Alysha M. De Livera

Forecasting Time Series With Complex Seasonal Patterns Using Exponential Smoothing

An innovations state space modeling framework is introduced for forecasting complex seasonal time series such as those with multiple seasonal periods, high-frequency seasonality, non-integer seasonality, and dual-calendar effects. The new framework incorporates Box–Cox transformations, Fourier representations with time varying coefficients, and ARMA error correction. Likelihood evaluation and analytical expressions for point forecasts and interval predictions under the assumption of Gaussian errors are derived, leading to a simple, comprehensive approach to forecasting complex seasonal time series. A key feature of the framework is that it relies on a new method that greatly reduces the computational burden in the maximum likelihood estimation. The modeling framework is useful for a broad range of applications, its versatility being illustrated in three empirical studies. In addition, the proposed trigonometric formulation is presented as a means of decomposing complex seasonal time series, and it is shown that this decomposition leads to the identification and extraction of seasonal components which are otherwise not apparent in the time series plot itself.

Normalizing and integrating metabolomics data.

Metabolomics research often requires the use of multiple analytical platforms, batches of samples, and laboratories, any of which can introduce a component of unwanted variation. In addition, every experiment is subject to within-platform and other experimental variation, which often includes unwanted biological variation. Such variation must be removed in order to focus on the biological information of interest. We present a broadly applicable method for the removal of unwanted variation arising from various sources for the identification of differentially abundant metabolites and, hence, for the systematic integration of data on the same quantities from different sources. We illustrate the versatility and the performance of the approach in four applications, and we show that it has several advantages over the existing normalization methods.

Genomic prediction of coronary heart disease

Aims Genetics plays an important role in coronary heart disease (CHD) but the clinical utility of genomic risk scores (GRSs) relative to clinical risk scores, such as the Framingham Risk Score (FRS), is unclear. Our aim was to construct and externally validate a CHD GRS, in terms of lifetime CHD risk and relative to traditional clinical risk scores. Methods and results We generated a GRS of 49 310 SNPs based on a CARDIoGRAMplusC4D Consortium meta-analysis of CHD, then independently tested it using five prospective population cohorts (three FINRISK cohorts, combined n = 12 676, 757 incident CHD events; two Framingham Heart Study cohorts (FHS), combined n = 3406, 587 incident CHD events). The GRS was associated with incident CHD (FINRISK HR = 1.74, 95% confidence interval (CI) 1.61–1.86 per S.D. of GRS; Framingham HR = 1.28, 95% CI 1.18–1.38), and was largely unchanged by adjustment for known risk factors, including family history. Integration of the GRS with the FRS or ACC/AHA13 scores improved the 10 years risk prediction (meta-analysis C-index: +1.5–1.6%, P < 0.001), particularly for individuals ≥60 years old (meta-analysis C-index: +4.6–5.1%, P < 0.001). Importantly, the GRS captured substantially different trajectories of absolute risk, with men in the top 20% of attaining 10% cumulative CHD risk 12–18 y earlier than those in the bottom 20%. High genomic risk was partially compensated for by low systolic blood pressure, low cholesterol level, and non-smoking. Conclusions A GRS based on a large number of SNPs improves CHD risk prediction and encodes different trajectories of lifetime risk not captured by traditional clinical risk scores.

Statistical methods for handling unwanted variation in metabolomics data

Metabolomics experiments are inevitably subject to a component of unwanted variation, due to factors such as batch effects, long runs of samples, and confounding biological variation. Although the removal of this unwanted variation is a vital step in the analysis of metabolomics data, it is considered a gray area in which there is a recognized need to develop a better understanding of the procedures and statistical methods required to achieve statistically relevant optimal biological outcomes. In this paper, we discuss the causes of unwanted variation in metabolomics experiments, review commonly used metabolomics approaches for handling this unwanted variation, and present a statistical approach for the removal of unwanted variation to obtain normalized metabolomics data. The advantages and performance of the approach relative to several widely used metabolomics normalization approaches are illustrated through two metabolomics studies, and recommendations are provided for choosing and assessing the most suitable normalization method for a given metabolomics experiment. Software for the approach is made freely available.

Elemental and metabolite profiling of nickel hyperaccumulators from New Caledonia.

Leaf material from nine Ni hyperaccumulating species was collected in New Caledonia: Homalium kanaliense (Vieill.) Briq., Casearia silvana Schltr, Geissois hirsuta Brongn. & Gris, Hybanthus austrocaledonicus Seem, Psychotria douarrei (G. Beauvis.) Däniker, Pycnandra acuminata (Pierre ex Baill.) Swenson & Munzinger (syn Sebertia acuminata Pierre ex Baill.), Geissois pruinosa Brongn. & Gris, Homalium deplanchei (Viell) Warb. and Geissois bradfordii (H.C. Hopkins). The elemental concentration was determined by inductively-coupled plasma optical emission spectrometry (ICP-OES) and from these results it was found that the species contained Ni concentrations from to 250-28,000 mg/kg dry mass. Gas chromatography mass spectrometry (GC-MS)-based metabolite profiling was then used to analyse leaves of each species. The aim of this study was to target Ni-binding ligands through correlation analysis of the metabolite levels and leaf Ni concentration. Approximately 258 compounds were detected in each sample. As has been observed before, a correlation was found between the citric acid and Ni concentrations in the leaves for all species collected. However, the strongest Ni accumulator, P. douarrei, has been found to contain particularly high concentrations of malonic acid, suggesting an additional storage mechanism for Ni. A size exclusion chromatography separation protocol for the separation of Ni-complexes in P. acuminata sap was also applied to aqueous leaf extracts of each species. A number of metabolites were identified in complexes with Ni including Ni-malonate from P. douarrei. Furthermore, the levels for some metabolites were found to correlate with the leaf Ni concentration. These data show that Ni ions can be bound by a range of small molecules in Ni hyperaccumulation in plants.

Quantifying Low Birth Weight, Preterm Birth and Small-for-Gestational-Age Effects of Malaria in Pregnancy: A Population Cohort Study

Background The association between malaria during pregnancy and low birth weight (LBW) is well described. This manuscript aims to quantify the relative contribution of malaria to small-for-gestational-age (SGA) infants and preterm birth (PTB) in pregnancies accurately dated by ultrasound on the Thai-Myanmar border at the Shoklo Malaria Research Unit. Methods and Findings From 2001 to 2010 in a population cohort of prospectively followed pregnancies, we analyzed all singleton newborns who were live born, normal, weighed in the first hour of life and with a gestational age (GA) between 28+0 and 41+6 weeks. Fractional polynomial regression was used to determine the mean birthweight and standard deviation as functions of GA. Risk differences and factors of LBW and SGA were studied across the range of GA for malaria and non-malaria pregnancies. From 10,264 newborns records, population centiles were created. Women were screened for malaria by microscopy a median of 22 [range 1–38] times and it was detected and treated in 12.6% (1,292) of pregnancies. Malaria was associated with LBW, PTB, and SGA compared to those without malaria. Nearly two-thirds of PTB were classified as LBW (68% (539/789)), most of which 83% (447/539) were not SGA. After GA 39 weeks, 5% (298/5,966) of non-LBW births were identified as SGA. Low body mass index, primigravida, hypertension, smoking and female sex of the newborn were also significantly and independently associated with LBW and SGA consistent with previous publications. Conclusions Treated malaria in pregnancy was associated with an increased risk for LBW, PTB, and SGA, of which the latter are most important for infant survival. Using LBW as an endpoint without adjusting for GA incorrectly estimated the effects of malaria in pregnancy. Ultrasound should be used for dating pregnancies and birth weights should be expressed as a function (or adjusted for GA) of GA in future malaria in pregnancy studies.

Cross-Platform Urine Metabolomics of Experimental Hyperglycemia in Type 2 Diabetes

Hyperglycemia causes diabetic nephropathy, a condition for which there are no specific diagnostic markers that predict progression to renal failure. Here we describe a multiplatform metabolomic analysis of urine from individuals with type 2 diabetes, collected before and immediately following experimental hyperglycemia. We used targeted nuclear magnetic resonance spectroscopy (NMR), liquid chromatography - mass spectrometry (LC-MS) and gas chromatography - MS (GC-MS) to identify markers of hyperglycemia. Following optimization of data normalisation and statistical analysis, we identified a reproducible NMR and LC-MS based urine signature of hyperglycemia. Significant increases of alanine, alloisoleucine, isoleucine, leucine, N-isovaleroylglycine, valine, choline, lactate and taurine and decreases of arginine, gamma-aminobutyric acid, hippurate, suberate and N-acetylglutamate were observed. GC-MS analysis identified a number of metabolites differentially present in post-glucose versus baseline urine, but these could not be identified using current metabolite libraries. This analysis is an important first step towards identifying biomarkers of early-stage diabetic nephropathy.

Associations of Lifestyle, Medication, and Socio-Demographic Factors with Disability in People with Multiple Sclerosis: An International Cross-Sectional Study

Objective Emerging evidence links modifiable lifestyle risk factors to disease progression in multiple sclerosis (MS). We sought further evidence around this hypothesis through detailed analysis of the association with disability of lifestyle behaviours of a large international sample of people with MS. Materials and Methods A total of 2469 people with MS from 57 countries provided self-reported data via cross-sectional online survey on lifestyle (mostly with validated tools) and the primary outcome measure, disability (Patient Determined Disease Steps), categorised from 8 steps into 3 categories, mild, moderate and major disability. Multinomial logistic regression modelling derived relative risk ratios (RRRs) for disability categories. Results RRRs of having moderate vs mild disability were: diet (per 30 points on 100 point scale) 0.72 (95%CI 0.52–0.98), ever smoking 1.32 (1.06–1.65), exercise (moderate/high vs low) 0.35 (0.28–0.44), latitude (per degree from the equator) 1.02 (1.01–1.04), and number of comorbidities (2 vs none) 1.43 (1.04–1.95), (3 vs none) 1.56 (1.13–2.16). RRRs of having major vs mild disability were: exercise (moderate/high vs low) 0.07 (0.04–0.11), alcohol consumption (moderate vs low) 0.45 (0.30–0.68), plant-based omega 3 supplementation 0.39 (0.18–0.86), and disease-modifying medication use 0.45 (0.29–0.70). Conclusions Healthier lifestyle has strong associations with disability in our large international sample of people with MS, supporting further investigation into the role of lifestyle risk factors in MS disease progression.

Statistical Analysis of Metabolomics Data

Statistical matters form an integral part of a metabolomics experiment. In this chapter we describe several important aspects in the analysis of metabolomics data such as the removal of unwanted variation and the identification of differentially abundant metabolites, along with a number of other essential statistical considerations.

Models for the analysis of repeated continuous outcome measures in clinical trials.

Repeated continuous outcome measures are common in clinical trials. In this tutorial style paper, using data collected from a trial evaluating an intervention for managing asthma and chronic obstructive pulmonary disease, we demonstrate ways of statistically analysing such data to answer frequently encountered clinical research questions. We illustrate the use of linear mixed effects modelling in doing so and discuss its advantages over several other commonly used approaches. The methods described in this paper can easily be carried out using standard statistical software.