Forensic bitemark identification: weak foundations, exaggerated claims.

Several forensic sciences, especially of the pattern-matching kind, are increasingly seen to lack the scientific foundation needed to justify continuing admission as trial evidence. Indeed, several have been abolished in the recent past. A likely next candidate for elimination is bitemark identification. A number of DNA exonerations have occurred in recent years for individuals convicted based on erroneous bitemark identifications. Intense scientific and legal scrutiny has resulted. An important National Academies review found little scientific support for the field. The Texas Forensic Science Commission recently recommended a moratorium on the admission of bitemark expert testimony. The California Supreme Court has a case before it that could start a national dismantling of forensic odontology. This article describes the (legal) basis for the rise of bitemark identification and the (scientific) basis for its impending fall. The article explains the general logic of forensic identification, the claims of bitemark identification, and reviews relevant empirical research on bitemark identification-highlighting both the lack of research and the lack of support provided by what research does exist. The rise and possible fall of bitemark identification evidence has broader implications-highlighting the weak scientific culture of forensic science and the law's difficulty in evaluating and responding to unreliable and unscientific evidence.

Scalable privacy-preserving data sharing methodology for genome-wide association studies.

The protection of privacy of individual-level information in genome-wide association study (GWAS) databases has been a major concern of researchers following the publication of "an attack" on GWAS data by Homer et al. (2008). Traditional statistical methods for confidentiality and privacy protection of statistical databases do not scale well to deal with GWAS data, especially in terms of guarantees regarding protection from linkage to external information. The more recent concept of differential privacy, introduced by the cryptographic community, is an approach that provides a rigorous definition of privacy with meaningful privacy guarantees in the presence of arbitrary external information, although the guarantees may come at a serious price in terms of data utility. Building on such notions, Uhler et al. (2013) proposed new methods to release aggregate GWAS data without compromising an individual's privacy. We extend the methods developed in Uhler et al. (2013) for releasing differentially-private χ(2)-statistics by allowing for arbitrary number of cases and controls, and for releasing differentially-private allelic test statistics. We also provide a new interpretation by assuming the controls' data are known, which is a realistic assumption because some GWAS use publicly available data as controls. We assess the performance of the proposed methods through a risk-utility analysis on a real data set consisting of DNA samples collected by the Wellcome Trust Case Control Consortium and compare the methods with the differentially-private release mechanism proposed by Johnson and Shmatikov (2013).

Reconceptualizing the classification of PNAS articles.

PNAS article classification is rooted in long-standing disciplinary divisions that do not necessarily reflect the structure of modern scientific research. We reevaluate that structure using latent pattern models from statistical machine learning, also known as mixed-membership models, that identify semantic structure in co-occurrence of words in the abstracts and references. Our findings suggest that the latent dimensionality of patterns underlying PNAS research articles in the Biological Sciences is only slightly larger than the number of categories currently in use, but it differs substantially in the content of the categories. Further, the number of articles that are listed under multiple categories is only a small fraction of what it should be. These findings together with the sensitivity analyses suggest ways to reconceptualize the organization of papers published in PNAS.

Detection of deception using fMRI: better than chance, but well below perfection.

Functional brain imaging has been considered a new and better technique for the detection of deception. The reasoning is that there is a neural locus or circuit for lying that is sensitive, specific, generalizable across individuals and measurement contexts, and robust to countermeasures. To determine the extent to which the group results predicted lying at the level of the individual, we reanalyzed data on 14 participants from a study that had previously identified regions involved in lying (thus satisfying the criterion for sensitivity). We assessed the efficacy of functionally determined brain regions based on the lie-truth contrast for N-1 participants to detect deception in the Nth individual. Results showed that no region could be used to correctly detect deception across all individuals. The best results were obtained for medial prefrontal cortex (mPFC), correctly identifying 71% of participants as lying with no false alarms. Lowering the threshold for a response increased hits and false alarms. The results suggest that although brain imaging is a more direct index of cognition than the traditional polygraph, it is subject to many of the same caveats and thus neuroimaging does not appear to reveal processes that are necessarily unique to deception.

Population size estimation using individual level mixture models.

We revisit the heterogeneous closed population multiple recapture problem, modeling individual-level heterogeneity using the Grade of Membership model (Woodbury et al., 1978). This strategy allows us to postulate the existence of homogeneous latent "ideal" or "pure" classes within the population, and construct a soft clustering of the individuals, where each one is allowed partial or mixed membership in all of these classes. We propose a full hierarchical Bayes specification and a MCMC algorithm to obtain samples from the posterior distribution. We apply the method to simulated data and to three real life examples.

Mixed Membership Stochastic Blockmodels.

Observations consisting of measurements on relationships for pairs of objects arise in many settings, such as protein interaction and gene regulatory networks, collections of author-recipient email, and social networks. Analyzing such data with probabilisic models can be delicate because the simple exchangeability assumptions underlying many boilerplate models no longer hold. In this paper, we describe a latent variable model of such data called the mixed membership stochastic blockmodel. This model extends blockmodels for relational data to ones which capture mixed membership latent relational structure, thus providing an object-specific low-dimensional representation. We develop a general variational inference algorithm for fast approximate posterior inference. We explore applications to social and protein interaction networks.

DESCRIBING DISABILITY THROUGH INDIVIDUAL-LEVEL MIXTURE MODELS FOR MULTIVARIATE BINARY DATA.

Data on functional disability are of widespread policy interest in the United States, especially with respect to planning for Medicare and Social Security for a growing population of elderly adults. We consider an extract of functional disability data from the National Long Term Care Survey (NLTCS) and attempt to develop disability profiles using variations of the Grade of Membership (GoM) model. We first describe GoM as an individual-level mixture model that allows individuals to have partial membership in several mixture components simultaneously. We then prove the equivalence between individual-level and population-level mixture models, and use this property to develop a Markov Chain Monte Carlo algorithm for Bayesian estimation of the model. We use our approach to analyze functional disability data from the NLTCS.

Statistical issues and challenges associated with rapid detection of bio-terrorist attacks.

The traditional focus for detecting outbreaks of an epidemic or bio-terrorist attack has been on the collection and analysis of medical and public health data. Although such data are the most direct indicators of symptoms, they tend to be collected, delivered, and analysed days, weeks, and even months after the outbreak. By the time this information reaches decision makers it is often too late to treat the infected population or to react in some other way. In this paper, we explore different sources of data, traditional and non-traditional, that can be used for detecting a bio-terrorist attack in a timely manner. We set our discussion in the context of state-of-the-art syndromic surveillance systems and we focus on statistical issues and challenges associated with non-traditional data sources and the timely integration of multiple data sources for detection purposes.

Bone lead levels in adjudicated delinquents. A case control study.

BACKGROUND: Lead exposure shares many risk factors with delinquent behavior, and bone lead levels are related to self-reports of delinquent acts. No data exist as to whether lead exposure is higher in arrested delinquents. The goal of this study is to evaluate the association between lead exposure, as reflected in bone lead levels, and adjudicated delinquency. METHODS: This is a case-control study of 194 youths aged 12-18, arrested and adjudicated as delinquent by the Juvenile Court of Allegheny County, PA and 146 nondelinquent controls from high schools in the city of Pittsburgh. Bone lead was measured by K-line X-ray fluorescence (XRF) spectroscopy of tibia. Logistic regression was used to model the association between delinquent status and bone lead concentration. Covariates entered into the model were race, parent education and occupation, presence of two parental figures in the home, number of children in the home and neighborhood crime rate. Separate regression analyses were also conducted after stratification on race. RESULTS: Cases had significantly higher mean concentrations of lead in their bones than controls (11.0+/-32.7 vs. 1.5+/-32.1 ppm). This was true for both Whites and African Americans. The unadjusted odds ratio for a lead level > or =25 vs. <25 ppm was 1.9 (95% CL: 1.1-3.2). After adjustment for covariates and interactions and removal of noninfluential covariates, adjudicated delinquents were four times more likely to have bone lead concentrations >25 ppm than controls (OR=4.0, 95% CL: 1.4-11.1). CONCLUSION: Elevated body lead burdens, measured by bone lead concentrations, are associated with elevated risk for adjudicated delinquency.

Early statistical detection of anthrax outbreaks by tracking over-the-counter medication sales.

The recent series of anthrax attacks has reinforced the importance of biosurveillance systems for the timely detection of epidemics. This paper describes a statistical framework for monitoring grocery data to detect a large-scale but localized bioterrorism attack. Our system illustrates the potential of data sources that may be more timely than traditional medical and public health data. The system includes several layers, each customized to grocery data and tuned to finding footprints of an epidemic. We also propose an evaluation methodology that is suitable in the absence of data on large-scale bioterrorist attacks and disease outbreaks.