Bayesian Networks for Probabilistic Inference and Decision Analysis in Forensic Science

Continuing developments in science and technology mean that the amounts of information forensic scientists are able to provide for criminal investigations is ever increasing. The commensurate increase in complexity creates difficulties for scientists and lawyers with regard to evaluation and interpretation, notably with respect to issues of inference and decision. Probability theory, implemented through graphical methods, and specifically Bayesian networks, provides powerful methods to deal with this complexity. Extensions of these methods to elements of decision theory provide further support and assistance to the judicial system. Bayesian Networks for Probabilistic Inference and Decision Analysis in Forensic Science provides a unique and comprehensive introduction to the use of Bayesian decision networks for the evaluation and interpretation of scientific findings in forensic science, and for the support of decision-makers in their scientific and legal tasks. Includes self-contained introductions to probability and decision theory. Develops the characteristics of Bayesian networks, object-oriented Bayesian networks and their extension to decision models. Features implementation of the methodology with reference to commercial and academically available software. Presents standard networks and their extensions that can be easily implemented and that can assist in the reader's own analysis of real cases. Provides a technique for structuring problems and organizing data based on methods and principles of scientific reasoning. Contains a method for the construction of coherent and defensible arguments for the analysis and evaluation of scientific findings and for decisions based on them. Is written in a lucid style, suitable for forensic scientists and lawyers with minimal mathematical background. Includes a foreword by Ian Evett. The clear and accessible style of this second edition makes this book ideal for all forensic scientists, applied statisticians and graduate students wishing to evaluate forensic findings from the perspective of probability and decision analysis. It will also appeal to lawyers and other scientists and professionals interested in the evaluation and interpretation of forensic findings, including decision making based on scientific information.

The use of classification and regression trees to predict the likelihood of seasonal influenza.

Background Individual signs and symptoms are of limited value for the diagnosis of influenza. Objective To develop a decision tree for the diagnosis of influenza based on a classification and regression tree (CART) analysis. Methods Data from two previous similar cohort studies were assembled into a single dataset. The data were randomly divided into a development set (70%) and a validation set (30%). We used CART analysis to develop three models that maximize the number of patients who do not require diagnostic testing prior to treatment decisions. The validation set was used to evaluate overfitting of the model to the training set. Results Model 1 has seven terminal nodes based on temperature, the onset of symptoms and the presence of chills, cough and myalgia. Model 2 was a simpler tree with only two splits based on temperature and the presence of chills. Model 3 was developed with temperature as a dichotomous variable (≥38°C) and had only two splits based on the presence of fever and myalgia. The area under the receiver operating characteristic curves (AUROCC) for the development and validation sets, respectively, were 0.82 and 0.80 for Model 1, 0.75 and 0.76 for Model 2 and 0.76 and 0.77 for Model 3. Model 2 classified 67% of patients in the validation group into a high- or low-risk group compared with only 38% for Model 1 and 54% for Model 3. Conclusions A simple decision tree (Model 2) classified two-thirds of patients as low or high risk and had an AUROCC of 0.76. After further validation in an independent population, this CART model could support clinical decision making regarding influenza, with low-risk patients requiring no further evaluation for influenza and high-risk patients being candidates for empiric symptomatic or drug therapy.

The likelihood approach to compare populations : a study on DNA evidence and pitfalls of intuitions

The paper follows on from earlier work [Taroni F and Aitken CGG. Probabilistic reasoning in the law, Part 1: assessment of probabilities and explanation of the value of DNA evidence. Science & Justice 1998; 38: 165-177]. Different explanations of the value of DNA evidence were presented to students from two schools of forensic science and to members of fifteen laboratories all around the world. The responses were divided into two groups; those which came from a school or laboratory identified as Bayesian and those which came from a school or laboratory identified as non-Bayesian. The paper analyses these responses using a likelihood approach. This approach is more consistent with a Bayesian analysis than one based on a frequentist approach, as was reported by Taroni F and Aitken CGG. [Probabilistic reasoning in the law, Part 1: assessment of probabilities and explanation of the value of DNA evidence] in Science & Justice 1998.

Hierarchical conditional random fields for parts based models matching

A parts based model is a parametrization of an object class using a collection of landmarks following the object structure. The matching of parts based models is one of the problems where pairwise Conditional Random Fields have been successfully applied. The main reason of their effectiveness is tractable inference and learning due to the simplicity of involved graphs, usually trees. However, these models do not consider possible patterns of statistics among sets of landmarks, and thus they sufffer from using too myopic information. To overcome this limitation, we propoese a novel structure based on a hierarchical Conditional Random Fields, which we explain in the first part of this memory. We build a hierarchy of combinations of landmarks, where matching is performed taking into account the whole hierarchy. To preserve tractable inference we effectively sample the label set. We test our method on facial feature selection and human pose estimation on two challenging datasets: Buffy and MultiPIE. In the second part of this memory, we present a novel approach to multiple kernel combination that relies on stacked classification. This method can be used to evaluate the landmarks of the parts-based model approach. Our method is based on combining responses of a set of independent classifiers for each individual kernel. Unlike earlier approaches that linearly combine kernel responses, our approach uses them as inputs to another set of classifiers. We will show that we outperform state-of-the-art methods on most of the standard benchmark datasets.

Uncertainty about the true source. A note on the likelihood ratio at the activity level.

This paper focuses on likelihood ratio based evaluations of fibre evidence in cases in which there is uncertainty about whether or not the reference item available for analysis - that is, an item typically taken from the suspect or seized at his home - is the item actually worn at the time of the offence. A likelihood ratio approach is proposed that, for situations in which certain categorical assumptions can be made about additionally introduced parameters, converges to formula described in existing literature. The properties of the proposed likelihood ratio approach are analysed through sensitivity analyses and discussed with respect to possible argumentative implications that arise in practice.

Inference about the number of contributors to a DNA mixture: Comparative analyses of a Bayesian network approach and the maximum allele count method.

In the forensic examination of DNA mixtures, the question of how to set the total number of contributors (N) presents a topic of ongoing interest. Part of the discussion gravitates around issues of bias, in particular when assessments of the number of contributors are not made prior to considering the genotypic configuration of potential donors. Further complication may stem from the observation that, in some cases, there may be numbers of contributors that are incompatible with the set of alleles seen in the profile of a mixed crime stain, given the genotype of a potential contributor. In such situations, procedures that take a single and fixed number contributors as their output can lead to inferential impasses. Assessing the number of contributors within a probabilistic framework can help avoiding such complication. Using elements of decision theory, this paper analyses two strategies for inference on the number of contributors. One procedure is deterministic and focuses on the minimum number of contributors required to 'explain' an observed set of alleles. The other procedure is probabilistic using Bayes' theorem and provides a probability distribution for a set of numbers of contributors, based on the set of observed alleles as well as their respective rates of occurrence. The discussion concentrates on mixed stains of varying quality (i.e., different numbers of loci for which genotyping information is available). A so-called qualitative interpretation is pursued since quantitative information such as peak area and height data are not taken into account. The competing procedures are compared using a standard scoring rule that penalizes the degree of divergence between a given agreed value for N, that is the number of contributors, and the actual value taken by N. Using only modest assumptions and a discussion with reference to a casework example, this paper reports on analyses using simulation techniques and graphical models (i.e., Bayesian networks) to point out that setting the number of contributors to a mixed crime stain in probabilistic terms is, for the conditions assumed in this study, preferable to a decision policy that uses categoric assumptions about N.

Adaptively weighted maximum likelihood estimation of discrete distributions

SummaryDiscrete data arise in various research fields, typically when the observations are count data.I propose a robust and efficient parametric procedure for estimation of discrete distributions. The estimation is done in two phases. First, a very robust, but possibly inefficient, estimate of the model parameters is computed and used to indentify outliers. Then the outliers are either removed from the sample or given low weights, and a weighted maximum likelihood estimate (WML) is computed.The weights are determined via an adaptive process such that if the data follow the model, then asymptotically no observation is downweighted.I prove that the final estimator inherits the breakdown point of the initial one, and that its influence function at the model is the same as the influence function of the maximum likelihood estimator, which strongly suggests that it is asymptotically fully efficient.The initial estimator is a minimum disparity estimator (MDE). MDEs can be shown to have full asymptotic efficiency, and some MDEs have very high breakdown points and very low bias under contamination. Several initial estimators are considered, and the performances of the WMLs based on each of them are studied.It results that in a great variety of situations the WML substantially improves the initial estimator, both in terms of finite sample mean square error and in terms of bias under contamination. Besides, the performances of the WML are rather stable under a change of the MDE even if the MDEs have very different behaviors.Two examples of application of the WML to real data are considered. In both of them, the necessity for a robust estimator is clear: the maximum likelihood estimator is badly corrupted by the presence of a few outliers.This procedure is particularly natural in the discrete distribution setting, but could be extended to the continuous case, for which a possible procedure is sketched.RésuméLes données discrètes sont présentes dans différents domaines de recherche, en particulier lorsque les observations sont des comptages.Je propose une méthode paramétrique robuste et efficace pour l'estimation de distributions discrètes. L'estimation est faite en deux phases. Tout d'abord, un estimateur très robuste des paramètres du modèle est calculé, et utilisé pour la détection des données aberrantes (outliers). Cet estimateur n'est pas nécessairement efficace. Ensuite, soit les outliers sont retirés de l'échantillon, soit des faibles poids leur sont attribués, et un estimateur du maximum de vraisemblance pondéré (WML) est calculé.Les poids sont déterminés via un processus adaptif, tel qu'asymptotiquement, si les données suivent le modèle, aucune observation n'est dépondérée.Je prouve que le point de rupture de l'estimateur final est au moins aussi élevé que celui de l'estimateur initial, et que sa fonction d'influence au modèle est la même que celle du maximum de vraisemblance, ce qui suggère que cet estimateur est pleinement efficace asymptotiquement.L'estimateur initial est un estimateur de disparité minimale (MDE). Les MDE sont asymptotiquement pleinement efficaces, et certains d'entre eux ont un point de rupture très élevé et un très faible biais sous contamination. J'étudie les performances du WML basé sur différents MDEs.Le résultat est que dans une grande variété de situations le WML améliore largement les performances de l'estimateur initial, autant en terme du carré moyen de l'erreur que du biais sous contamination. De plus, les performances du WML restent assez stables lorsqu'on change l'estimateur initial, même si les différents MDEs ont des comportements très différents.Je considère deux exemples d'application du WML à des données réelles, où la nécessité d'un estimateur robuste est manifeste : l'estimateur du maximum de vraisemblance est fortement corrompu par la présence de quelques outliers.La méthode proposée est particulièrement naturelle dans le cadre des distributions discrètes, mais pourrait être étendue au cas continu.

Subject-based steroid profiling and the determination of novel biomarkers for DHT and DHEA misuse in sports.

Doping with natural steroids can be detected by evaluating the urinary concentrations and ratios of several endogenous steroids. Since these biomarkers of steroid doping are known to present large inter-individual variations, monitoring of individual steroid profiles over time allows switching from population-based towards subject-based reference ranges for improved detection. In an Athlete Biological Passport (ABP), biomarkers data are collated throughout the athlete's sporting career and individual thresholds defined adaptively. For now, this approach has been validated on a limited number of markers of steroid doping, such as the testosterone (T) over epitestosterone (E) ratio to detect T misuse in athletes. Additional markers are required for other endogenous steroids like dihydrotestosterone (DHT) and dehydroepiandrosterone (DHEA). By combining comprehensive steroid profiles composed of 24 steroid concentrations with Bayesian inference techniques for longitudinal profiling, a selection was made for the detection of DHT and DHEA misuse. The biomarkers found were rated according to relative response, parameter stability, discriminative power, and maximal detection time. This analysis revealed DHT/E, DHT/5β-androstane-3α,17β-diol and 5α-androstane-3α,17β-diol/5β-androstane-3α,17β-diol as best biomarkers for DHT administration and DHEA/E, 16α-hydroxydehydroepiandrosterone/E, 7β-hydroxydehydroepiandrosterone/E and 5β-androstane-3α,17β-diol/5α-androstane-3α,17β-diol for DHEA. The selected biomarkers were found suitable for individual referencing. A drastic overall increase in sensitivity was obtained. The use of multiple markers as formalized in an Athlete Steroidal Passport (ASP) can provide firm evidence of doping with endogenous steroids.

The evolutionary history of the CD209 (DC-SIGN) family in humans and non-human primates

The CD209 gene family that encodes C-type lectins in primates includes CD209 (DC-SIGN), CD209L (L-SIGN) and CD209L2. Understanding the evolution of these genes can help understand the duplication events generating this family, the process leading to the repeated neck region and identify protein domains under selective pressure. We compiled sequences from 14 primates representing 40 million years of evolution and from three non-primate mammal species. Phylogenetic analyses used Bayesian inference, and nucleotide substitutional patterns were assessed by codon-based maximum likelihood. Analyses suggest that CD209 genes emerged from a first duplication event in the common ancestor of anthropoids, yielding CD209L2 and an ancestral CD209 gene, which, in turn, duplicated in the common Old World primate ancestor, giving rise to CD209L and CD209. K(A)/K(S) values averaged over the entire tree were 0.43 (CD209), 0.52 (CD209L) and 0.35 (CD209L2), consistent with overall signatures of purifying selection. We also assessed the Toll-like receptor (TLR) gene family, which shares with CD209 genes a common profile of evolutionary constraint. The general feature of purifying selection of CD209 genes, despite an apparent redundancy (gene absence and gene loss), may reflect the need to faithfully recognize a multiplicity of pathogen motifs, commensals and a number of self-antigens

Protocol of the COSMIN study: consensus-based standards for the selection of health measurement instruments

Background: Choosing an adequate measurement instrument depends on the proposed use of the instrument, the concept to be measured, the measurement properties (e.g. internal consistency, reproducibility, content and construct validity, responsiveness, and interpretability), the requirements, the burden for subjects, and costs of the available instruments. As far as measurement properties are concerned, there are no sufficiently specific standards for the evaluation of measurement properties of instruments to measure health status, and also no explicit criteria for what constitutes good measurement properties. In this paper we describe the protocol for the COSMIN study, the objective of which is to develop a checklist that contains COnsensus-based Standards for the selection of health Measurement INstruments, including explicit criteria for satisfying these standards. We will focus on evaluative health related patient-reported outcomes (HR-PROs), i.e. patient-reported health measurement instruments used in a longitudinal design as an outcome measure, excluding health care related PROs, such as satisfaction with care or adherence. The COSMIN standards will be made available in the form of an easily applicable checklist.Method: An international Delphi study will be performed to reach consensus on which and how measurement properties should be assessed, and on criteria for good measurement properties. Two sources of input will be used for the Delphi study: (1) a systematic review of properties, standards and criteria of measurement properties found in systematic reviews of measurement instruments, and (2) an additional literature search of methodological articles presenting a comprehensive checklist of standards and criteria. The Delphi study will consist of four (written) Delphi rounds, with approximately 30 expert panel members with different backgrounds in clinical medicine, biostatistics, psychology, and epidemiology. The final checklist will subsequently be field-tested by assessing the inter-rater reproducibility of the checklist.Discussion: Since the study will mainly be anonymous, problems that are commonly encountered in face-to-face group meetings, such as the dominance of certain persons in the communication process, will be avoided. By performing a Delphi study and involving many experts, the likelihood that the checklist will have sufficient credibility to be accepted and implemented will increase.

Vitamin D levels and associated factors: a population-based study in Switzerland.

QUESTIONS UNDER STUDY: To update the prevalence of vitamin D insufficiency and to identify factors associated with vitamin D status in the Swiss adult population. METHODS: Data from the 2010-2011 Swiss Study on Salt intake, a population-based study in the Swiss population, was used. Vitamin D concentration in serum was measured by liquid chromatography- tandem mass spectrometry. Major factors that influence vitamin D levels were taken into account. Survey statistical procedures were used to estimate means and prevalences of vitamin D levels and status. Monthly-specific tertiles of vitamin D and ordinal logistic regression were used to determine the associations of covariates of interest with vitamin D status. RESULTS: The prevalences of vitamin D insufficiency (serum 25-hydroxyvitamin D: 20-29.9 ng/ml) and deficiency (<20 ng/ml) were the highest in the January-March period; 26.4% (95%CI: 21.6-31.7) and 61.6% (95%CI: 56.0-67.0), respectively. In the same period, more than 9 of ten men were vitamin D insufficient or deficient. Each unit increase of Body Mass Index was associated with an 8% decreased likelihood of being in a higher vitamin D tertiles. Oral contraceptive, altitude, urinary excretion of calcium, use of vitamin D supplement or treatment, high wine consumption, physical activity were associated with vitamin D tertiles. Compared to the French-speaking region, the Italian-speaking region was independently associated with a higher likelihood of being in higher vitamin D tertiles (OR: 1.66, 95%CI: 1.14-2.43). CONCLUSIONS: Low levels of vitamin D are common among Swiss adults, in particular during winter months and outside the Italian-speaking region.

Bringing game theory to hypothesis testing: Establishing finite sample bounds on inference

Small sample properties are of fundamental interest when only limited data is avail-able. Exact inference is limited by constraints imposed by speci.c nonrandomizedtests and of course also by lack of more data. These e¤ects can be separated as we propose to evaluate a test by comparing its type II error to the minimal type II error among all tests for the given sample. Game theory is used to establish this minimal type II error, the associated randomized test is characterized as part of a Nash equilibrium of a .ctitious game against nature.We use this method to investigate sequential tests for the di¤erence between twomeans when outcomes are constrained to belong to a given bounded set. Tests ofinequality and of noninferiority are included. We .nd that inference in terms oftype II error based on a balanced sample cannot be improved by sequential sampling or even by observing counter factual evidence providing there is a reasonable gap between the hypotheses.

Robust response transformations based on optimal prediction

Nonlinear regression problems can often be reduced to linearity by transforming the response variable (e.g., using the Box-Cox family of transformations). The classic estimates of the parameter defining the transformation as well as of the regression coefficients are based on the maximum likelihood criterion, assuming homoscedastic normal errors for the transformed response. These estimates are nonrobust in the presence of outliers and can be inconsistent when the errors are nonnormal or heteroscedastic. This article proposes new robust estimates that are consistent and asymptotically normal for any unimodal and homoscedastic error distribution. For this purpose, a robust version of conditional expectation is introduced for which the prediction mean squared error is replaced with an M scale. This concept is then used to develop a nonparametric criterion to estimate the transformation parameter as well as the regression coefficients. A finite sample estimate of this criterion based on a robust version of smearing is also proposed. Monte Carlo experiments show that the new estimates compare favorably with respect to the available competitors.

Inference with the lognormal distribution

Several estimators of the expectation, median and mode of the lognormal distribution are derived. They aim to be approximately unbiased, efficient, or have a minimax property in the class of estimators we introduce. The small-sample properties of these estimators are assessed by simulations and, when possible, analytically. Some of these estimators of the expectation are far more efficient than the maximum likelihood or the minimum-variance unbiased estimator, even for substantial samplesizes.

Global Genomic Diversity of Human Papillomavirus 6 Based on 724 Isolates and 190 Complete Genome Sequences.

Human papillomavirus type 6 (HPV6) is the major etiological agent of anogenital warts and laryngeal papillomas and has been included in both the quadrivalent and nonavalent prophylactic HPV vaccines. This study investigated the global genomic diversity of HPV6, using 724 isolates and 190 complete genomes from six continents, and the association of HPV6 genomic variants with geographical location, anatomical site of infection/disease, and gender. Initially, a 2,800-bp E5a-E5b-L1-LCR fragment was sequenced from 492/530 (92.8%) HPV6-positive samples collected for this study. Among them, 130 exhibited at least one single nucleotide polymorphism (SNP), indel, or amino acid change in the E5a-E5b-L1-LCR fragment and were sequenced in full. A global alignment and maximum likelihood tree of 190 complete HPV6 genomes (130 fully sequenced in this study and 60 obtained from sequence repositories) revealed two variant lineages, A and B, and five B sublineages: B1, B2, B3, B4, and B5. HPV6 (sub)lineage-specific SNPs and a 960-bp representative region for whole-genome-based phylogenetic clustering within the L2 open reading frame were identified. Multivariate logistic regression analysis revealed that lineage B predominated globally. Sublineage B3 was more common in Africa and North and South America, and lineage A was more common in Asia. Sublineages B1 and B3 were associated with anogenital infections, indicating a potential lesion-specific predilection of some HPV6 sublineages. Females had higher odds for infection with sublineage B3 than males. In conclusion, a global HPV6 phylogenetic analysis revealed the existence of two variant lineages and five sublineages, showing some degree of ethnogeographic, gender, and/or disease predilection in their distribution. IMPORTANCE: This study established the largest database of globally circulating HPV6 genomic variants and contributed a total of 130 new, complete HPV6 genome sequences to available sequence repositories. Two HPV6 variant lineages and five sublineages were identified and showed some degree of association with geographical location, anatomical site of infection/disease, and/or gender. We additionally identified several HPV6 lineage- and sublineage-specific SNPs to facilitate the identification of HPV6 variants and determined a representative region within the L2 gene that is suitable for HPV6 whole-genome-based phylogenetic analysis. This study complements and significantly expands the current knowledge of HPV6 genetic diversity and forms a comprehensive basis for future epidemiological, evolutionary, functional, pathogenicity, vaccination, and molecular assay development studies.