of CKB and UK Biobank for specific proteins were conducted to assess their causal relevance for T2D, along with colocalization analyses to examine shared causal variants between proteins and T2D. RESULTS: Overall, 33 proteins were significantly associated (false discovery rate <0.05) with risk of incident T2D, including IGFBP1, GHR, and amylase. The addition of these 33 proteins to a conventional risk prediction model improved AUC from 0.77 (0.73-0.82) to 0.88 (0.85-0.91) and NRI by 38%, with predicted risks well calibrated with observed risks. MR analyses provided support for the causal relevance for T2D of ENTR1, LPL, and PON3, with replication of ENTR1 and LPL in Europeans using different genetic instruments. Moreover, colocalization analyses showed strong evidence (pH4 > 0.6) of shared genetic variants of LPL and PON3 with T2D. CONCLUSIONS: Proteomic analyses in Chinese adults identified novel associations of multiple proteins with T2D with strong genetic evidence supporting their causal relevance and potential as novel drug targets for prevention and treatment of T2D.

Correlates and consequences of atrial fibrillation in a prospective study of 25 000 participants in the China Kadoorie Biobank.

Aims: The prevalence of atrial fibrillation (AF) is positively correlated with prior cardiovascular diseases (CVD) and CVD risk factors but is lower in Chinese than Europeans despite their higher burden of CVD. We examined the prevalence and prognosis of AF and other electrocardiogram (ECG) abnormalities in the China Kadoorie Biobank. Methods and results: A random sample of 25 239 adults (mean age 59.5 years, 62% women) had a 12-lead ECG recorded and interpreted using a Mortara VERITAS™ algorithm in 2013-14. Participants were followed up for 5 years for incident stroke, ischaemic heart disease, heart failure (HF), and all CVD, overall and by CHA2DS2-VASc scores, age, sex, and area. Overall, 1.2% had AF, 13.6% had left ventricular hypertrophy (LVH), and 28.1% had ischaemia (two-thirds of AF cases also had ischaemia or LVH). The prevalence of AF increased with age, prior CVD, and levels of CHA2DS2-VASc scores (0.5%, 1.3%, 2.1%, 2.9%, and 4.4% for scores <2, 2, 3, 4, and ≥5, respectively). Atrial fibrillation was associated with two-fold higher hazard ratios (HR) for CVD (2.15; 95% CI, 1.71-2.69) and stroke (1.88; 1.44-2.47) and a four-fold higher HR for HF (3.79; 2.21-6.49). The 5-year cumulative incidence of CVD was comparable for AF, prior CVD, and CHA2DS2-VASc scores ≥ 2 (36.7% vs. 36.2% vs. 37.7%, respectively) but was two-fold greater than for ischaemia (19.4%), LVH (18.0%), or normal ECG (14.1%), respectively. Conclusion: The findings highlight the importance of screening for AF together with estimation of CHA2DS2-VASc scores for prevention of CVD in Chinese adults.

Iron Status and Risk of Heart Disease, Stroke, and Diabetes: A Mendelian Randomization Study in European Adults.

BACKGROUND: The relevance of iron status biomarkers for coronary artery disease (CAD), heart failure (HF), ischemic stroke (IS), and type 2 diabetes (T2D) is uncertain. We compared the observational and Mendelian randomization (MR) analyses of iron status biomarkers and hemoglobin with these diseases. METHODS AND RESULTS: Observational analyses of hemoglobin were compared with genetically predicted hemoglobin with cardiovascular diseases and diabetes in the UK Biobank. Iron biomarkers included transferrin saturation, serum iron, ferritin, and total iron binding capacity. MR analyses assessed associations with CAD (CARDIOGRAMplusC4D [Coronary Artery Disease Genome Wide Replication and Meta-Analysis Plus The Coronary Artery Disease Genetics], n=181 522 cases), HF (HERMES [Heart Failure Molecular Epidemiology for Therapeutic Targets), n=115 150 cases), IS (GIGASTROKE, n=62 100 cases), and T2D (DIAMANTE [Diabetes Meta-Analysis of Trans-Ethnic Association Studies], n=80 154 cases) genome-wide consortia. Observational analyses demonstrated J-shaped associations of hemoglobin with CAD, HF, IS, and T2D. In contrast, MR analyses demonstrated linear positive associations of higher genetically predicted hemoglobin levels with 8% higher risk per 1 SD higher hemoglobin for CAD, 10% to 13% for diabetes, but not with IS or HF in UK Biobank. Bidirectional MR analyses confirmed the causal relevance of iron biomarkers for hemoglobin. Further MR analyses in global consortia demonstrated modest protective effects of iron biomarkers for CAD (7%-14% lower risk for 1 SD higher levels of iron biomarkers), adverse effects for T2D, but no associations with IS or HF. CONCLUSIONS: Higher levels of iron biomarkers were protective for CAD, had adverse effects on T2D, but had no effects on IS or HF. Randomized trials are now required to assess effects of iron supplements on risk of CAD in high-risk older people.

Use of Artificial Intelligence in Improving Outcomes in Heart Disease: A Scientific Statement From the American Heart Association.

A major focus of academia, industry, and global governmental agencies is to develop and apply artificial intelligence and other advanced analytical tools to transform health care delivery. The American Heart Association supports the creation of tools and services that would further the science and practice of precision medicine by enabling more precise approaches to cardiovascular and stroke research, prevention, and care of individuals and populations. Nevertheless, several challenges exist, and few artificial intelligence tools have been shown to improve cardiovascular and stroke care sufficiently to be widely adopted. This scientific statement outlines the current state of the art on the use of artificial intelligence algorithms and data science in the diagnosis, classification, and treatment of cardiovascular disease. It also sets out to advance this mission, focusing on how digital tools and, in particular, artificial intelligence may provide clinical and mechanistic insights, address bias in clinical studies, and facilitate education and implementation science to improve cardiovascular and stroke outcomes. Last, a key objective of this scientific statement is to further the field by identifying best practices, gaps, and challenges for interested stakeholders.

PCSK9 genetic variants and risk of vascular and non-vascular diseases in Chinese and UK populations.

AIM: Lowering low-density lipoprotein cholesterol (LDL-C) through PCSK9 inhibition represents a new therapeutic approach to preventing and treating cardiovascular disease (CVD). Phenome-wide analyses of PCSK9 genetic variants in large biobanks can help to identify unexpected effects of PCSK9 inhibition. METHODS: In the prospective China Kadoorie Biobank, we constructed a genetic score using three variants at the PCSK9 locus associated with directly-measured LDL-C (PCSK9-GS). Logistic regression gave estimated odds ratios (ORs) for PCSK9-GS associations with CVD and non-CVD outcomes, scaled to 1SD lower LDL-C. RESULTS: PCSK9-GS was associated with lower risks of carotid plaque (n=8340 cases; OR=0.61 [95%CI: 0.45-0.83]; P=0.0015), major occlusive vascular events (n=15,752; 0.80 [0.67-0.95]; P=0.011), and ischaemic stroke (n=11,467; 0.80 [0.66-0.98]; P=0.029). However, PCSK9-GS was also associated with higher risk of hospitalisation with chronic obstructive pulmonary disease (COPD: n=6836; 1.38 [1.08-1.76]; P=0.0089), and with even higher risk of fatal exacerbations among individuals with pre-existing COPD (n=730; 3.61 [1.71-7.60]; P=7.3x10-4). We also replicated associations for a PCSK9 variant, reported in UK Biobank, with increased risks of acute upper respiratory tract infection (URTI) (pooled OR after meta-analysis of 1.87 ([1.38-2.54]; P=5.4x10-5) and self-reported asthma (pooled OR 1.17 ([1.04-1.30]; P=0.0071). There was no association of a polygenic LDL-C score with COPD hospitalisation, COPD exacerbation, or URTI. CONCLUSIONS: LDL-C-lowering PCSK9 genetic variants are associated with lower risk of subclinical and clinical atherosclerotic vascular disease, but higher risks of respiratory diseases. Pharmacovigilance studies may be required to monitor patients treated with therapeutic PCSK9 inhibitors for exacerbations of respiratory diseases or respiratory tract infections.

Genetic analyses of over 100,000 participants of the China Kadoorie Biobank, mimicking the effect of new drugs intended to reduce cholesterol by targeting the PCSK9 protein, have identified potential severe effects of lower PCSK9 activity in patients with existing respiratory disease. PCSK9 genetic variants that are associated with lower cholesterol and reduced rates of cardiovascular disease are also associated with increased risk of a range of respiratory diseases, including asthma, upper respiratory tract infections, and hospitalisation with chronic obstructive respiratory disease (COPD). These genetic variants are not associated with whether or not individuals have COPD; instead they are specifically associated with an increase in the chance of those who already have COPD being hospitalised and even dying, suggesting that careful monitoring of such patients should be considered during development of and treatment with anti-PCSK9 medication.

Global heterogeneity in folic acid fortification policies and implications for prevention of neural tube defects and stroke: a systematic review.

Background: Folic acid (pteroylmonoglutamic acid) supplements are highly effective for prevention of neural tube defects (NTD) prompting implementation of mandatory or voluntary folic acid fortification for prevention of NTDs. We used plasma folate levels in population studies by country and year to compare effects of folic acid fortification types (mandatory or voluntary folic acid fortification policies) on plasma folate levels, NTD prevalence and stroke mortality rates. Methods: We conducted systematic reviews of (i) implementation of folic acid fortification in 193 countries that were member states of the World Health Organization by country and year, and (ii) estimated population mean plasma folate levels by year and type of folic acid fortification. We identified relevant English language reports published between Jan 1, 1990 and July 31, 2023 using Google Scholar, Medline, Embase and Global Health. Eligibility criteria were observational or interventional studies with >1000 participants. Studies of pregnant women or children <15 years were excluded. Using an ecological study design, we examined the associations of folic acid fortification types with NTD prevalence (n = 108 studies) and stroke mortality rates (n = 3 countries). Findings: Among 193 countries examined up to 31 July 2023, 69 implemented mandatory folic acid fortification, 47 had voluntary fortification, but 77 had no fortification (accounting for 32%, 53% and 15% of worldwide population, respectively). Mean plasma folate levels were 36, 21 and 17 nmol/L in populations with mandatory, voluntary and no fortification, respectively (and proportions with mean folate levels >25 nmol/L were 100%, 15% and 7%, respectively). Among 75 countries with NTD prevalence, mean (95% CI) prevalence per 10,000 population were 4.19 (4.11-4.28), 7.61 (7.47-7.75) and 9.66 (9.52-9.81) with mandatory, voluntary and no folic acid fortification, respectively. However, age-standardised trends in stroke mortality rates were unaltered by the introduction of folic acid fortification. Interpretation: There is substantial heterogeneity in folic acid fortification policies worldwide where folic acid fortification are associated with 50-100% higher population mean plasma folate levels and 25-50% lower NTD prevalence compared with no fortification. Many thousand NTD pregnancies could be prevented yearly if all countries implemented mandatory folic acid fortification. Further trials of folic acid for stroke prevention are required in countries without effective folic acid fortification policies. Funding: Medical Research Council (UK) and British Heart Foundation.

Plasma Proteomics to Identify Drug Targets for Ischemic Heart Disease.

BACKGROUND: Integrated analyses of plasma proteomic and genetic markers in prospective studies can clarify the causal relevance of proteins and discover novel targets for ischemic heart disease (IHD) and other diseases. OBJECTIVES: The purpose of this study was to examine associations of proteomics and genetics data with IHD in population studies to discover novel preventive treatments. METHODS: We conducted a nested case-cohort study in the China Kadoorie Biobank (CKB) involving 1,971 incident IHD cases and 2,001 subcohort participants who were genotyped and free of prior cardiovascular disease. We measured 1,463 proteins in the stored baseline samples using the OLINK EXPLORE panel. Cox regression yielded adjusted HRs for IHD associated with individual proteins after accounting for multiple testing. Moreover, cis-protein quantitative loci (pQTLs) identified for proteins in genome-wide association studies of CKB and of UK Biobank were used as instrumental variables in separate 2-sample Mendelian randomization (MR) studies involving global CARDIOGRAM+C4D consortium (210,842 IHD cases and 1,378,170 controls). RESULTS: Overall 361 proteins were significantly associated at false discovery rate <0.05 with risk of IHD (349 positively, 12 inversely) in CKB, including N-terminal prohormone of brain natriuretic peptide and proprotein convertase subtilisin/kexin type 9. Of these 361 proteins, 212 had cis-pQTLs in CKB, and MR analyses of 198 variants in CARDIOGRAM+C4D identified 13 proteins that showed potentially causal associations with IHD. Independent MR analyses of 307 cis-pQTLs identified in Europeans replicated associations for 4 proteins (FURIN, proteinase-activated receptor-1, Asialoglycoprotein receptor-1, and matrix metalloproteinase-3). Further downstream analyses showed that FURIN, which is highly expressed in endothelial cells, is a potential novel target and matrix metalloproteinase-3 a potential repurposing target for IHD. CONCLUSIONS: Integrated analyses of proteomic and genetic data in Chinese and European adults provided causal support for FURIN and multiple other proteins as potential novel drug targets for treatment of IHD.

Device-measured movement behaviours in over 20,000 China Kadoorie Biobank participants.

BACKGROUND: Movement behaviours, including physical activity, sedentary behaviour, and sleep have been shown to be associated with several chronic diseases. However, they have not been objectively measured in large-scale prospective cohort studies in low-and middle-income countries. We aim to describe the patterns of device-measured movement behaviours collected in the China Kadoorie Biobank (CKB) study. METHODS: During 2020 and 2021, a random subset of 25,087 surviving CKB individuals participated in the 3rd resurvey of the CKB. Among them, 22,511 (89.7%) agreed to wear an Axivity AX3 wrist-worn triaxial accelerometer for seven consecutive days to assess their habitual movement behaviours. We developed a machine-learning model to infer time spent in four movement behaviours [i.e. sleep, sedentary behaviour, light intensity physical activity (LIPA), and moderate-to-vigorous physical activity (MVPA)]. Descriptive analyses were performed for wear-time compliance and patterns of movement behaviours by different participant characteristics. RESULTS: Data from 21,897 participants (aged 65.4 ± 9.1 years; 35.4% men) were received for demographic and wear-time analysis, with a median wear-time of 6.9 days (IQR: 6.1-7.0). Among them, 20,370 eligible participants were included in movement behavior analyses. On average, they had 31.1 mg/day (total acceleration) overall activity level, accumulated 7.7 h/day (32.3%) of sleep time, 8.8 h/day (36.6%) sedentary, 5.7 h/day (23.9%) in light physical activity, and 104.4 min/day (7.2%) in moderate-to-vigorous physical activity. There was an inverse relationship between age and overall acceleration with an observed decline of 5.4 mg/day (17.4%) per additional decade. Women showed a higher activity level than men (32.3 vs 28.8 mg/day) and there was a marked geographical disparity in the overall activity level and time allocation. CONCLUSIONS: This is the first large-scale accelerometer data collected among Chinese adults, which provides rich and comprehensive information about device-measured movement behaviour patterns. This resource will enhance our knowledge about the potential relevance of different movement behaviours for chronic disease in Chinese adults.

Alcohol intake and cause-specific mortality: conventional and genetic evidence in a prospective cohort study of 512 000 adults in China.

BACKGROUND: Genetic variants that affect alcohol use in East Asian populations could help assess the causal effects of alcohol consumption on cause-specific mortality. We aimed to investigate the associations between alcohol intake and cause-specific mortality using conventional and genetic epidemiological methods among more than 512 000 adults in China. METHODS: The prospective China Kadoorie Biobank cohort study enrolled 512 724 adults (210 205 men and 302 519 women) aged 30-79 years, during 2004-08. Residents with no major disabilities from ten diverse urban and rural areas of China were invited to participate, and alcohol use was self-reported. During 12 years of follow-up, 56 550 deaths were recorded through linkage to death registries, including 23 457 deaths among 168 050 participants genotyped for ALDH2-rs671 and ADH1B-rs1229984. Adjusted hazard ratios (HRs) for cause-specific mortality by self-reported and genotype-predicted alcohol intake were estimated using Cox regression. FINDINGS: 33% of men drank alcohol most weeks. In conventional observational analyses, ex-drinkers, non-drinkers, and heavy drinkers had higher risks of death from most major causes than moderate drinkers. Among current drinkers, each 100 g/week higher alcohol intake was associated with higher mortality risks from cancers (HR 1·18 [95% CI 1·14-1·22]), cardiovascular disease (CVD; HR 1·19 [1·15-1·24]), liver diseases (HR 1·51 [1·27-1·78]), non-medical causes (HR 1·15 [1·08-1·23]), and all causes (HR 1·18 [1·15-1·20]). In men, ALDH2-rs671 and ADH1B-rs1229984 genotypes predicted 60-fold differences in mean alcohol intake (4 g/week in the lowest group vs 255 g/week in the highest). Genotype-predicted alcohol intake was uniformly and positively associated with risks of death from all causes (n=12 939; HR 1·07 [95% CI 1·05-1·10]) and from pre-defined alcohol-related cancers (n=1274; 1·12 [1·04-1·21]), liver diseases (n=110; 1·31 [1·02-1·69]), and CVD (n=6109; 1·15 [1·10-1·19]), chiefly due to stroke (n=3285; 1·18 [1·12-1·24]) rather than ischaemic heart disease (n=2363; 1·06 [0·99-1·14]). Results were largely consistent using a polygenic score to predict alcohol intake, with higher intakes associated with higher risks of death from alcohol-related cancers, CVD, and all causes. Approximately 2% of women were current drinkers, and although power was low to assess observational associations of alcohol with mortality, the genetic evidence suggested that the excess risks in men were due to alcohol, not pleiotropy. INTERPRETATION: Higher alcohol intake increased the risks of death overall and from major diseases for men in China. There was no genetic evidence of protection from moderate drinking for all-cause and cause-specific mortality, including CVD. FUNDING: Kadoorie Charitable Foundation, National Natural Science Foundation of China, British Heart Foundation, Cancer Research UK, GlaxoSmithKline, Wellcome Trust, Medical Research Council, and Chinese Ministry of Science and Technology.

Independent relevance of adiposity measures to coronary heart disease risk among 0.5 million adults in UK Biobank.

BACKGROUND: Evidence on body fat distribution shows opposing effects of waist circumference (WC) and hip circumference (HC) for coronary heart disease (CHD). We aimed to investigate the causality and the shape of such associations. METHODS: UK Biobank is a prospective cohort study of 0.5 million adults aged 40-69 years recruited between 2006 and 2010. Adjusted hazard ratios (HRs) for the associations of measured and genetically predicted body mass index (BMI), WC, HC and waist-to-hip ratio with incident CHD were obtained from Cox models. Mendelian randomization (MR) was used to assess causality. The analysis included 456 495 participants (26 225 first-ever CHD events) without prior CHD. RESULTS: All measures of adiposity demonstrated strong, positive and approximately log-linear associations with CHD risk over a median follow-up of 12.7 years. For HC, however, the association became inverse given the BMI and WC (HR per usual SD 0.95, 95% CI 0.93-0.97). Associations for BMI and WC remained independently positive after adjustment for other adiposity measures and were similar (1.14, 1.13-1.16 and 1.18, 1.15-1.20, respectively), with WC displaying stronger associations among women. Blood pressure, plasma lipids and dysglycaemia accounted for much of the observed excess risk. MR results were generally consistent with the observational, implying causality. CONCLUSIONS: Body fat distribution measures displayed similar associations with CHD risk as BMI except for HC, which was inversely associated with CHD risk (given WC and BMI). These findings suggest that different measures of body fat distribution likely influence CHD risk through both overlapping and independent mechanisms.

MTHFR and risk of stroke and heart disease in a low-folate population: a prospective study of 156 000 Chinese adults.

BACKGROUND: The relevance of folic acid for stroke prevention in low-folate populations such as in China is uncertain. Genetic studies of the methylenetetrahydrofolate reductase (MTHFR) C677T polymorphism, which increases plasma homocysteine (tHcy) levels, could clarify the causal relevance of elevated tHcy levels for stroke, ischaemic heart disease (IHD) and other diseases in populations without folic acid fortification. METHODS: In the prospective China Kadoorie Biobank, 156 253 participants were genotyped for MTHFR and 12 240 developed a stroke during the 12-year follow-up. Logistic regression was used to estimate region-specific odds ratios (ORs) for total stroke and stroke types, IHD and other diseases comparing TT genotype for MTHFR C677T (two thymine alleles at position 677 of MTHFR C677T polymorphism) vs CC (two cytosine alleles) after adjustment for age and sex, and these were combined using inverse-variance weighting. RESULTS: Overall, 21% of participants had TT genotypes, but this varied from 5% to 41% across the 10 study regions. Individuals with TT genotypes had 13% (adjusted OR 1.13, 95% CI 1.09-1.17) higher risks of any stroke [with a 2-fold stronger association with intracerebral haemorrhage (1.24, 1.17-1.32) than for ischaemic stroke (1.11, 1.07-1.15)] than the reference CC genotype. In contrast, MTHFR C677T was unrelated to risk of IHD or any other non-vascular diseases, including cancer, diabetes and chronic obstructive lung disease. CONCLUSIONS: In Chinese adults, the MTHFR C677T polymorphism was associated with higher risks of stroke. The findings warrant corroboration by further trials of folic acid and implementation of mandatory folic acid fortification programmes for stroke prevention in low-folate populations.

Conventional and genetic associations of adiposity with 1463 proteins in relatively lean Chinese adults.

Adiposity is associated with multiple diseases and traits, but little is known about the causal relevance and mechanisms underlying these associations. Large-scale proteomic profiling, especially when integrated with genetic data, can clarify mechanisms linking adiposity with disease outcomes. We examined the associations of adiposity with plasma levels of 1463 proteins in 3977 Chinese adults, using measured and genetically-instrumented BMI. We further used two-sample bi-directional MR analyses to assess if certain proteins influenced adiposity, along with other (e.g. enrichment) analyses to clarify possible mechanisms underlying the observed associations. Overall, the mean (SD) baseline BMI was 23.9 (3.3) kg/m2, with only 6% being obese (i.e. BMI ≥ 30 kg/m2). Measured and genetically-instrumented BMI was significantly associated at FDR < 0.05 with levels of 1096 (positive/inverse: 826/270) and 307 (positive/inverse: 270/37) proteins, respectively, with FABP4, LEP, IL1RN, LSP1, GOLM2, TNFRSF6B, and ADAMTS15 showing the strongest positive and PON3, NCAN, LEPR, IGFBP2 and MOG showing the strongest inverse genetic associations. These associations were largely linear, in adiposity-to-protein direction, and replicated (> 90%) in Europeans of UKB (mean BMI 27.4 kg/m2). Enrichment analyses of the top > 50 BMI-associated proteins demonstrated their involvement in atherosclerosis, lipid metabolism, tumour progression and inflammation. Two-sample bi-directional MR analyses using cis-pQTLs identified in CKB GWAS found eight proteins (ITIH3, LRP11, SCAMP3, NUDT5, OGN, EFEMP1, TXNDC15, PRDX6) significantly affect levels of BMI, with NUDT5 also showing bi-directional association. The findings among relatively lean Chinese adults identified novel pathways by which adiposity may increase disease risks and novel potential targets for treatment of obesity and obesity-related diseases.

Self-supervised learning of accelerometer data provides new insights for sleep and its association with mortality.

Background: Sleep is essential to life. Accurate measurement and classification of sleep/wake and sleep stages is important in clinical studies for sleep disorder diagnoses and in the interpretation of data from consumer devices for monitoring physical and mental well-being. Existing non-polysomnography sleep classification techniques mainly rely on heuristic methods developed in relatively small cohorts. Thus, we aimed to establish the accuracy of wrist-worn accelerometers for sleep stage classification and subsequently describe the association between sleep duration and efficiency (proportion of total time asleep when in bed) with mortality outcomes. Methods: We developed and validated a self-supervised deep neural network for sleep stage classification using concurrent laboratory-based polysomnography and accelerometry data from three countries (Australia, the UK, and the USA). The model was validated within-cohort using subject-wise five-fold cross-validation for sleep-wake classification and in a three-class setting for sleep stage classification wake, rapid-eye-movement sleep (REM), non-rapid-eye-movement sleep (NREM) and by external validation. We assessed the face validity of our model for population inference by applying the model to the UK Biobank with 100,000 participants, each of whom wore a wristband for up to seven days. The derived sleep parameters were used in a Cox regression model to study the association of sleep duration and sleep efficiency with all-cause mortality. Findings: After exclusion, 1,448 participant nights of data were used to train the sleep classifier. The difference between polysomnography and the model classifications on the external validation was 34.7 minutes (95% limits of agreement (LoA): -37.8 to 107.2 minutes) for total sleep duration, 2.6 minutes for REM duration (95% LoA: -68.4 to 73.4 minutes) and 32.1 minutes (95% LoA: -54.4 to 118.5 minutes) for NREM duration. The derived sleep architecture estimate in the UK Biobank sample showed good face validity. Among 66,214 UK Biobank participants, 1,642 mortality events were observed. Short sleepers (<6 hours) had a higher risk of mortality compared to participants with normal sleep duration (6 to 7.9 hours), regardless of whether they had low sleep efficiency (Hazard ratios (HRs): 1.69; 95% confidence intervals (CIs): 1.28 to 2.24 ) or high sleep efficiency (HRs: 1.42; 95% CIs: 1.14 to 1.77). Interpretation: Deep-learning-based sleep classification using accelerometers has a fair to moderate agreement with polysomnography. Our findings suggest that having short overnight sleep confers mortality risk irrespective of sleep continuity.

Development and Validation of a Machine Learning Wrist-worn Step Detection Algorithm with Deployment in the UK Biobank.

Background: Step count is an intuitive measure of physical activity frequently quantified in a range of health-related studies; however, accurate quantification of step count can be difficult in the free-living environment, with step counting error routinely above 20% in both consumer and research-grade wrist-worn devices. This study aims to describe the development and validation of step count derived from a wrist-worn accelerometer and to assess its association with cardiovascular and all-cause mortality in a large prospective cohort study. Methods: We developed and externally validated a hybrid step detection model that involves self-supervised machine learning, trained on a new ground truth annotated, free-living step count dataset (OxWalk, n=39, aged 19-81) and tested against other open-source step counting algorithms. This model was applied to ascertain daily step counts from raw wrist-worn accelerometer data of 75,493 UK Biobank participants without a prior history of cardiovascular disease (CVD) or cancer. Cox regression was used to obtain hazard ratios and 95% confidence intervals for the association of daily step count with fatal CVD and all-cause mortality after adjustment for potential confounders. Findings: The novel step algorithm demonstrated a mean absolute percent error of 12.5% in free-living validation, detecting 98.7% of true steps and substantially outperforming other recent wrist-worn, open-source algorithms. Our data are indicative of an inverse dose-response association, where, for example, taking 6,596 to 8,474 steps per day was associated with a 39% [24-52%] and 27% [16-36%] lower risk of fatal CVD and all-cause mortality, respectively, compared to those taking fewer steps each day. Interpretation: An accurate measure of step count was ascertained using a machine learning pipeline that demonstrates state-of-the-art accuracy in internal and external validation. The expected associations with CVD and all-cause mortality indicate excellent face validity. This algorithm can be used widely for other studies that have utilised wrist-worn accelerometers and an open-source pipeline is provided to facilitate implementation.

DAS28(3)CRP is a reliable measure of disease activity in pregnant women with rheumatoid arthritis.

OBJECTIVES: The disease activity of rheumatoid arthritis (RA) in pregnancy is most commonly assessed with the modified Disease Activity Score (DAS)-28, the DAS28(3)CRP. However, the performance of the DAS28(3)CRP in pregnancy has not been compared to musculoskeletal ultrasound (MSK-US) as a gold standard. We performed a prospective pilot study to test the hypothesis that pregnancy-related factors limit the reliability of the DAS28(3)CRP. METHODS: Pregnant women with RA were recruited from an Obstetric Rheumatology clinic and assessed during pregnancy (second (T2) and third (T3) trimesters) and postpartum with DAS28(3)CRP and MSK-US scores, with quantification of power Doppler (PD) signal in small joints (hands and feet). Age-matched non-pregnant women with RA underwent equivalent assessments. PD scores were calculated as mean scores of all joints scanned. RESULTS: We recruited 27 pregnant and 20 non-pregnant women with RA. DAS28(3)CRP was sensitive and specific for active RA in pregnancy and postpartum as defined by positive PD signal, but not in non-pregnancy. There were significant correlations between DAS28(3)CRP and PD scores throughout pregnancy (T2, r=0.82 (95% CI [0.42, 0.95], p<0.01); T3, r=0.68 (95% CI [0.38, 0.86], p<0.01)) and postpartum, r=0.84 (95% CI [0.60, 0.94], p<0.01), while this correlation in non-pregnancy was weaker (r=0.47 (95% CI [0, 0.77], p<0.05). CONCLUSIONS: This pilot study found that DAS28(3)CRP is a reliable measure of disease activity in pregnant women with RA. Based on these data, pregnancy does not appear to confound clinical evaluation of the tender and/or swollen joint counts.

WHO cardiovascular disease risk prediction model performance in 10 regions, China.

Objective: To validate the World Health Organization (WHO) non-laboratory-based cardiovascular disease risk prediction model in regions of China. Methods: We performed an external validation of the WHO model for East Asia using the data set of China Kadoorie Biobank, an ongoing cohort study with 512 725 participants recruited from 10 regions of China from 2004-2008. We also recalculated the recalibration parameters for the WHO model in each region and evaluated the predictive performance of the model before and after recalibration. We assessed discrimination performance by Harrell's C index. Findings: We included 412 225 participants aged 40-79 years. During a median follow-up of 11 years, 58 035 and 41 262 incident cardiovascular disease cases were recorded in women and men, respectively. Harrell's C of the WHO model was 0.682 in women and 0.700 in men but varied among regions. The WHO model underestimated the 10-year cardiovascular disease risk in most regions. After recalibration in each region, discrimination and calibration were both improved in the overall population. Harrell's C increased from 0.674 to 0.749 in women and from 0.698 to 0.753 in men. The ratios of predicted to observed cases before and after recalibration were 0.189 and 1.027 in women and 0.543 and 1.089 in men. Conclusion: The WHO model for East Asia yielded moderate discrimination for cardiovascular disease in the Chinese population and had limited prediction for cardiovascular disease risk in different regions in China. Recalibration for diverse regions greatly improved discrimination and calibration in the overall population.