Adaptive model selection using empirical complexities

Given $n$ independent replicates of a jointly distributed pair $(X,Y)\in {\cal R}^d \times {\cal R}$, we wish to select from a fixed sequence of model classes ${\cal F}_1, {\cal F}_2, \ldots$ a deterministic prediction rule $f: {\cal R}^d \to {\cal R}$ whose risk is small. We investigate the possibility of empirically assessingthe {\em complexity} of each model class, that is, the actual difficulty of the estimation problem within each class. The estimated complexities are in turn used to define an adaptive model selection procedure, which is based on complexity penalized empirical risk.The available data are divided into two parts. The first is used to form an empirical cover of each model class, and the second is used to select a candidate rule from each cover based on empirical risk. The covering radii are determined empirically to optimize a tight upper bound on the estimation error. An estimate is chosen from the list of candidates in order to minimize the sum of class complexity and empirical risk. A distinguishing feature of the approach is that the complexity of each model class is assessed empirically, based on the size of its empirical cover.Finite sample performance bounds are established for the estimates, and these bounds are applied to several non-parametric estimation problems. The estimates are shown to achieve a favorable tradeoff between approximation and estimation error, and to perform as well as if the distribution-dependent complexities of the model classes were known beforehand. In addition, it is shown that the estimate can be consistent,and even possess near optimal rates of convergence, when each model class has an infinite VC or pseudo dimension.For regression estimation with squared loss we modify our estimate to achieve a faster rate of convergence.

Obesity-related phenotypes are associated to parental longevity in a Swiss population-based study

PURPOSE. Longevity has been attributed to decreased cardiovascular mortality. Subjects with long-lived parents may represent a valuable group to study cardiovascular risk factors (CVRF) associated with longevity, possibly leading to new ways of preventing cardiovascular disease. Purpose: Longevity has been attributed to decreased cardiovascular mortality. Subjects with long-lived parents may represent a valuable group to study cardiovascular risk factors (CVRF) associated with longevity, possibly leading to new ways of preventing cardiovascular disease. Methods: We analyzed data from a population-based sample of 2561 participants (1163 men and 1398 women) aged 55--75 years from the city of Lausanne, Switzerland (CoLaus study). Participants were stratified by the number of parents (0, 1, 2) who survived to 85 years or more. Trend across these strata was assessed using a non-parametric kmean test. The associations of parental age (independent covariate used as a proxy for longevity) with fasting blood glucose, blood pressures, blood lipids, body mass index (BMI), weight, height or liver enzymes (continuous dependent variables) were analyzed using multiple linear regressions. Models were adjusted for age, sex, alcohol consumption, smoking and educational level, and BMI for liver enzymes. Results: For subjects with 0 (N=1298), 1 (N=991) and 2 (N=272) long-lived parents, median BMI (interquartile range) was 25.4 (6.5), 24.9 (6.1) and 23.7 (4.8) kg/m2 in women (P<0.001), and 27.3 (4.8), 27.0 (4.5) and 25.9 (4.9) kg/m2 in men (P=0.04), respectively; median weight was 66.5 (16.1), 65.0 (16.4) and 63.4 (13.7) kg in women (P=0.003), and 81.5 (17.0), 81.4 (16.4) and 80.3 (17.1) kg in men (P=0.36). Median height was 161 (8), 162 (9) and 163 (8) cm in women (P=0.005), and 173 (9), 174 (9) and 174 (11) cm in men (P=0.09). The corresponding medians for AST (Aspartate Aminotransferase) were 31 (13), 29 (11) and 28 (10) U/L (P=0.002), and 28 (17), 27 (14) and 26 (19) U/L for ALT (Alanin Aminotransferase, P=0.053) in men. In multivariable analyses, greater parental longevity was associated with lower BMI, lower weight and taller stature in women (P<0.01) and lower AST in men (P=0.011). No significant associations were observed for the other variables analyzed. Sensitivity analyses restricted to subjects whose parents were dead (N=1844) led to similar results, with even stronger associations of parental longevity with liver enzymes in men. Conclusion: In women, increased parental longevity was associated with smaller BMI, attributable to lower weight and taller stature. In men, the association of increased parental longevity with lower liver enzymes, independently of BMI, suggests that parental longevity may be associated with decreased nonalcoholic fatty liver disease.

Examining non-response bias in substance use research-Are late respondents proxies for non-respondents?

BACKGROUND: Non-response is a major concern among substance use epidemiologists. When differences exist between respondents and non-respondents, survey estimates may be biased. Therefore, researchers have developed time-consuming strategies to convert non-respondents to respondents. The present study examines whether late respondents (converted former non-participants) differ from early respondents, non-consenters or silent refusers (consent givers but non-participants) in a cohort study, and whether non-response bias can be reduced by converting former non-respondents. METHODS: 6099 French- and 5720 German-speaking Swiss 20-year-old males (more than 94% of the source population) completed a short questionnaire on substance use outcomes and socio-demographics, independent of any further participation in a cohort study. Early respondents were those participating in the cohort study after standard recruitment procedures. Late respondents were non-respondents that were converted through individual encouraging telephone contact. Early respondents, non-consenters and silent refusers were compared to late respondents using logistic regressions. Relative non-response biases for early respondents only, for respondents only (early and late) and for consenters (respondents and silent refusers) were also computed. RESULTS: Late respondents showed generally higher patterns of substance use than did early respondents, but lower patterns than did non-consenters and silent refusers. Converting initial non-respondents to respondents reduced the non-response bias, which might be further reduced if silent refusers were converted to respondents. CONCLUSION: Efforts to convert refusers are effective in reducing non-response bias. However, converted late respondents cannot be seen as proxies of non-respondents, and are at best only indicative of existing response bias due to persistent non-respondents.

Neurally adjusted ventilatory assist can improve arterial oxygenation

INTRODUCTION. Neurally Adjusted Ventilatory Assist (NAVA) [1] is a new spontaneousassisted ventilatory mode which uses the diaphragmatic electrical activity (Eadi) to pilot the ventilator. Eadi is used to initiate the ventilator's pressurization and cycling off. Delivered inspiratory assistance is proportional to Eadi. NAVA can improve patient-ventilator synchrony [2] compared to pressure support (PS), but little is known about its effect on minute ventilation and oxygenation. OBJECTIVES. To compare the effects of NAVA and PS on minute ventilation and oxygenation and to analyze potential determinant factors for oxygenation. METHODS. Comparison between two 20-min periods under NAVA and PS. NAVA gain (proportionality factor between Eadi and delivered pressure) set as to obtain the same peak pressure as in PS. FIO2 and positive end-expiratory pressure (PEEP) were the same in NAVA and PS. Blood gas analyses were performed at the end of both recording periods. Statistical analysis: groups were compared with paired t tests or non parametric Wilcoxon signed-rank tests. p\0.05 was considered significant. RESULTS. [Mean ± SD]: 22 patients (age 66 ± 12 year, 7 M/15F, BMI 23.4 ± 3.1 kg/m2), 8 patients with COPD. Initial settings: PS 13 ± 3 cmH2O, PEEP 7 ± 2 cmH2O, NAVA gain 2.2 ± 1.8. Minute ventilation and PaCO2 were the same with both modes (p = 0.296 and 0.848, respectively). Tidal volume was lower with NAVA (427 ± 102 vs. 477 ± 102 ml, p\0.001). In contrast respiratory rate was higher with NAVA (25.6 ± 9.5 vs. 22.3 ± 8.9 cycles/min). Arterial oxygenation was improved with NAVA (PaO2 85.1 ± 28.9 vs. 75.8 ± 11.9 mmHg, p = 0.017, PaO2/FIO2 210 ± 53 vs. 195 ± 58 mmHg, p = 0.019). Neural inspiratory time (Tin) was comparable between NAVA and PS (p = 0.566). Among potential determinant factors for oxygenation, mean airway pressure (Pmean) was lower with NAVA (10.6 ± 2.6 vs. 11.1 ± 2.4 cmH2O, p = 0.006), as was the pressure time product (PTP) (6.8 ± 3.0 vs. 9.2 ± 3.5 cmH2O 9 s, p = 0.004). There were less asynchrony events with NAVA (2.3 ± 2.0 vs. 4.4 ± 3.8, p = 0.009).Tidal volume variability was higher with NAVA (variation coefficient: 30 ± 19.5 vs. 13.5 ± 8.6, p\0.001). Inspiratory time in excess (Tiex) was lower with NAVA (56 ± 23 vs. 202 ± 200 ms, p = 0.001). CONCLUSION. Despite lower Pmean and PTP in NAVA, arterial oxygenation was improved compared to PS. As asynchronies may be associated with an increased work of breathing and a higher oxygen consumption, their decrease in number with NAVA could be an explanation for oxygenation improvement. Another explanation could be the increase in VT variability. Further studies should now be performed to confirm the potential of NAVA in improving arterial oxygenation and explore the underlying mechanisms.

Effect of once-yearly zoledronic acid on the spine and hip as measured by quantitative computed tomography: results of the HORIZON Pivotal Fracture Trial.

Changes in bone mineral density and bone strength following treatment with zoledronic acid (ZOL) were measured by quantitative computed analysis (QCT) or dual-energy X-ray absorptiometry (DXA). ZOL treatment increased spine and hip BMD vs placebo, assessed by QCT and DXA. Changes in trabecular bone resulted in increased bone strength. INTRODUCTION: To investigate bone mineral density (BMD) changes in trabecular and cortical bone, estimated by quantitative computed analysis (QCT) or dual-energy X-ray absorptiometry (DXA), and whether zoledronic acid 5 mg (ZOL) affects bone strength. METHODS: In 233 women from a randomized, controlled trial of once-yearly ZOL, lumbar spine, total hip, femoral neck, and trochanter were assessed by DXA and QCT (baseline, Month 36). Mean percentage changes from baseline and between-treatment differences (ZOL vs placebo, t-test) were evaluated. RESULTS: Mean between-treatment differences for lumbar spine BMD were significant by DXA (7.0%, p < 0.01) and QCT (5.7%, p < 0.0001). Between-treatment differences were significant for trabecular spine (p = 0.0017) [non-parametric test], trabecular trochanter (10.7%, p < 0.0001), total hip (10.8%, p < 0.0001), and compressive strength indices at femoral neck (8.6%, p = 0.0001), and trochanter (14.1%, p < 0.0001). CONCLUSIONS: Once-yearly ZOL increased hip and spine BMD vs placebo, assessed by QCT vs DXA. Changes in trabecular bone resulted in increased indices of compressive strength.

Pericardial fat volume correlates with coronary vasomotion

Purpose: Recent studies showed that pericardial fat was independently correlated with the development of coronary artery disease (CAD). The mechanism remains unclear. We aimed at assessing a possible relationship between pericardial fat volume and endothelium-dependent coronary vasomotion, a surrogate of future cardiovascular events.Methods: Fifty healthy volunteers without known CAD or cardiovascular risk factors (CRF) were enrolled. They all underwent a dynamic Rb- 82 cardiac PET/CT to quantify myocardial blood flow (MBF) at rest, during MBF response to cold pressure test (CPT-MBF) and adenosine stress. Pericardial fat volume (PFV) was measured using a 3D volumetric CT method and common biological CRF (glucose and insulin levels, HOMA-IR, cholesterol, triglyceride, hs-CRP). Relationships between MBF response to CPT, PFV and other CRF were assessed using non-parametric Spearman correlation and multivariate regression analysis of variables with significant correlation on univariate analysis (Stata 11.0).Results: All of the 50 participants had normal MBF response to adenosine (2.7±0.6 mL/min/g; 95%CI: 2.6−2.9) and myocardial flow reserve (2.8±0.8; 95%CI: 2.6−3.0) excluding underlying CAD. Simple regression analysis revealed a significant correlation between absolute CPTMBF and triglyceride level (rho = −0.32, p = 0.024) fasting blood insulin (rho = −0.43, p = 0.0024), HOMA-IR (rho = −0.39, p = 0.007) and PFV (rho = −0.52, p = 0.0001). MBF response to adenosine was only correlated with PFV (rho = −0.32, p = 0.026). On multivariate regression analysis PFV emerged as the only significant predictor of MBF response to CPT (p = 0.002).Conclusion: PFV is significantly correlated with endothelium-dependent coronary vasomotion. High PF burden might negatively influence MBF response to CPT, as well as to adenosine stress, even in persons with normal hyperemic myocardial perfusion imaging, suggesting a link between PF and future cardiovascular events. While outside-to-inside adipokines secretion through the arterial wall has been described, our results might suggest an effect upon NO-dependent and -independent vasodilatation. Further studies are needed to elucidate this mechanism.

Benchmarking therapeutic drug monitoring software: A systematic evaluation of available computer tools

Objectives: Therapeutic drug monitoring (TDM) aims at optimizing treatment by individualizing dosage regimen based on blood concentrations measurement. Maintaining concentrations within a target range requires pharmacokinetic (PK) and clinical capabilities. Bayesian calculation represents a gold standard in TDM approach but requires computing assistance. The aim of this benchmarking was to assess and compare computer tools designed to support TDM clinical activities.¦Methods: Literature and Internet were searched to identify software. Each program was scored against a standardized grid covering pharmacokinetic relevance, user-friendliness, computing aspects, interfacing, and storage. A weighting factor was applied to each criterion of the grid to consider its relative importance. To assess the robustness of the software, six representative clinical vignettes were also processed through all of them.¦Results: 12 software tools were identified, tested and ranked. It represents a comprehensive review of the available software characteristics. Numbers of drugs handled vary from 2 to more than 180, and integration of different population types is available for some programs. Nevertheless, 8 programs offer the ability to add new drug models based on population PK data. 10 computer tools incorporate Bayesian computation to predict dosage regimen (individual parameters are calculated based on population PK models). All of them are able to compute Bayesian a posteriori dosage adaptation based on a blood concentration while 9 are also able to suggest a priori dosage regimen, only based on individual patient covariates. Among those applying Bayesian analysis, MM-USC*PACK uses a non-parametric approach. The top 2 programs emerging from this benchmark are MwPharm and TCIWorks. Others programs evaluated have also a good potential but are less sophisticated or less user-friendly.¦Conclusions: Whereas 2 software packages are ranked at the top of the list, such complex tools would possibly not fit all institutions, and each program must be regarded with respect to individual needs of hospitals or clinicians. Programs should be easy and fast for routine activities, including for non-experienced users. Although interest in TDM tools is growing and efforts were put into it in the last years, there is still room for improvement, especially in terms of institutional information system interfacing, user-friendliness, capability of data storage and automated report generation.

Intérêt de l'échographie du diamètre de l'enveloppe du nerf optique pour la détection non invasive de l'hypertension intracrânienne [Interest of optic nerve sheath diameter ultrasonography in dectecting non-invasively raised intracranial pressure].

Intracranial hypertension is an emergency suspected from clinical symptoms, imaging data and ophthalomologic signs. Intracranial hypertension is confirmed by invasive intracranial monitoring, which is the gold standard technique to measure intracranial pressure (ICP). Because of complications, hemorrhage or infection, non-invasive methods have been developed such as neuroimaging, transcranial Doppler sonography and optic nerve sheath diameter (ONSD) ultrasonography. We have reviewed ONSD technique that detects intracranial hypertension related volume variations of subarachnoid space along the retro bulbar segment of the optic nerve. Technique, indications and prospects are discussed.

An active contour-based atlas registration model applied to automatic subthalamic nucleus targeting on MRI: method and validation.

This paper presents a new non parametric atlas registration framework, derived from the optical flow model and the active contour theory, applied to automatic subthalamic nucleus (STN) targeting in deep brain stimulation (DBS) surgery. In a previous work, we demonstrated that the STN position can be predicted based on the position of surrounding visible structures, namely the lateral and third ventricles. A STN targeting process can thus be obtained by registering these structures of interest between a brain atlas and the patient image. Here we aim to improve the results of the state of the art targeting methods and at the same time to reduce the computational time. Our simultaneous segmentation and registration model shows mean STN localization errors statistically similar to the most performing registration algorithms tested so far and to the targeting expert's variability. Moreover, the computational time of our registration method is much lower, which is a worthwhile improvement from a clinical point of view.

A multicentre, retrospective case-control study assessing the role of trabecular bone score (TBS) in menopausal Caucasian women with low areal bone mineral density (BMDa): Analysing the odds of vertebral fracture.

INTRODUCTION: The trabecular bone score (TBS) is a new parameter that is determined from grey level analysis of DXA images. It relies on the mean thickness and volume fraction of trabecular bone microarchitecture. This was a preliminary case-control study to evaluate the potential diagnostic value of TBS, both alone and combined with bone mineral density (BMDa), in the assessment of vertebral fracture. METHODS: Out of a subject pool of 441 Caucasian, postmenopausal women between the ages of 50 and 80 years, we identified 42 women with osteoporosis-related vertebral fractures, and compared them with 126 age-matched women without any fractures (1 case: 3 controls). Primary outcomes were BMDa and TBS. Inter-group comparisons were undertaken using Student's t-tests and Wilcoxon signed ranks tests for parametric and non-parametric data, respectively. Odds ratios for vertebral fracture were calculated for each incremental one standard deviation decrease in BMDa and TBS, and areas under the receiver operating curve (AUC) calculated and sensitivity analysis were conducted to compare BMDa alone, TBS alone, and the combination of BMDa and TBS. Subgroup analyses were performed specifically for women with osteopenia, and for women with T-score-defined osteoporosis. RESULTS: Across all subjects (n=42, 126) weight and body mass index were greater and BMDa and TBS both less in women with fractures. The odds of vertebral fracture were 3.20 (95% CI, 2.01-5.08) for each incremental decrease in TBS, 1.95 (1.34-2.84) for BMDa, and 3.62 (2.32-5.65) for BMDa + TBS combined. The AUC was greater for TBS than for BMDa (0.746 vs. 0.662, p=0.011). At iso-specificity (61.9%) or iso-sensitivity (61.9%) for both BMDa and TBS, TBS + BMDa sensitivity or specificity was 19.1% or 16.7% greater than for either BMDa or TBS alone. Among subjects with osteoporosis (n=11, 40) both BMDa (p=0.0008) and TBS (p=0.0001) were lower in subjects with fractures, and both OR and AUC (p=0.013) for BMDa + TBS were greater than for BMDa alone (OR=4.04 [2.35-6.92] vs. 2.43 [1.49-3.95]; AUC=0.835 [0.755-0.897] vs. 0.718 [0.627-0.797], p=0.013). Among subjects with osteopenia, TBS was lower in women with fractures (p=0.0296), but BMDa was not (p=0.75). Similarly, the OR for TBS was statistically greater than 1.00 (2.82, 1.27-6.26), but not for BMDa (1.12, 0.56-2.22), as was the AUC (p=0.035), but there was no statistical difference in specificity (p=0.357) or sensitivity (p=0.678). CONCLUSIONS: The trabecular bone score warrants further study as to whether it has any clinical application in osteoporosis detection and the evaluation of fracture risk.

Is it appropriate to index glomerular filtration rate for body surface area

Purpose: Obesity is an established independent risk factor for chronic kidney disease. Thus, measurement of glomerular filtration rate (GFR) is important in this population. Traditionally, GFR has been indexed for body surface area (BSA), but this indexation may not be appropriate in obese individuals. Therefore, the objective of the study was to compare absolute GFR with GFR indexed for BSA and with GFR indexed for height. Methods and materials: The study was conducted in 66 families from the Seychelles islands that included several members with hypertension. GFR and effective renal plasma flow (ERPF) were measured using inulin and PAH clearances, respectively. Antihypertensive treatment, if used, was withheld 2 weeks before conducting the clearances. Participants with diabetes mellitus were excluded from the analysis. BSA was calculated using the Dubois formula. We assessed trend across BMI categories using a non parametric test. Results: Participants included 174 women and 127 men. The prevalence of hypertension was 61%, of which 68% were treated. The table shows that absolute GFR, GFR indexed for height, ERPF, filtration fraction were significantly higher across BMI categories. When GFR was indexed for BSA, the association between GFR and BMI categories was lost. Conclusion: Indexing GFR for BSA in overweight and obese individuals leads to a substantial underestimation of GFR. Filtration fraction, which does not depend on BSA, is higher in obese individuals, which suggests glomerular hyperfiltration. Indexing GFR for BSA therefore would mask the underlying glomerular hyperfiltration. As the number of nephrons does not increase with weight gain, absolute GFR represents a better marker of single nephron GFR and is more appropriate.

Parametric Approach to Blind Deconvolution of Nonlinear Channels

A parametric procedure for the blind inversion of nonlinear channels is proposed, based on a recent method of blind source separation in nonlinear mixtures. Experiments show that the proposed algorithms perform efficiently, even in the presence of hard distortion. The method, based on the minimization of the output mutual information, needs the knowledge of log-derivative of input distribution (the so-called score function). Each algorithm consists of three adaptive blocks: one devoted to adaptive estimation of the score function, and two other blocks estimating the inverses of the linear and nonlinear parts of the channel, (quasi-)optimally adapted using the estimated score functions. This paper is mainly concerned by the nonlinear part, for which we propose two parametric models, the first based on a polynomial model and the second on a neural network, while [14, 15] proposed non-parametric approaches.

Frequency and chronology of extraintestinal manifestations in the Swiss Inflammatory Bowel Disease Cohort

Background: Data on the frequency of extraintestinal manifestations (EIM) in inflammatory bowel disease (IBD) is scarce, especially the one evaluating the time of occurrence of the EIM relative to IBD diagnosis. Aim: To assess the type and frequency of EIM in IBD patients and to evaluate when EIMs occur relative to IBD diagnosis. Methods: Analysis of data from the Swiss Inflammatory Bowel Disease Cohort (SIBDCS) which collects data on a large sample of IBD patients from hospitals and private practices across Switzerland starting in 2005. While parametric data are shown as mean ± SD, non-parametric data are presented as median and interquartile range (IQR). Results: A total of 1143 patients were analyzed (572 (50%) female, mean age 42.1 ± 14.4 years): 629 (55%) with Crohn's disease (CD), 501 (44%) with ulcerative colitis (UC), and 13 (1%) with indeterminate colitis (IC). Of 1143 patients, 374 (32.7%) presented with EIM (65% with CD, 33% with UC, 2% with IC). Of those patients suffering from EIMs, 37.4% presented with one, 41.7% with two, 12.4% with three, 5.3% with four, and 3.2% with five EIM during their lifetime. The IBD patients initially presented with the following EIMs: peripheral arthritis (PA) 63.4%, ankylosing spondylitis (AS) 8.1%, primary sclerosing cholangitis (PSC) 6.0%, uveitis 5.7%, oral aphthosis 5.7%, erythema nodosum (EN) 5.0%, pyoderma gangrenosum 1.8%, psoriasis 0.7%. While 92.9% of EIM occurred once IBD diagnosis was established (median 72 months, IQR 9-147 months, p < 0.001), 7.1% of EIMs preceded IBD diagnosis (median time 28 months before IBD diagnosis, IQR 7-60 months). Over a course of a lifetime, IBD patients presented with the following EIM (total exceeds 100 due to potential presence of multiple EIM): peripheral arthritis 69.3%, oral aphthosis 23%, ankylosing spondylitis 19.4%, uveitis 15.5%, erythema nodosum 14.5%, PSC 7.8%, pyoderma gangrenosum 6%, psoriasis 2.8%. Conclusion: EIMs frequently occur in a lifetime of IBD patients. The vast majority of patients present with EIMs once IBD diagnosis has been established. IBD patients most often present with peripheral arthritis, ankylosing spondylitis and PSC as their first EIM. However, peripheral arthritis, oral aphthosis, and ankylosing spondylitis are the most common EIMs in a lifetime of IBD patients.

Prevalence of cognitive disorders in HIV plus patients with long-term suppression of viremia

Background: HAART has contributed to decrease the HIV-related mortality and morbidity. However, the prevalence of HIV-associated neurocognitive disorders (HAND) seems to have increased. The aim of this study was to determine the prevalence of cognitive complaint and of HAND in a cohort of aviremic HIV_patients in the South-western part of Switzerland. Design/Methods: Two hundred HIV_ patients who had (1) undetectable HIV RNA concentrations in the plasma for_3 months, (2) no history of major opportunistic infection of the CNS in the past three years, (3) no current use of IV drugs and (4) no signs of major depression according to the DSM-IV criteria, answered a questionnaire designed to elicit cognitive complaints. Cognitive functions of a subset of HIV_ patients with or without cognitive complaints were assessed using the HIV Dementia scale (HDS) and a battery of neuropsychological tests evaluating the sub-cortical functions. Cognitive impairment was defined according to the revised diagnostic criteria for HAND. Non-parametric tests were used for statistics and a Bonferroni corrected standard p level of pB0.002 was applied for multiple comparisons. Results: The prevalence of cognitive complaints was 27% (54 patients) among the 200 questioned patients. At the time of writing this abstract, cognitive functions of 50 complaining and 28 noncomplaining aviremic patients had been assessed with the HDS and the full neuropsychological battery. The prevalence of HAND producing at least mild interference in daily functioning (mild neurocognitive disorders [MND] or HIV-associated dementia [HAD]) was 44% (34/78 patients) in the group who underwent neuropsychological testing. Objective evidences of HAND were more frequent in complaining than in non-complaining patients (pB0.001). Using a ROC curve, a cut-off of 13 on the HDS was found to have a sensitivity of 74% and a specificity of 71% (p_0.001) for the diagnosis of HAND. A trend for lower CNS Penetrating-Effectiveness scores for HAART in patients with MND or HAD as compared to the others was present (1.59 0.6 vs. 1.990.6; p_0.006 [Bonferroni correction]). Conclusions/Relevance: So far, our results suggest that (1) the prevalence of HAND is high in HIV_ patients with a long-term suppression of viremia, and (2) cognitive complaints expressed by aviremic HIV_ patients should be carefully investigated as they correlate with objective evidences of cognitive decline in a neuropsychological testing. HAART with a high CNS penetrating-effectiveness may contribute to prevent HAND. Funding: Swiss HIV Cohort Study.

Alternative tests for correct specification of conditional predictive densities

We propose new methods for evaluating predictive densities that focus on the models' actual predictive ability in finite samples. The tests offer a simple way of evaluatingthe correct specification of predictive densities, either parametric or non-parametric.The results indicate that our tests are well sized and have good power in detecting mis-specification in predictive densities. An empirical application to the Survey ofProfessional Forecasters and a baseline Dynamic Stochastic General Equilibrium modelshows the usefulness of our methodology.