Concordance Probability and Discriminatory Power in Proportional Hazards Regression

The concordance probability is used to evaluate the discriminatory power and the predictive accuracy of nonlinear statistical models. We derive an analytic expression for the concordance probability in the Cox proportional hazards model. The proposed estimator is a function of the regression parameters and the covariate distribution only and does not use the observed event and censoring times. For this reason it is asymptotically unbiased, unlike Harrell's c-index based on informative pairs. The asymptotic distribution of the concordance probability estimate is derived using U-statistic theory and the methodology is applied to a predictive model in lung cancer.

Non-classical karyotypic features in relapsed childhood B-cell precursor acute lymphoblastic leukemia

Karyotype analysis of acute lymphoblastic leukemia (ALL) at diagnosis has provided valuable prognostic markers for treatment stratification. However, reports of cytogenetic studies of relapsed ALL samples are limited. We compared the karyotypes from 436 nonselected B-cell precursor ALL patients at initial diagnosis and of 76 patients at first relapse. We noticed a relative increase of karyotypes that did not fall into the classic ALL cytogenetic subgroups (high hyperdiploidy, t(12;21), t(9;22), 11q23, t(1;19), <45 chromosomes) in a group of 29 patients at relapse (38%) compared to 130 patients at presentation (30%). Non-classical cytogenetic aberrations in these 29 patients were mostly found on chromosomes 1, 2, 7, 9, 13, 14, and 17. We also describe six rare reciprocal translocations, three of which involved 14q32. The most frequent abnormalities were found in 9p (12/29 cases) and were associated with a marked decrease in the duration of the second remission, but not of the probability of 10-year event-free survival after relapse treatment. From 29 patients with non-classical cytogenetic aberrations, only 8 (28%) had been stratified to a high risk-arm on the first treatment protocol, suggesting that this subgroup might benefit from the identification of new prognostic markers in future studies.

STATISTICAL METHODS FOR RARE AND COMMON VARIANT ASSOCIATION STUDIES

Complex human diseases are a major challenge for biological research. The goal of my research is to develop effective methods for biostatistics in order to create more opportunities for the prevention and cure of human diseases. This dissertation proposes statistical technologies that have the ability of being adapted to sequencing data in family-based designs, and that account for joint effects as well as gene-gene and gene-environment interactions in the GWA studies. The framework includes statistical methods for rare and common variant association studies. Although next-generation DNA sequencing technologies have made rare variant association studies feasible, the development of powerful statistical methods for rare variant association studies is still underway. Chapter 2 demonstrates two adaptive weighting methods for rare variant association studies based on family data for quantitative traits. The results show that both proposed methods are robust to population stratification, robust to the direction and magnitude of the effects of causal variants, and more powerful than the methods using weights suggested by Madsen and Browning [2009]. In Chapter 3, I extended the previously proposed test for Testing the effect of an Optimally Weighted combination of variants (TOW) [Sha et al., 2012] for unrelated individuals to TOW &ndash F, TOW for Family &ndash based design. Simulation results show that TOW &ndash F can control for population stratification in wide range of population structures including spatially structured populations, is robust to the directions of effect of causal variants, and is relatively robust to percentage of neutral variants. In GWA studies, this dissertation consists of a two &ndash locus joint effect analysis and a two-stage approach accounting for gene &ndash gene and gene &ndash environment interaction. Chapter 4 proposes a novel two &ndash stage approach, which is promising to identify joint effects, especially for monotonic models. The proposed approach outperforms a single &ndash marker method and a regular two &ndash stage analysis based on the two &ndash locus genotypic test. In Chapter 5, I proposed a gene &ndash based two &ndash stage approach to identify gene &ndash gene and gene &ndash environment interactions in GWA studies which can include rare variants. The two &ndash stage approach is applied to the GAW 17 dataset to identify the interaction between KDR gene and smoking status.

Auditory event-related potentials, bispectral index, and entropy for the discrimination of different levels of sedation in intensive care unit patients

BACKGROUND: Sedation protocols, including the use of sedation scales and regular sedation stops, help to reduce the length of mechanical ventilation and intensive care unit stay. Because clinical assessment of depth of sedation is labor-intensive, performed only intermittently, and interferes with sedation and sleep, processed electrophysiological signals from the brain have gained interest as surrogates. We hypothesized that auditory event-related potentials (ERPs), Bispectral Index (BIS), and Entropy can discriminate among clinically relevant sedation levels. METHODS: We studied 10 patients after elective thoracic or abdominal surgery with general anesthesia. Electroencephalogram, BIS, state entropy (SE), response entropy (RE), and ERPs were recorded immediately after surgery in the intensive care unit at Richmond Agitation-Sedation Scale (RASS) scores of -5 (very deep sedation), -4 (deep sedation), -3 to -1 (moderate sedation), and 0 (awake) during decreasing target-controlled sedation with propofol and remifentanil. Reference measurements for baseline levels were performed before or several days after the operation. RESULTS: At baseline, RASS -5, RASS -4, RASS -3 to -1, and RASS 0, BIS was 94 [4] (median, IQR), 47 [15], 68 [9], 75 [10], and 88 [6]; SE was 87 [3], 46 [10], 60 [22], 74 [21], and 87 [5]; and RE was 97 [4], 48 [9], 71 [25], 81 [18], and 96 [3], respectively (all P < 0.05, Friedman Test). Both BIS and Entropy had high variabilities. When ERP N100 amplitudes were considered alone, ERPs did not differ significantly among sedation levels. Nevertheless, discriminant ERP analysis including two parameters of principal component analysis revealed a prediction probability PK value of 0.89 for differentiating deep sedation, moderate sedation, and awake state. The corresponding PK for RE, SE, and BIS was 0.88, 0.89, and 0.85, respectively. CONCLUSIONS: Neither ERPs nor BIS or Entropy can replace clinical sedation assessment with standard scoring systems. Discrimination among very deep, deep to moderate, and no sedation after general anesthesia can be provided by ERPs and processed electroencephalograms, with similar P(K)s. The high inter- and intraindividual variability of Entropy and BIS precludes defining a target range of values to predict the sedation level in critically ill patients using these parameters. The variability of ERPs is unknown.

Event-related potential and EEG measures in Parkinson's disease without and with dementia

Nondemented Parkinson’s disease (PD) patients showed increased amplitude of event-related potential component P3. We recorded 18-channel spontaneous eyes-closed resting EEG and auditory oddball event-related potentials in 29 PD patients and 11 age-matched controls. Combining Mini-Mental State Examination score and oddball P3 counting performance, 15 patients were intellectually normal, 7 moderately, and 7 severely demented. P3 and N1 amplitude and latency, mean amplitude of 1,024 ms post-stimulus (separate after rare and after frequent stimuli), and resting EEG total power for 40 s were computed, and linearly regressed for age, sex, and L-dopa dosage. In nondemented PD patients, increased P3 amplitude was confirmed, but N1 amplitude and mean amplitude after rare and frequent stimuli were also increased as well as – most important – resting EEG total power. With increasing dementia, amplitude and power decreased, and P3 latency increased. Task demands cannot explain increased P3 amplitude, since similarly increased EEG total power was found during no-task resting. Prospective studies must determine whether P3 amplitude and EEG power in nondemented PD patients can serve as predictors of dementia.

Fixed and Flexible: Coexistence of Obligate and Facultative Migratory Strategies in a Freshwater Fish

Migration is an important event in many animal life histories, but the degree to which individual animals participate in seasonal migrations often varies within populations. The powerful ecological and evolutionary consequences of such partial migration are now well documented, but the underlying mechanisms are still heavily debated. One potential mechanism of partial migration is between-individual variation in body condition, where animals in poor condition cannot pay the costs of migration and hence adopt a resident strategy. However, underlying intrinsic traits may overrule such environmental influence, dictating individual consistency in migratory patterns. Unfortunately, field tests of individual consistency compared to the importance of individual condition on migratory propensity are rare. Here we analyse 6 years of field data on roach migration, gathered by tagging almost 3000 individual fish and monitoring their seasonal migrations over extended periods of time. Our aims were to provide a field test of the role of condition in wild fish for migratory decisions, and also to assess individual consistency in migratory tendency. Our analyses reveal that (1) migratory strategy, in terms of migration/residency, is highly consistent within individuals over time and (2) there is a positive relationship between condition and the probability of migration, but only in individuals that adopt a migratory strategy at some point during their lives. However, life-long residents do not differ in condition to migrants, hence body condition is only a good predictor of migratory tendency in fish with migratory phenotypes and not a more general determinant of migratory tendency for the population. As resident individuals can achieve very high body condition and still remain resident, we suggest that our data provides some of the first field evidence to show that both facultative and obligate strategies can co-exist within populations of migratory animals.

The Use of New Technology and Occupational Mobility: An Event History Analysis of the Swiss Labor Market

The investigation of the consequences of new technologies has a long standing tradition within economics. Particularly, labor economists are wondering how the introduction of new technologies, e.g. Personal Computers, have shaped labor markets. Former research has concentrated on the question of whether on-the-job use of PCs creates a wage bonus for employees. In this paper, we investigate whether the use of PCs increases employees’ probability of an upward shift in their employment status and whether it reduces the risk of involuntary labor market exits. We do so by applying event history analysis to the Swiss Labor Market Survey, a random sample of 3028 respondents, and by analyzing a Panel sub-sample of 650 respondents conducted recently in Switzerland. Our results show that on-the-job use of PCs was beneficial for employees in the past by increasing their probability of an upward shift by approximately 50%. The analysis also suggests that PC use reduces the risk and duration of unemployment. However, these latter results fail to reach statistical significance.

A logarithmic efficient estimator of the probability of ruin with recuperation for spectrally negative Lévy risk processes

This article provides an importance sampling algorithm for computing the probability of ruin with recuperation of a spectrally negative Lévy risk process with light-tailed downwards jumps. Ruin with recuperation corresponds to the following double passage event: for some t∈(0,∞)t∈(0,∞), the risk process starting at level x∈[0,∞)x∈[0,∞) falls below the null level during the period [0,t][0,t] and returns above the null level at the end of the period tt. The proposed Monte Carlo estimator is logarithmic efficient, as t,x→∞t,x→∞, when y=t/xy=t/x is constant and below a certain bound.

Potential Impact of Sexual Transmission on Ebola Virus Epidemiology: Sierra Leone as a Case Study.

BACKGROUND Sexual transmission of Ebola virus disease (EVD) 6 months after onset of symptoms has been recently documented, and Ebola virus RNA has been detected in semen of survivors up to 9 months after onset of symptoms. As countries affected by the 2013-2015 epidemic in West Africa, by far the largest to date, are declared free of Ebola virus disease (EVD), it remains unclear what threat is posed by rare sexual transmission events that could arise from survivors. METHODOLOGY/PRINCIPAL FINDINGS We devised a compartmental mathematical model that includes sexual transmission from convalescent survivors: a SEICR (susceptible-exposed-infectious-convalescent-recovered) transmission model. We fitted the model to weekly incidence of EVD cases from the 2014-2015 epidemic in Sierra Leone. Sensitivity analyses and Monte Carlo simulations showed that a 0.1% per sex act transmission probability and a 3-month convalescent period (the two key unknown parameters of sexual transmission) create very few additional cases, but would extend the epidemic by 83 days [95% CI: 68-98 days] (p < 0.0001) on average. Strikingly, a 6-month convalescent period extended the average epidemic by 540 days (95% CI: 508-572 days), doubling the current length, despite an insignificant rise in the number of new cases generated. CONCLUSIONS/SIGNIFICANCE Our results show that reductions in the per sex act transmission probability via abstinence and condom use should reduce the number of sporadic sexual transmission events, but will not significantly reduce the epidemic size and may only minimally shorten the length of time the public health community must maintain response preparedness. While the number of infectious survivors is expected to greatly decline over the coming months, our results show that transmission events may still be expected for quite some time as each event results in a new potential cluster of non-sexual transmission. Precise measurement of the convalescent period is thus important for planning ongoing surveillance efforts.