Evolutionary methods for variable selection in the epidemiological modeling of cardiovascular diseases

The association between five socioeconomic indices (lifetime occupation, education, income, ownership of material possessions, and childhood socioeconomic status) and plasma fibrinogen levels was investigated in middle-aged Finnish men who were part of the Kuopio Ischemic Heart Disease Risk Factor Study. The Kuopio Ischemic Heart Disease Risk Factor Study is based on a representative age-stratified sample of 2,682 men aged 42, 48, 54, and 60 years. The data were collected between 1984 and 1989. The present analysis is restricted to the 2,011 men for whom information on fibrinogen and all covariates was available. The covariates were alcohol consumption, body mass index, physical fitness, smoking, coffee consumption, high density lipoprotein cholesterol, low density lipoprotein cholesterol, blood leukocyte count, and prevalent disease (at least one sign of ischemic heart disease, hypertension, diabetes, or previous stroke). An age-adjusted inverse association was found between levels of plasma fibrinogen and four of the five socioeconomic indices: current income, education, lifetime occupation status, and current material possessions. After adjustment for the covariates, the association persisted for education, current income, and lifetime occupation. Analysis of the joint effect of childhood and adult socioeconomic status indicated that those who were economically disadvantaged at both times had the highest fibrinogen levels, but the fibrinogen levels of those who were not poor as adults had no variation by childhood socioeconomic status.

Most common chronic diseases have a multifaceted aetiological background. Because currently used statistical methods have severe limitations in describing complex non-linear processes, the authors evaluated the usefulness of a multivariate method which is able to describe non-linear phenomena, the self-organizing map (SOM). The study subjects were the 1650 participants of the Kuopio Ischemic Heart Disease Risk Factor Study (KIHD). The SOM model was constructed using 25 continuous biochemical and physiological variables. The aim of the SOM algorithm, together with Sammon's mapping, is to group the data into reduced but representative format and divide the study population into homogeneous subgroups. The study population consisted of four groups (clusters) according to the method used. In the clusters C1 to C4 were 637, 445, 275 and 121 men, respectively. There were eight neurons (n = 172) which were not included to the four main clusters. The mean values of the variables related to insulin resistance syndrome in the identified SOM map were 32.1 (kg/m(2)) for body mass index (BMI), 1.01 for waist-to-hip ratio (WHR), 158.7 mmHg and 103.8 mmHg for systolic (SBP) and diastolic blood pressure (DBP), 2.8 mmol/l for triglycerides, 6.2 mmol/l for blood glucose and 22.4 mU/l for serum insulin. There was a statistically significant difference in the mean values of BMI, WHR, SBP, DBP, HDL, triglycerides and blood glucose between the cluster representing the insulin resistance syndrome and the normal cluster. This study shows that the multidimensional structures of insulin resistance syndrome can be visualized and identified at qualitative and quantitative level using the SOM algorithm.

Egg consumption, cholesterol intake, and risk of incident stroke in men: the Kuopio Ischaemic Heart Disease Risk Factor Study

Dietary proteins and protein sources and risk of death: the Kuopio Ischaemic Heart Disease Risk Factor Study

Association of dietary cholesterol and egg intakes with the risk of incident dementia or Alzheimer disease: the Kuopio Ischaemic Heart Disease Risk Factor Study ,

PurposeTo investigate associations of total dietary choline intake and its major dietary form, phosphatidylcholine, with type 2 diabetes risk.MethodsWe included 2332 men aged 42–60 years at baseline in 1984–1989 from the Kuopio Ischaemic Heart Disease Risk Factor Study in eastern Finland. Dietary intakes were assessed with 4-d food recording at baseline. Type 2 diabetes diagnosis was based on self-administered questionnaires, fasting and 2-h oral glucose tolerance test blood glucose measurements, or by record linkage to national health registries. Multivariable-adjusted Cox proportional hazards regression models were used for statistical analysis.ResultsDuring the mean 19.3-year follow-up, 432 men had type 2 diabetes diagnosis. After multivariable adjustments, those in the highest vs. lowest choline intake quartile had 25% (95% CI 2–43%) lower relative risk (P trend across quartiles = 0.02) and those in the highest vs. lowest phosphatidylcholine quartile had 41% (95% CI 22–55%) lower relative risk (P trend < 0.001) of type 2 diabetes.ConclusionsHigher choline intake, especially phosphatidylcholine, was associated with lower type 2 diabetes risk among men.

The role of flavonoids in CVD, especially in strokes, is unclear. Our aim was to study the role of flavonoids in CVD. We studied the association between the intakes of five subclasses (flavonols, flavones, flavanones, flavan-3-ols and anthocyanidins), a total of twenty-six flavonoids, on the risk of ischaemic stroke and CVD mortality. The study population consisted of 1950 eastern Finnish men aged 42-60 years free of prior CHD or stroke as part of the prospective population-based Kuopio Ischaemic Heart Disease Risk Factor Study. During an average follow-up time of 15.2 years, 102 ischaemic strokes and 153 CVD deaths occurred. In the Cox proportional hazards model adjusted for age and examination years, BMI,systolic blood pressure, hypertension medication, serum HDL- and LDL-cholesterol, serum TAG, maximal oxygen uptake, smoking, family history of CVD, diabetes, alcohol intake, energy-adjusted intake of folate, vitamin E, total fat and saturated fat intake (percentage of energy), men in the highest quartile of flavonol and flavan-3-ol intakes had a relative risk of 0.55 (95% CI 0.31, 0.99) and 0.59 (95% CI 0.30, 1.14) for ischaemic stroke, respectively, as compared with the lowest quartile. After multivariate adjustment, the relative risk for CVD death in the highest quartile of flavanone and flavone intakes were 0.54 (95% CI 0.32, 0.92) and 0.65 (95% CI 0.40, 1.05), respectively. The present results suggest that high intakes of flavonoids may be associated with decreased risk of ischaemic stroke and possibly with reduced CVD mortality.

Association between serum zinc and later development of metabolic syndrome in middle aged and older men: The Kuopio Ischaemic Heart Disease Risk Factor Study

Based on case-control and prospective studies elevated blood total homocysteine (tHcy) has been suggested to be an independent risk factor for cardiovascular diseases (CVD). The purpose of the study was to explore the joint effect of increased serum tHcy concentration and other risk factors on the risk of CVD mortality in middle-aged men without a history of heart disease or stroke. A prospective, population-based Kuopio Ischaemic Heart Disease Risk Factor (KIHD) Study. Eastern Finland. Subjects. A total of 802 men aged 46-64 years, examined in 1991-93. CVD mortality event. The mean serum tHcy concentration was 10.8 micromol L(-1) (SD 3.3). During the average follow-up time of 10.8 years 50 men experienced a CVD death. The hazard rate ratio for CVD mortality was 1.80 (95% confidence interval: 1.02-3.19) in men in the highest serum tHcy third versus lower thirds after adjustment for cardiovascular risk factors. Furthermore, elevated serum tHcy concentration appeared to increase the risk of CVD death in men who smoke or who have high circulating concentrations of serum total or LDL cholesterol, apo-B apolipoprotein or plasma fibrinogen. We conclude that homocysteine may increase the risk of CVD mortality in middle-aged men from Eastern Finland, and it may especially increase the risk when present with other risk factors for CVD.

Serum Lycopene and the Risk of Cancer: The Kuopio Ischaemic Heart Disease Risk Factor (KIHD) Study

In this paper we compare a number of two-criterion filtering techniques for feature selection in cardiovascular predictive modelling. We design two-objective schemes based on different combinations of four criteria describing the quality of reduced feature sets. To find attribute subsystems meeting the introduced criteria in an optimal way, we suggest applying a cooperative multi-objective genetic algorithm. It includes various search strategies working in a parallel way, which allows additional experiments to be avoided when choosing the most effective heuristic for the problem considered. The performance of filtering techniques was investigated in combination with the SVM model on a population-based epidemiological database called KIHD (Kuopio Ischemic Heart Disease Risk Factor Study). The dataset consists of a large number of variables on various characteristics of the study participants. These baseline measures were collected at the beginning of the study. In addition, all major cardiovascular events that had occurred among the participants over an average of 27 years of follow-up were collected from the national health registries. As a result, we found that the usage of the filtering technique including intra- and inter-class distances led to a significant reduction of the feature set (up to 11 times, from 433 to 38 features) without detriment to the predictive ability of the SVM model. This implies that there is a possibility to cut down on the clinical tests needed to collect the data, which is relevant to the prediction of cardiovascular diseases.

To investigate the association between serum 25-hydroxyvitamin D [25(OH)D] concentration, a marker of vitamin D status, and risk of all-cause and cardiovascular mortality in a general older population with relatively low average serum 25(OH)D concentrations. The study population included 552 men and 584 women aged 53-73 years who were free of CVD and cancer at baseline in 1998-2001 from the prospective, population-based Kuopio Ischaemic Heart Disease Risk Factor (KIHD) Study. Deaths were ascertained by a computer linkage to the national cause of death register. All deaths that occurred from the study entry to December 31, 2008, were included. Cox proportional hazards regression models were used to analyze the association between serum 25(OH)D and risk of death. The mean serum 25(OH)D concentration was 43.7 nmol/L (SD 17.8), with a strong seasonal variation. During the average follow-up of 9.1 years, 87 participants died, 35 from cardiovascular disease (CVD). After multivariable-adjustments, the hazard ratios (HR) for all cause death in the tertiles of serum 25(OH)D were 1, 1.68 (95% CI: 0.92, 3.07) and 2.06 (95% CI: 1.12, 3.80), p for trend = 0.02. Our study supports the accumulating evidence from epidemiological studies that vitamin D deficiency is associated with increased risk of death. Large-scale primary prevention trials with vitamin D supplementation are warranted.

Predictive models driven by time-to-event data are commonly used in survival analysis. Owing to the availability of high-dimensional epidemiological cohorts, there is a need for models and learning algorithms capable of utilizing hundreds or even thousands of predictors. Advanced machine learning tools with embedded variable selection are being modified for use with time-to-event data in the presence of competing risks and censoring. In this study, random survival forests were compared to the cause-specific and sub-distribution hazard models widely applied in survival analysis with competing risks. Using the extensive cohort data of 2,682 subjects and 950 predictors collected within the Kuopio Ischemic Heart Disease Risk Factor Study (1984–1989), we built models to predict cardiovascular and non-cardiovascular deaths for a 30-year prediction horizon in two scenarios: first, on a subset of manually selected risk factors and, second, on hundreds of more freely chosen available predictors. The experimental results show that without a manual pre-selection of predictors, the random forest reaches the highest area under the curve (AUC). In addition, this study presents a modeling pipeline, including a non-parametric analysis, missing data imputation, model validation, and result assessment.

Animal and plant protein intakes have indicated opposite associations with cardiovascular mortality risk. Whether dietary proteins are associated with risk of heart failure (HF) is unclear. Thus, we examined the associations of proteins from different food sources with risk of HF. The study included 2441 men aged 42 to 60 years at the baseline examinations in 1984 to 1989 in the Kuopio Ischaemic Heart Disease Risk Factor Study. Protein intakes at baseline were assessed with 4-day dietary records. Data on incident HF cases were obtained from national registers. HF risk according to protein intake was estimated by Cox proportional hazard ratios. During the mean follow-up of 22.2 years, 334 incident HF cases occurred. Higher intake of total protein indicated a trend toward increased risk of HF (multivariable-adjusted hazard ratio in the highest versus lowest quartile=1.33; 95% confidence interval: 0.95-1.85; P-trend=0.05). The associations between specific types and sources of protein with incident HF were consistent with this overall finding although not all associations reached statistical significance. For example, the hazard ratio in the highest versus lowest quartile was 1.43 (95% confidence interval: 1.00-2.03; P-trend=0.07) for total animal protein and 1.17 (95% confidence interval: 0.72-1.91; P-trend=0.35) for total plant protein. In middle-aged men, higher protein intake was marginally associated with increased risk of HF. URL: https://www.clinicaltrials.gov. Unique identifier: NCT03221127.

Benefits of life-long physical activity (PA) are widely recognized. Fewer studies have investigated how being young competing athlete reflects to later life body composition and PA. PURPOSE: To investigate the impact of participation in competitive sports in youth on the body composition and subjectively and objectively measured PA at middle- age. METHODS: The study participants are 47-to-55-year old Finnish women (n=985) attending to the Estrogenic Regulation of Muscle Apoptosis study (ERMA). Their participation in competitive sports at the age of 13-to-16 was assessed with retrospective self-report. Midlife lean body mass (LBM), skeletal muscle mass (SMM) and fat mass (FM) were measured with bioimpedance (Inbody 720) after overnight fast (n=866). Midlife PA included 7-scaled self-estimate that was reclassified to form sedentary, light, moderate and vigorous PA categories (n=985), a detailed assessment of leisure-time PA volume, duration and intensity (a modified Kuopio Ischemic Heart Disease Risk Factor Study Questionnaire, n=982) enabling calculation of mean daily metabolic equivalent (MET-h/day) index as a proxy of the previous 12 months PA and objectively measured total PA over seven-days by hip-worn accelerometer (Actigraph GT3X+ or wGT3X+, n=734). Statistical differences were tested using chi-squared test, independent samples t-test or Mann-Whitney U test. RESULTS: Participants were assigned into youth athlete (n=136) or non-athlete (n=849) groups based on their participation in competitive sports at the age of 13-to-16. Youth athlete group had 1.5 kg higher LBM (p=0.002) and 1.0 kg higher SMM (p=0.001) at middle-age, but no statistically significant difference in FM. Youth athletes reported more vigorous PA (32.4% vs. 20.3%, p=0.007) and higher leisure-time MET index (4.3 vs. 3.4 h/day, p<0.001) than non-athletes. No difference was observed in accelerometer-measured sedentary time, light PA, moderate-to-vigorous PA or total PA (counts) between groups. CONCLUSION: Participation in sport competitions as a teenager has long lasting health benefits that associates with healthier body composition and higher leisure-time PA later in life. However, the discrepancy between self-reports and objective measures needs further studies to delineate the benefits of different components of PA.