Review of New Statistical Techniques for Analysis of Cardiovascular Trial and Registry Data.

Cardiovascular research generates a large variety of data including big and small datasets that pose interesting challenges to researchers in biomedical statistics. This paper covers a number of important methodology issues and the solutions that have been applied so far to try to solve them. The first issue is multiple testing and the use of false discovery rates to correct for multiplicity. This is illustrated with an application related to genotyping heart diseases. The second is measuring life extension at present and in the future. The third one is the issue of U and J curves and the questions they present. Finally, this paper studies visit-to-visit blood pressure variability as a predictor of cardiovascular events.

Title-molecular diagnostics of dystrophinopathies in Sri Lanka towards phenotype predictions: an insight from a South Asian resource limited setting.

BACKGROUND: The phenotype of Duchenne muscular dystrophy (DMD) and Becker muscular dystrophy (BMD) patients is determined by the type of DMD gene variation, its location, effect on reading frame, and its size. The primary objective of this investigation was to determine the frequency and distribution of DMD gene variants (deletions/duplications) in Sri Lanka through the utilization of a combined approach involving multiplex polymerase chain reaction (mPCR) followed by Multiplex Ligation Dependent Probe Amplification (MLPA) and compare to the international literature. The current consensus is that MLPA is a labor efficient yet expensive technique for identifying deletions and duplications in the DMD gene. METHODOLOGY: Genetic analysis was performed in a cohort of 236 clinically suspected pediatric and adult myopathy patients in Sri Lanka, using mPCR and MLPA. A comparative analysis was conducted between our findings and literature data. RESULTS: In the entire patient cohort (n = 236), mPCR solely was able to identify deletions in the DMD gene in 131/236 patients (DMD-120, BMD-11). In the same cohort, MLPA confirmed deletions in 149/236 patients [DMD-138, BMD -11]. These findings suggest that mPCR has a detection rate of 95% (131/138) among all patients who received a diagnosis. The distal and proximal deletion hotspots for DMD were exons 45-55 and 6-15. Exon 45-60 identified as a novel in-frame variation hotspot. Exon 45-59 was a hotspot for BMD deletions. Comparisons with the international literature show significant variations observed in deletion and duplication frequencies in DMD gene across different populations. CONCLUSION: DMD gene deletions and duplications are concentrated in exons 45-55 and 2-20 respectively, which match global variation hotspots. Disparities in deletion and duplication frequencies were observed when comparing our data to other Asian and Western populations. Identified a 95% deletion detection rate for mPCR, making it a viable initial molecular diagnostic approach for low-resource countries where MLPA could be used to evaluate negative mPCR cases and cases with ambiguous mutation borders. Our findings may have important implications in the early identification of DMD with limited resources in Sri Lanka and to develop tailored molecular diagnostic algorithms that are regional and population specific and easily implemented in resource limited settings.

Duchenne Muscular Dystrophy from Brain to Muscle: The Role of Brain Dystrophin Isoforms in Motor Functions.

Brain function and its effect on motor performance in Duchenne muscular dystrophy (DMD) is an emerging concept. The present study explored how cumulative dystrophin isoform loss, age, and a corticosteroid treatment affect DMD motor outcomes. A total of 133 genetically confirmed DMD patients from Sri Lanka were divided into two groups based on whether their shorter dystrophin isoforms (Dp140, Dp116, and Dp71) were affected: Group 1, containing patients with Dp140, Dp116, and Dp71 affected (n = 98), and Group 2, containing unaffected patients (n = 35). A subset of 52 patients (Group 1, n = 38; Group 2, n = 14) was followed for up to three follow-ups performed in an average of 28-month intervals. The effect of the cumulative loss of shorter dystrophin isoforms on the natural history of DMD was analyzed. A total of 74/133 (56%) patients encountered developmental delays, with 66/74 (89%) being in Group 1 and 8/74 (11%) being in Group 2 (p < 0.001). Motor developmental delays were predominant. The hip and knee muscular strength, according to the Medical Research Council (MRC) scale and the North Star Ambulatory Assessment (NSAA) activities, "standing on one leg R", "standing on one leg L", and "walk", declined rapidly in Group 1 (p < 0.001 In the follow-up analysis, Group 1 patients became wheelchair-bound at a younger age than those of Group 2 (p = 0.004). DMD motor dysfunction is linked to DMD mutations that affect shorter dystrophin isoforms. When stratifying individuals for clinical trials, considering the DMD mutation site and its impact on a shorter dystrophin isoform is crucial.

Integrated genomic, proteomic and cognitive assessment in Duchenne Muscular Dystrophy suggest astrocyte centric pathology.

Introduction: Documented Duchenne Muscular Dystrophy (DMD) biomarkers are confined to Caucasians and are poor indicators of cognitive difficulties and neuropsychological alterations. Materials and methods: This study correlates serum protein signatures with cognitive performance in DMD patients of South Asian origin. Study included 25 DMD patients aged 6-16 years. Cognitive profiles were assessed by Wechsler Intelligence Scale for Children. Serum proteome profiling of 1317 proteins was performed in eight DMD patients and eight age-matched healthy volunteers. Results: Among the several novel observations we report, better cognitive performance in DMD was associated with increased serum levels of MMP9 and FN1 but decreased Siglec-3, C4b, and C3b. Worse cognitive performance was associated with increased serum levels of LDH-H1 and PDGF-BB but reduced GDF-11, MMP12, TPSB2, and G1B. Secondly, better cognitive performance in Processing Speed (PSI) and Perceptual Reasoning (PRI) domains was associated with intact Dp116, Dp140, and Dp71 dystrophin isoforms while better performance in Verbal Comprehension (VCI) and Working Memory (WMI) domains was associated with intact Dp116 and Dp140 isoforms. Finally, functional pathways shared with Alzheimer's Disease (AD) point towards an astrocyte-centric model for DMD. Conclusion: Astrocytic dysfunction leading to synaptic dysfunction reported previously in AD may be a common pathogenic mechanism underlying both AD and DMD, linking protein alterations to cognitive impairment. This new insight may pave the path towards novel therapeutic approaches targeting reactive astrocytes.

Socio-economic impact on COVID-19 cases and deaths and its evolution in New Jersey.

Socio-economic factors could impact how epidemics spread. In this study, we investigated the possible effect of several local socio-economic factors on the case count and time course of confirmed Covid-19 cases and Covid-19-related deaths across the twenty one counties of New Jersey. Socio-economic and geographic factors considered included population, percentage of elders in the population, percentage of low-income households, access to food and health facilities and distance to New York. We found that the counties could be clustered into three groups based on (a) the case totals, (b) the total number of deaths, (c) the time course of the cases and (d) the time course of the deaths. The four groupings were very similar to one another and could all be largely explained by the county population, the percentage of low-income population, and the distance of the county from New York. As for food and health factors, the numbers of local restaurants and pharmacies significantly influenced the total number of cases and deaths as well as the epidemic's evolution. In particular, the number of cases and deaths showed a decrease with the number of McDonald's within the county in contrast to other fast-food or non-fast food restaurants. Overall, our study found that the evolution of the epidemic was influenced by certain socio-economic factors, which could be helpful for the formulation of public health policies.

Informative or noninformative calls for gene expression: a latent variable approach.

The strength and weakness of microarray technology can be attributed to the enormous amount of information it is generating. To fully enhance the benefit of microarray technology for testing differentially expressed genes and classification, there is a need to minimize the amount of irrelevant genes present in microarray data. A major interest is to use probe-level data to call genes informative or noninformative based on the trade-off between the array-to-array variability and the measurement error. Existing works in this direction include filtering likely uninformative sets of hybridization (FLUSH; Calza et al., 2007) and I/NI calls for the exclusion of noninformative genes using FARMS (I/NI calls; Talloen et al., 2007; Hochreiter et al., 2006). In this paper, we propose a linear mixed model as a more flexible method that performs equally good as I/NI calls and outperforms FLUSH. We also introduce other criteria for gene filtering, such as, R2 and intra-cluster correlation. Additionally, we include some objective criteria based on likelihood ratio testing, the Akaike information criteria (AIC; Akaike, 1973) and the Bayesian information criterion (BIC; Schwarz, 1978 ). Based on the HGU-133A Spiked-in data set, it is shown that the linear mixed model approach outperforms FLUSH, a method that filters genes based on a quantile regression. The linear model is equivalent to a factor analysis model when either the factor loadings are set to a constant with the variance of the latent factor equal to one, or if the factor loadings are set to one together with unconstrained variance of the latent factor. Filtering based on conditional variance calls a probe set informative when the intensity of one or more probes is consistent across the arrays, while filtering using R2 or intra-cluster correlation calls a probe set informative only when average intensity of a probe set is consistent across the arrays. Filtering based on likelihood ratio test AIC and BIC are less stringent compared to the other criteria.

Types of the cerebral arterial circle (circle of Willis) in a Sri Lankan population.

BACKGROUND: The variations of the circle of Willis (CW) are clinically important as patients with effective collateral circulations have a lower risk of transient ischemic attack and stroke than those with ineffective collaterals. The aim of the present cadaveric study was to investigate the anatomical variations of the CW and to compare the frequency of prevalence of the different variations with previous autopsy studies as variations in the anatomy of the CW as a whole have not been studied in the Indian subcontinent. METHODS: The external diameter of all the arteries forming the CW in 225 normal Sri Lankan adult cadaver brains was measured using a calibrated grid to determine the prevalence in the variation in CW. Chisquared tests and a correspondence analysis were performed to compare the relative frequencies of prevalence of anatomical variations in the CW across 6 studies of diverse ethnic populations. RESULTS: We report 15 types of variations of CW out of 22 types previously described and one additional type: hypoplastic precommunicating part of the anterior cerebral arteries (A1) and contralateral posterior communicating arteries (PcoA) 5(2%). Statistically significant differences (p < 0.0001) were found between most of the studies except for the Moroccan study. An especially notable difference was observed in the following 4 configurations: 1) hypoplastic precommunicating part of the posterior cerebral arteries (P1), and contralateral A1, 2) hypoplastic PcoA and contralateral P1, 3) hypoplastic PcoA, anterior communicating artery (AcoA) and contralateral P1, 4) bilateral hypoplastic P1s and AcoA in a Caucasian dominant study by Fisher versus the rest of the studies. CONCLUSION: The present study reveals that there are significant variations in the CW among intra and inter ethnic groups (Caucasian, African and Asian: Iran and Sri Lanka dominant populations), and warrants further studies keeping the methods of measurements, data assessment, and the definitions of hypoplasia the same.

Uses and opportunities for machine learning in hypertension research.

BACKGROUND: Artificial intelligence (AI) promises to provide useful information to clinicians specializing in hypertension. Already, there are some significant AI applications on large validated data sets. METHODS AND RESULTS: This review presents the use of AI to predict clinical outcomes in big data i.e. data with high volume, variety, veracity, velocity and value. Four examples are included in this review. In the first example, deep learning and support vector machine (SVM) predicted the occurrence of cardiovascular events with 56%-57% accuracy. In the second example, in a data base of 378,256 patients, a neural network algorithm predicted the occurrence of cardiovascular events during 10 year follow up with sensitivity (68%) and specificity (71%). In the third example, a machine learning algorithm classified 1,504,437 patients on the presence or absence of hypertension with 51% sensitivity, 99% specificity and area under the curve 87%. In example four, wearable biosensors and portable devices were used in assessing a person's risk of developing hypertension using photoplethysmography to separate persons who were at risk of developing hypertension with sensitivity higher than 80% and positive predictive value higher than 90%. The results of the above studies were adjusted for demographics and the traditional risk factors for atherosclerotic disease. CONCLUSION: These examples describe the use of artificial intelligence methods in the field of hypertension.

Enriched random forests.

Although the random forest classification procedure works well in datasets with many features, when the number of features is huge and the percentage of truly informative features is small, such as with DNA microarray data, its performance tends to decline significantly. In such instances, the procedure can be improved by reducing the contribution of trees whose nodes are populated by non-informative features. To some extent, this can be achieved by prefiltering, but we propose a novel, yet simple, adjustment that has demonstrably superior performance: choose the eligible subsets at each node by weighted random sampling instead of simple random sampling, with the weights tilted in favor of the informative features. This results in an 'enriched random forest'. We illustrate the superior performance of this procedure in several actual microarray datasets.

Genome-wide copy number alterations detection in fresh frozen and matched FFPE samples using SNP 6.0 arrays.

SNP arrays offer the opportunity to get a genome-wide view on copy number alterations and are increasingly used in oncology. DNA from formalin-fixed paraffin-embedded material (FFPE) is partially degraded which limits the application of those technologies for retrospective studies. We present the use of Affymetrix GeneChip SNP6.0 for identification of copy number alterations in fresh frozen (FF) and matched FFPE samples. Fifteen pairs of adenocarcinomas with both frozen and FFPE embedded material were analyzed. We present an optimization of the sample preparation and show the importance of correcting the measured intensities for fragment length and GC-content when using FFPE samples. The absence of GC content correction results in a chromosome specific "wave pattern" which may lead to the misclassification of genomic regions as being altered. The highest concordance between FFPE and matched FF were found in samples with the highest call rates. Nineteen of the 23 high level amplifications (83%) seen using FF samples were also detected in the corresponding FFPE material. For limiting the rate of "false positive" alterations, we have chosen a conservative False Discovery Rate (FDR). We observed better results using SNP probes than CNV probes for copy number analysis of FFPE material. This is the first report on the detection of copy number alterations in FFPE samples using Affymetrix GeneChip SNP6.0.

I/NI-calls for the exclusion of non-informative genes: a highly effective filtering tool for microarray data.

MOTIVATION: DNA microarray technology typically generates many measurements of which only a relatively small subset is informative for the interpretation of the experiment. To avoid false positive results, it is therefore critical to select the informative genes from the large noisy data before the actual analysis. Most currently available filtering techniques are supervised and therefore suffer from a potential risk of overfitting. The unsupervised filtering techniques, on the other hand, are either not very efficient or too stringent as they may mix up signal with noise. We propose to use the multiple probes measuring the same target mRNA as repeated measures to quantify the signal-to-noise ratio of that specific probe set. A Bayesian factor analysis with specifically chosen prior settings, which models this probe level information, is providing an objective feature filtering technique, named informative/non-informative calls (I/NI calls). RESULTS: Based on 30 real-life data sets (including various human, rat, mice and Arabidopsis studies) and a spiked-in data set, it is shown that I/NI calls is highly effective, with exclusion rates ranging from 70% to 99%. Consequently, it offers a critical solution to the curse of high-dimensionality in the analysis of microarray data. AVAILABILITY: This filtering approach is publicly available as a function implemented in the R package FARMS (www.bioinf.jku.at/software/farms/farms.html).

The high-level similarity of some disparate gene expression measures.

Probe-level data from Affymetrix GeneChips can be summarized in many ways to produce probe-set level gene expression measures (GEMs). Disturbingly, the different approaches not only generate quite different measures but they could also yield very different analysis results. Here, we explore the question of how much the analysis results really do differ, first at the gene level, then at the biological process level. We demonstrate that, even though the gene level results may not necessarily match each other particularly well, as long as there is reasonably strong differentiation between the groups in the data, the various GEMs do in fact produce results that are similar to one another at the biological process level. Not only that the results are biologically relevant. As the extent of differentiation drops, the degree of concurrence weakens, although the biological relevance of findings at the biological process level may yet remain.

A consolidated approach to analyzing data from high-throughput protein microarrays with an application to immune response profiling in humans.

MOTIVATION: DNA microarrays are a well-known and established technology in biological and pharmaceutical research providing a wealth of information essential for understanding biological processes and aiding drug development. Protein microarrays are quickly emerging as a follow-up technology, which will also begin to experience rapid growth as the challenges in protein to spot methodologies are overcome. Like DNA microarrays, their protein counterparts produce large amounts of data that must be suitably analyzed in order to yield meaningful information that should eventually lead to novel drug targets and biomarkers. Although the statistical management of DNA microarray data has been well described, there is no available report that offers a successful consolidated approach to the analysis of high-throughput protein microarray data. We describe the novel application of a statistical methodology to analyze the data from an immune response profiling assay using human protein microarray with over 5000 proteins on each chip.

Associations between Lifestyle Factors and Parkinson's Disease in an Urban Sri Lankan Clinic Study.

BACKGROUND: Associations between certain environmental and lifestyle factors and Parkinson's disease (PD) have been reported in several studies, but information on these factors and Parkinson's Disease (PD) in South Asia, is limited. OBJECTIVE: To determine associations between lifestyle factors and PD in an urban clinic-based study in Sri Lanka. METHODS: In this case-control study, demographic and lifestyle factor data (including diet, coffee/tea drinking, smoking, alcohol status) was collected from an unselected cohort of PD patients and age and gender-matched controls attending clinics in Greater Colombo, Sri Lanka. Associations between lifestyle factors and PD status were assessed using Logistic Regression analysis, while links with age of PD onset were explored with Kaplan Meier and Cox Regression survival analyses. Results with p<0.05 were considered to be statistically significant. FINDINGS: Of 229 patients with parkinsonism, 144 had Idiopathic PD using standard diagnostic criteria. Controls numbered 102. Coffee drinkers and smokers were significantly less likely to have PD (coffee, p<0.001; Odds Ratio (OR)=0.264; smoking, p=0.043; OR=0.394). Coffee drinkers were older at PD onset (p<0.001). Similar trends seen with tea drinking were not statistically significant. CONCLUSIONS: This is the first formal study of PD and these lifestyle factors in South Asia. It demonstrates an inverse association between coffee drinking, smoking and PD, and an association between coffee drinking and later age of PD onset. This is in line with other studies done worldwide, suggesting biological associations with global relevance.

Early stages of Alzheimer's disease are alarming signs in injury deaths caused by traffic accidents in elderly people (≥60 years of age): A neuropathological study.

BACKGROUND: There is little information available in the literature concerning the contribution of dementia in injury deaths in elderly people (≥60 years). AIM: This study was intended to investigate the extent of dementia-related pathologies in the brains of elderly people who died in traffic accidents or by suicide and to compare our findings with age- and sex-matched natural deaths in an elderly population. MATERIALS AND METHODS: Autopsy-derived human brain samples from nine injury death victims (5 suicide and 4 traffic accidents) and nine age- and sex-matched natural death victims were screened for neurodegenerative and cerebrovascular pathologies using histopathological and immunohistochemical techniques. For the analysis, Statistical Package for the Social Sciences (SPSS) version 16.0 was used. RESULTS: There was a greater likelihood for Alzheimer's disease (AD)-related changes in the elders who succumbed to traffic accidents (1 out of 4) compared to age- and sex-matched suicides (0 out of 5) or natural deaths (0 out of 9) as assessed by the National Institute on Aging - Alzheimer's Association guidelines. Actual burden of both neurofibrillary tangles (NFTs) and (SPs) was comparatively higher in the brains of traffic accidents, and the mean NFT counts were significantly higher in the region of entorhinal cortex (P < 0.05). However, associations obtained for other dementia-related pathologies were not statistically important. CONCLUSION: Our findings suggest that early Alzheimer stages may be a contributing factor to injury deaths caused by traffic accidents in elderly people whereas suicidal brain neuropathologies resembled natural deaths.

ABC gene-ranking for prediction of drug-induced cholestasis in rats.

As legacy toxicogenomics databases have become available, improved data mining approaches are now key to extracting and visualizing subtle relationships between toxicants and gene expression. In the present study, a novel "aggregating bundles of clusters" (ABC) procedure was applied to separate cholestatic from non-cholestatic drugs and model toxicants in the Johnson & Johnson (Janssen) rat liver toxicogenomics database [3]. Drug-induced cholestasis is an important issue, particularly when a new compound enters the market with this liability, with standard preclinical models often mispredicting this toxicity. Three well-characterized cholestasis-responsive genes (Cyp7a1, Mrp3 and Bsep) were chosen from a previous in-house Janssen gene expression signature; these three genes show differing, non-redundant responses across the 90+ paradigm compounds in our database. Using the ABC procedure, extraneous contributions were minimized in comparisons of compound gene responses. All genes were assigned weights proportional to their correlations with Cyp7a1, Mrp3 and Bsep, and a resampling technique was used to derive a stable measure of compound similarity. The compounds that were known to be associated with rat cholestasis generally had small values of this measure relative to each other but also had large values of this measure relative to non-cholestatic compounds. Visualization of the data with the ABC-derived signature showed a very tight, essentially identically behaving cluster of robust human cholestatic drugs and experimental cholestatic toxicants (ethinyl estradiol, LPS, ANIT and methylene dianiline, disulfiram, naltrexone, methapyrilene, phenacetin, alpha-methyl dopa, flutamide, the NSAIDs--indomethacin, flurbiprofen, diclofenac, flufenamic acid, sulindac, and nimesulide, butylated hydroxytoluene, piperonyl butoxide, and bromobenzene), some slightly less active compounds (3'-acetamidofluorene, amsacrine, hydralazine, tannic acid), some drugs that behaved very differently, and were distinct from both non-cholestatic and cholestatic drugs (ketoconazole, dipyridamole, cyproheptadine and aniline), and many postulated human cholestatic drugs that in rat showed no evidence of cholestasis (chlorpromazine, erythromycin, niacin, captopril, dapsone, rifampicin, glibenclamide, simvastatin, furosemide, tamoxifen, and sulfamethoxazole). Most of these latter drugs were noted previously by other groups as showing cholestasis only in humans. The results of this work suggest that the ABC procedure and similar statistical approaches can be instrumental in combining data to compare toxicants across toxicogenomics databases, extract similarities among responses and reduce unexplained data varation.

Cytoskeletal Pathologies of Age-Related Diseases between Elderly Sri Lankan (Colombo) and Indian (Bangalore) Brain Samples.

Within South Asia, Sri Lanka represents fastest aging with 13% of the population was aged over 60's in 2011, whereas in India it was 8%. Majority of the Sri Lankan population based genetic studies have confirmed their origin on Indian mainland. As there were inadequate data on aging cytoskeletal pathologies of these two nations with their close genetic affiliations, we performed a comparison on their elderly. Autopsy brain samples of 50 individuals from Colombo, Sri Lanka (mean age 72.1 yrs ± 7.8, mean ± S.D.) and 42 individuals from Bangalore, India (mean age 65.9 yrs ± 9.3) were screened for neurodegenerative pathologies using immunohistochemical techniques. A total of 79 cases with incomplete clinical history (Colombo- 47 and Bangalore- 32) were subjected to statistical analysis and 13 cases, clinically diagnosed with dementia and/or Parkinsonism disorders were excluded. As per National Institute on Aging- Alzheimer's Association guidelines, between Colombo and Bangalore samples, Alzheimer's disease neuropathologic change for intermediate/ high level was 4.25% vs. 3.12% and low level was 19.15% vs. 15.62% respectively. Pathologies associated with Parkinsonism including brainstem predominant Lewy bodies- 6.4% and probable progressive supra nuclear palsy- 2.13% were found solely in Colombo samples. Alzheimer related pathologies were not different among elders, however, in Colombo males, neurofibrillary tangle grade was significantly higher in the region of hippocampus (odds ratio = 1.46, 95% confidence interval = 0.07-0.7) and at risk in midbrain substantia nigra (p = 0.075). Other age-related pathologies including spongiform changes (p < 0.05) and hippocampus cell loss in dentate gyrus region (p < 0.05) were also identified prominently in Colombo samples. Taken together, aging cytoskeletal pathologies are comparatively higher in elderly Sri Lankans and this might be due to their genetic, dietary and/ or environmental variations.

Vascular Contributions in Alzheimer's Disease-Related Neuropathological Changes: First Autopsy Evidence from a South Asian Aging Population.

BACKGROUND: Evidence from various consortia on vascular contributions has been inconsistent in determining the etiology of sporadic Alzheimer's disease (AD). OBJECTIVE: To investigate vascular risk factors and cerebrovascular pathologies associated in manifestation of AD-related neuropathological changes of an elderly population. METHODS: Postmortem brain samples from 76 elderly subjects (≥50 years) were used to study genetic polymorphisms, intracranial atherosclerosis of the circle of Willis (IASCW), and microscopic infarcts in deep white matters. From this cohort, 50 brains (≥60 years) were subjected to neuropathological diagnosis using immunohistopathological techniques. RESULTS: Besides the association with age, the apolipoprotein E ɛ4 allele was significantly and strongly associated with Thal amyloid-ß phases ≥1 [odds ratio (OR) = 6.76, 95% confidence interval (CI) 1.37-33.45] and inversely with Braak neurofibrillary tangle (NFT) stages ≥III (0.02, 0.0-0.47). Illiterates showed a significant positive association for Braak NFT stages ≥IV (14.62, 1.21-176.73) and a significant negative association for microscopic infarcts (0.15, 0.03-0.71) in deep white matters. With respect to cerebrovascular pathologies, cerebral small vessel lesions (white matter hyperintensities and cerebral amyloid angiopathy) showed a higher degree of associations among them and with AD-related neuropathological changes (p < 0.05) compared to large vessel pathology (IASCW), which showed a significant association only with Braak NFT stages ≥I (p = 0.050). CONCLUSION: These findings suggest that besides age, education, and genetic factors, other vascular risk factors were not associated with AD-related neuropathological changes and urge prompt actions be taken against cerebral small vessel diseases since evidence for effective prevention is still lacking.

Resampling-based similarity measures for high-dimensional data.

An important issue in classification is the assessment of sample similarity. This is nontrivial in high-dimensional or megavariate datasets--datasets that are comprised of simultaneous measurements on thousands of features, many of which carry little or no information regarding consistent sample differences. Conventional similarity measures do not work particularly well for such data. As an alternative, we propose a distance measure that is based on a refiltering process: at each step of the process a random subset of features is selected and a cluster analysis is performed using only this subset; the relative frequency with which a pair of samples clusters together across several such random subsets forms the similarity measure. The features chosen at any step may be completely random or enriched by awarding the more informative features a higher chance of selection; this enrichment turns out to be particularly effective. We use actual datasets from the burgeoning genomics literature to demonstrate the superior performance of this similarity measure, especially the enriched form of the similarity measure, compared to more conventional measures such as Euclidean distance or correlation, or, if the data are categorical, Hamming distance.