Objective To improve classification of movement behaviours in free-living accelerometer data using machine-learning methods, and to investigate the association between machine-learned movement behaviours and risk of incident cardiovascular disease (CVD) in adults. Methods Using free-living data from 152 participants, we developed a machine-learning model to classify movement behaviours (moderate-to-vigorous physical activity behaviours (MVPA), light physical activity behaviours, sedentary behaviour, sleep) in wrist-worn accelerometer data. Participants in UK Biobank, a prospective cohort, were asked to wear an accelerometer for 7 days, and we applied our machine-learning model to classify their movement behaviours. Using compositional data analysis Cox regression, we investigated how reallocating time between movement behaviours was associated with CVD incidence. Results In leave-one-participant-out analysis, our machine-learning method classified free-living movement behaviours with mean accuracy 88% (95% CI 87% to 89%) and Cohen’s kappa 0.80 (95% CI 0.79 to 0.82). Among 87 498 UK Biobank participants, there were 4105 incident CVD events. Reallocating time from any behaviour to MVPA, or reallocating time from sedentary behaviour to any behaviour, was associated with lower CVD risk. For an average individual, reallocating 20 min/day to MVPA from all other behaviours proportionally was associated with 9% (95% CI 7% to 10%) lower risk, while reallocating 1 hour/day to sedentary behaviour from all other behaviours proportionally was associated with 5% (95% CI 3% to 7%) higher risk. Conclusion Machine-learning methods classified movement behaviours accurately in free-living accelerometer data. Reallocating time from other behaviours to MVPA, and from sedentary behaviour to other behaviours, was associated with lower risk of incident CVD, and should be promoted by interventions and guidelines.
The availability of protein measurements and whole exome sequence data in the UK Biobank enables investigation of potential observational and genetic protein-cancer risk associations. We investigated associations of 1463 plasma proteins with incidence of 19 cancers and 9 cancer subsites in UK Biobank participants (average 12 years follow-up). Emerging protein-cancer associations were further explored using two genetic approaches, cis-pQTL and exome-wide protein genetic scores (exGS). We identify 618 protein-cancer associations, of which 107 persist for cases diagnosed more than seven years after blood draw, 29 of 618 were associated in genetic analyses, and four had support from long time-to-diagnosis ( > 7 years) and both cis-pQTL and exGS analyses: CD74 and TNFRSF1B with NHL, ADAM8 with leukemia, and SFTPA2 with lung cancer. We present multiple blood protein-cancer risk associations, including many detectable more than seven years before cancer diagnosis and that had concordant evidence from genetic analyses, suggesting a possible role in cancer development.
International differences in the incidence of many cancer types indicate the existence of carcinogen exposures that have not yet been identified by conventional epidemiology make a substantial contribution to cancer burden1. In clear cell renal cell carcinoma, obesity, hypertension and tobacco smoking are risk factors, but they do not explain the geographical variation in its incidence2. Underlying causes can be inferred by sequencing the genomes of cancers from populations with different incidence rates and detecting differences in patterns of somatic mutations. Here we sequenced 962 clear cell renal cell carcinomas from 11 countries with varying incidence. The somatic mutation profiles differed between countries. In Romania, Serbia and Thailand, mutational signatures characteristic of aristolochic acid compounds were present in most cases, but these were rare elsewhere. In Japan, a mutational signature of unknown cause was found in more than 70% of cases but in less than 2% elsewhere. A further mutational signature of unknown cause was ubiquitous but exhibited higher mutation loads in countries with higher incidence rates of kidney cancer. Known signatures of tobacco smoking correlated with tobacco consumption, but no signature was associated with obesity or hypertension, suggesting that non-mutagenic mechanisms of action underlie these risk factors. The results of this study indicate the existence of multiple, geographically variable, mutagenic exposures that potentially affect tens of millions of people and illustrate the opportunities for new insights into cancer causation through large-scale global cancer genomics.
Uncertainty remains regarding the role of diet in colorectal cancer development. We examined associations of 97 dietary factors with colorectal cancer risk in 542,778 Million Women Study participants (12,251 incident cases over 16.6 years), and conducted a targeted genetic analysis in the ColoRectal Transdisciplinary Study, Colon Cancer Family Registry, and Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO). Alcohol (relative risk per 20 g/day=1.15, 95% confidence interval 1.09-1.20) and calcium (per 300 mg/day=0.83, 0.77–0.89) intakes had the strongest associations, followed by six dairy-related factors associated with calcium. We showed a positive association with red and processed meat intake and weaker inverse associations with breakfast cereal, fruit, wholegrains, carbohydrates, fibre, total sugars, folate, and vitamin C. Genetically predicted milk consumption was inversely associated with risk of colorectal, colon, and rectal cancers. We conclude that dairy products help protect against colorectal cancer, and that this is driven largely or wholly by calcium.
Most patients who receive implantable cardioverter defibrillators (ICDs) for primary prevention do not receive therapy during the lifespan of the ICD, whilst up to 50% of sudden cardiac death (SCD) occur in individuals who are considered low risk by conventional criteria. Machine learning offers a novel approach to risk stratification for ICD assignment. Systematic search was performed in MEDLINE, Embase, Emcare, CINAHL, Cochrane Library, OpenGrey, MedrXiv, arXiv, Scopus, and Web of Science. Studies modelling SCD risk prediction within days to years using machine learning were eligible for inclusion. Transparency and quality of reporting (TRIPOD) and risk of bias (PROBAST) were assessed. A total of 4356 studies were screened with 11 meeting the inclusion criteria with heterogeneous populations, methods, and outcome measures preventing meta-analysis. The study size ranged from 122 to 124 097 participants. Input data sources included demographic, clinical, electrocardiogram, electrophysiological, imaging, and genetic data ranging from 4 to 72 variables per model. The most common outcome metric reported was the area under the receiver operator characteristic (n = 7) ranging between 0.71 and 0.96. In six studies comparing machine learning models and regression, machine learning improved performance in five. No studies adhered to a reporting standard. Five of the papers were at high risk of bias. Machine learning for SCD prediction has been under-applied and incorrectly implemented but is ripe for future investigation. It may have some incremental utility in predicting SCD over traditional models. The development of reporting standards for machine learning is required to improve the quality of evidence reporting in the field.