Kernel-PCA data integration with enhanced interpretability

Background Nowadays, combining the different sources of information to improve the biological knowledge available is a challenge in bioinformatics. One of the most powerful methods for integrating heterogeneous data types are kernel-based methods. Kernel-based data integration approaches consist of two basic steps: firstly the right kernel is chosen for each data set; secondly the kernels from the different data sources are combined to give a complete representation of the available data for a given statistical task. Results We analyze the integration of data from several sources of information using kernel PCA, from the point of view of reducing dimensionality. Moreover, we improve the interpretability of kernel PCA by adding to the plot the representation of the input variables that belong to any dataset. In particular, for each input variable or linear combination of input variables, we can represent the direction of maximum growth locally, which allows us to identify those samples with higher/lower values of the variables analyzed. Conclusions The integration of different datasets and the simultaneous representation of samples and variables together give us a better understanding of biological knowledge.

Kernel-PCA data integration with enhanced interpretability

Background Nowadays, combining the different sources of information to improve the biological knowledge available is a challenge in bioinformatics. One of the most powerful methods for integrating heterogeneous data types are kernel-based methods. Kernel-based data integration approaches consist of two basic steps: firstly the right kernel is chosen for each data set; secondly the kernels from the different data sources are combined to give a complete representation of the available data for a given statistical task. Results We analyze the integration of data from several sources of information using kernel PCA, from the point of view of reducing dimensionality. Moreover, we improve the interpretability of kernel PCA by adding to the plot the representation of the input variables that belong to any dataset. In particular, for each input variable or linear combination of input variables, we can represent the direction of maximum growth locally, which allows us to identify those samples with higher/lower values of the variables analyzed. Conclusions The integration of different datasets and the simultaneous representation of samples and variables together give us a better understanding of biological knowledge.

Patient-reported outcomes in meta-analyses - part 2: methods for improving interpretability for decision-makers

Systematic reviews and meta-analyses of randomized trials that include patient-reported outcomes (PROs) often provide crucial information for patients, clinicians and policy-makers facing challenging health care decisions. Based on emerging methods, guidance on improving the interpretability of meta-analysis of patient-reported outcomes, typically continuous in nature, is likely to enhance decision-making. The objective of this paper is to summarize approaches to enhancing the interpretability of pooled estimates of PROs in meta-analyses. When differences in PROs between groups are statistically significant, decision-makers must be able to interpret the magnitude of effect. This is challenging when, as is often the case, clinical trial investigators use different measurement instruments for the same construct within and between individual randomized trials. For such cases, in addition to pooling results as a standardized mean difference, we recommend that systematic review authors use other methods to present results such as relative (relative risk, odds ratio) or absolute (risk difference) dichotomized treatment effects, complimented by presentation in either: natural units (e.g. overall depression reduced by 2.4 points when measured on a 50-point Hamilton Rating Scale for Depression); minimal important difference units (e.g. where 1.0 unit represents the smallest difference in depression that patients, on average, perceive as important the depression score was 0.38 (95%CI 0.30 to 0.47) units less than the control group); or a ratio of means (e.g. where the mean in the treatment group is divided by the mean in the control group, the ratio of means is 1.27, representing a 27%relative reduction in the mean depression score).

A new model construction by making a detour via intuitionistic theories II: Interpretability lower bound of Feferman's explicit mathematics T0

We partially solve a long-standing problem in the proof theory of explicit mathematics or the proof theory in general. Namely, we give a lower bound of Feferman’s system T0 of explicit mathematics (but only when formulated on classical logic) with a concrete interpretat ion of the subsystem Σ12-AC+ (BI) of second order arithmetic inside T0. Whereas a lower bound proof in the sense of proof-theoretic reducibility or of ordinalanalysis was already given in 80s, the lower bound in the sense of interpretability we give here is new. We apply the new interpretation method developed by the author and Zumbrunnen (2015), which can be seen as the third kind of model construction method for classical theories, after Cohen’s forcing and Krivine’s classical realizability. It gives us an interpretation between classical theories, by composing interpretations between intuitionistic theories.

Digital Forensics AI: on Practicality, Optimality, and Interpretability of Digital Evidence Mining Techniques

Digital forensics as a field has progressed alongside technological advancements over the years, just as digital devices have gotten more robust and sophisticated. However, criminals and attackers have devised means for exploiting the vulnerabilities or sophistication of these devices to carry out malicious activities in unprecedented ways. Their belief is that electronic crimes can be committed without identities being revealed or trails being established. Several applications of artificial intelligence (AI) have demonstrated interesting and promising solutions to seemingly intractable societal challenges. This thesis aims to advance the concept of applying AI techniques in digital forensic investigation. Our approach involves experimenting with a complex case scenario in which suspects corresponded by e-mail and deleted, suspiciously, certain communications, presumably to conceal evidence. The purpose is to demonstrate the efficacy of Artificial Neural Networks (ANN) in learning and detecting communication patterns over time, and then predicting the possibility of missing communication(s) along with potential topics of discussion. To do this, we developed a novel approach and included other existing models. The accuracy of our results is evaluated, and their performance on previously unseen data is measured. Second, we proposed conceptualizing the term “Digital Forensics AI” (DFAI) to formalize the application of AI in digital forensics. The objective is to highlight the instruments that facilitate the best evidential outcomes and presentation mechanisms that are adaptable to the probabilistic output of AI models. Finally, we enhanced our notion in support of the application of AI in digital forensics by recommending methodologies and approaches for bridging trust gaps through the development of interpretable models that facilitate the admissibility of digital evidence in legal proceedings.

Pyrolysis-Gas Chromatography/Mass Spectrometry of Soil Organic Matter Extracted from a Brazilian Mangrove and Spanish Salt Marshes

The soil organic matter (SOM) extracted under different vegetation types from a Brazilian mangrove (Pai Matos Island, Sao Paulo State) and from three Spanish salt marshes (Betanzos Ria and Corrubedo Natural Parks, Galicia, and the Albufera Natural Park, Valencia) was investigated by pyrolysis-gas chromatography/mass spectrometry (Py-GC/MS). The chemical variation was larger in SOM from the Spanish marshes than in the SOM of the Brazilian mangroves, possibly because the marshes included sites with both tidal and nontidal variation, whereas the mangrove forest underwent just tidal variation. Thus, plant-derived organic matter was better preserved under permanently anoxic environments. Moreover, given the low number of studied profiles and sedimentary-vegetation sequences in both areas, depth trends remain unclear. The chemical data also allow distinction between the contributions of woody and nonwoody vegetation inputs. Soil organic matter decomposition was found to cause: (i) a decrease in lignin contents and a relative increase in aliphatics; (ii) an increase in short-chain aliphatics at the expense of longer ones; (iii) a loss of odd-over-even dominance in alkanes and alkenes; and (iv) an increase in microbial products, including proteins, sterols, short-chain fatty acids, and alkanes. Pyrolysis-gas chromatography/mass spectrometry is a useful tool to study the behavior and composition of SOM in wetland environments such as mangroves and salt marshes. Additional profiles need to be studied for each vegetation type, however, to improve the interpretability of the chemical data.

Strategies of accommodation: Development of a coding system for conversational interaction

This study describes a coding system developed to operationalize the sociolinguistic strategies proposed by communication accommodation theory (CAT) in an academic context. Fifty interactions between two students (of Australian or Chinese ethnic background) or a student and faculty member were videotaped. A turn- and episode-based coding system was developed, focusing on verbal and nonverbal behavior. The development of this system is described in detail, before results are presented. Results indicated that status was the main influence on choice of strategies, particularly the extent and type of discourse management and interpersonal control. Participants' sew and ethnicity also played a role: Male participants made more use of interpretability (largely questions), whereas female participants used discourse management to develop a shared perspective. The results make clear that there is no automatic correspondence between behaviors and the strategies they constitute, and they point to the appropriateness of conceptualizing behavior and strategies separately in CAT.

Contributo para a validação da versão portuguesa da Quebec Back Pain Disability Scale : fiabilidade e poder de resposta em utentes com dor lombar crónica

RESUMO: Um dos principais resultados das intervenções de Fisioterapia dirigidas a utentes com Dor Lombar Crónica (DLC) é reduzir a incapacidade funcional. A Quebec Back Pain Disability Scale (QBPDS) é um instrumento amplamente aceite a nível internacional na medição do nível de incapacidade funcional reportada pelos indivíduos com DLC. O objetivo deste estudo é dar um contributo para a adaptação cultural da versão portuguesa da QBPDS (QBPDS-VP) e investigar o poder de resposta e interpretabilidade da escala. Metodologia: Realizou-se um estudo metodológico, multicentro, baseado num coorte prospetivo de 132 utentes com DLC. Os utentes foram recrutados a partir da lista de espera de 16 serviços de Medicina Física e de Reabilitação/Fisioterapia de várias áreas geográficas de Portugal. A QBPDS- VP foi administrada 3 vezes, em 3 momentos de recolha de dados distintos: T0 - momento inicial (utentes em lista de espera); T1 - 1 semana de intervalo (início dos tratamentos de Fisioterapia); e T2 - 6 semanas de intervalo (pós-intervenção de Fisioterapia). Os dados recolhidos em T0 foram utilizados para a análise fatorial e para o estudo da consistência interna (n=132); os dados da amostra emparelhada de T0 e T1 (n=132) para a fiabilidade teste-reteste; e os dados da amostra emparelhada de T0 e T2 (n=120) para a análise do poder de resposta e interpretabilidade. A âncora externa utilizada foi a perceção global de mudança, neste caso a PGIC- VP, que foi respondida em T1 e T2. O nível de significância para o qual os valores se consideraram satisfatórios foi de p≤ 0,05. O tratamento dos dados foi realizado no software IBM SPSS Statistics (versão 20). Resultados: A QBPDS- VP é uma escala unidimensional, que revela uma excelente consistência interna (α de Cronbach= 0,95) e uma fiabilidade teste-reteste satisfatória (CCI= 0,696; IC 95%: 0,581–0,783). Esta escala demonstrou um poder de resposta moderado, quando aplicada em utentes com DLC ( = 0,426 e AAC= 0,741; IC 95%: 0,645 – 0,837). A Diferença Mínima Detetável (DMD) estimada foi de 19 pontos e as estimativas da Diferença Mínima Clinicamente Importante (DMCI) variaram entre 7 (pelo método curva ROC) e 8 pontos (pelo método “diferença média de pontuação”). A estimativa pela curva ROC deriva do ponto ótimo de corte de 6,5 pontos, com Área Abaixo da Curva (AAC)= 0,741, sensibilidade de 72%, e especificidade de 71%. Uma análise complementar da curva ROC baseada nas diferenças de pontuações da QBPDS, expressa em percentagem, revelou um ponto ótimo de corte de - 24% (AAC= 0,737, sensibilidade de 71%, e especificidade de 71%). Para pontuações iniciais da QBPDS- VP mais altas (≥34 pontos), foi encontrado um ponto ótimo de corte de 10,5 pontos (AAC= 0,738, sensibilidade de 73%, e especificidade de 67%). Conclusão: A QBPDS-VP demonstrou bons níveis de fiabilidade e poder de resposta, recomendando-se o seu uso na medição e avaliação da incapacidade funcional de utentes com DLC. A DMD estimada, de 19 pontos, determinou uma amplitude válida da QBPDS-VP de 19 a 81 pontos. Este estudo propõe estimativas de DMCI da QBPDS- VP numa aplicação específica da escala (em utentes com DLC que são referidos para a intervenção de Fisioterapia). A pontuação inicial da QBPDS- VP deve ser considerada na interpretação de mudanças de pontuação, após a intervenção de Fisioterapia.------------ ABSTRACT: One of the main results of physiotherapy interventions for patients with Chronic Low Back Pain (CLBP) is decrease the functional disability. The Quebec Back Pain Disability Scale (QBPDS) is an instrument widely accepted internationally, in measuring the level of disability reported by individuals with CLBP. The purpose of this study is to contribute to the cultural adaptation of the Portuguese version of QBPDS (QBPDS - PV) and investigate the Responsiveness and Interpretability of QBPDS-PV. Methodology: This was a methodological and multicenter study, based on a sample of 132 subjects with CLBP. The patients were recruited from the waiting lists of 16 medicine rehabilitation service, in many Portugal districts. The Quebec Back Pain Disability Scale was administered in three different moments: T0 – baseline (patients in the waiting list); T1- one week after T0 (the beginning of treatment); and T2 – six weeks after T1 (the posttreatment). The data collected at T0 were used for factor analysis and to study the internal consistency (n = 132); paired sample data of T0 and T1 (n=132) were used for test-retest reliability, and sample data paired for T0 and T2 (n=120) used for responsiveness and interpretability analysis. The external anchor was the global perception of change, measured by the Portuguese version of Patient’s Global Impression of Change (PGIC) Scale. The minimal level of significance established was p ≤ 0,05. Data analysis was performed using the IBM SPSS Statistics software (version 20). Results: The QBPDS-PV is a unidimensional scale, demonstrates an excellent internal consistency (Cronbach's α=0.95) and satisfactory test-retest reliability (ICC= 0.696, 95% CI: 0.581–0.783). The scale revealed moderate responsiveness when applied to patients with CLBP ( = 0.426 and AUC= 0.741, 95% CI: 0.645 - 0.837). The Smallest Detectable Change (SDC) was 19 points, whereas the Minimal Clinically Important Change (MCIC) ranged between 7 (ROC curve method) and 8 points (by the "mean difference score"). The estimate was derived from the ROC curve by an optimal cutoff point of 6.5 points, with Area Under the Curve (AUC)= 0.741, sensitivity 72%, and specificity of 71%. A complementary analysis of the ROC curve based on differences in QBPDS scores from baseline, expressed in percentage, revealed an optimal cutoff point of -24% (AUC= 0.737, sensitivity of 71%, and specificity of 71%). For the highest initial scores of QBPDS-PV (≥ 34 points) was found an optimal cutoff of 10.5 points (AUC= 0.738, sensitivity of 73%, and specificity 67%). Conclusion: The QBPDS-PV demonstrated good levels of reliability and responsiveness, being recommended its use in the measurement and evaluation of disability of patients with CLBP. The SDC of 19 points determined the QBPDS‟ scale width of 19 to 81. This study proposes MCIC values for QBPDS –PV for this specific setting (in CLBP patients who are referred for physiotherapy intervention). The QBPDS –PV baseline score have to be taken into account while interpreting the score change after physiotherapy intervention.

Automatically estimating iSAX parameters

The Symbolic Aggregate Approximation (iSAX) is widely used in time series data mining. Its popularity arises from the fact that it largely reduces time series size, it is symbolic, allows lower bounding and is space efficient. However, it requires setting two parameters: the symbolic length and alphabet size, which limits the applicability of the technique. The optimal parameter values are highly application dependent. Typically, they are either set to a fixed value or experimentally probed for the best configuration. In this work we propose an approach to automatically estimate iSAX’s parameters. The approach – AutoiSAX – not only discovers the best parameter setting for each time series in the database, but also finds the alphabet size for each iSAX symbol within the same word. It is based on simple and intuitive ideas from time series complexity and statistics. The technique can be smoothly embedded in existing data mining tasks as an efficient sub-routine. We analyze its impact in visualization interpretability, classification accuracy and motif mining. Our contribution aims to make iSAX a more general approach as it evolves towards a parameter-free method.

Data-driven system identification via evolutionary retrieval of Takagi-Sugeno fuzzy models

Systemidentification, evolutionary automatic, data-driven model, fuzzy Takagi-Sugeno grammar, genotype interpretability, toxicity-prediction

Multi-phase post-mortem CT angiography: development of a standardized protocol.

The objective of this work was to develop an easily applicable technique and a standardized protocol for high-quality post-mortem angiography. This protocol should (1) increase the radiological interpretation by decreasing artifacts due to the perfusion and by reaching a complete filling of the vascular system and (2) ease and standardize the execution of the examination. To this aim, 45 human corpses were investigated by post-mortem computed tomography (CT) angiography using different perfusion protocols, a modified heart-lung machine and a new contrast agent mixture, specifically developed for post-mortem investigations. The quality of the CT angiographies was evaluated radiologically by observing the filling of the vascular system and assessing the interpretability of the resulting images and by comparing radiological diagnoses to conventional autopsy conclusions. Post-mortem angiography yielded satisfactory results provided that the volumes of the injected contrast agent mixture were high enough to completely fill the vascular system. In order to avoid artifacts due to the post-mortem perfusion, a minimum of three angiographic phases and one native scan had to be performed. These findings were taken into account to develop a protocol for quality post-mortem CT angiography that minimizes the risk of radiological misinterpretation. The proposed protocol is easy applicable in a standardized way and yields high-quality radiologically interpretable visualization of the vascular system in post-mortem investigations.

Methods for testing association between uncertain genotypes and quantitative traits.

Interpretability and power of genome-wide association studies can be increased by imputing unobserved genotypes, using a reference panel of individuals genotyped at higher marker density. For many markers, genotypes cannot be imputed with complete certainty, and the uncertainty needs to be taken into account when testing for association with a given phenotype. In this paper, we compare currently available methods for testing association between uncertain genotypes and quantitative traits. We show that some previously described methods offer poor control of the false-positive rate (FPR), and that satisfactory performance of these methods is obtained only by using ad hoc filtering rules or by using a harsh transformation of the trait under study. We propose new methods that are based on exact maximum likelihood estimation and use a mixture model to accommodate nonnormal trait distributions when necessary. The new methods adequately control the FPR and also have equal or better power compared to all previously described methods. We provide a fast software implementation of all the methods studied here; our new method requires computation time of less than one computer-day for a typical genome-wide scan, with 2.5 M single nucleotide polymorphisms and 5000 individuals.

Topographic ERP analyses: a step-by-step tutorial review.

In this tutorial review, we detail both the rationale for as well as the implementation of a set of analyses of surface-recorded event-related potentials (ERPs) that uses the reference-free spatial (i.e. topographic) information available from high-density electrode montages to render statistical information concerning modulations in response strength, latency, and topography both between and within experimental conditions. In these and other ways these topographic analysis methods allow the experimenter to glean additional information and neurophysiologic interpretability beyond what is available from canonical waveform analyses. In this tutorial we present the example of somatosensory evoked potentials (SEPs) in response to stimulation of each hand to illustrate these points. For each step of these analyses, we provide the reader with both a conceptual and mathematical description of how the analysis is carried out, what it yields, and how to interpret its statistical outcome. We show that these topographic analysis methods are intuitive and easy-to-use approaches that can remove much of the guesswork often confronting ERP researchers and also assist in identifying the information contained within high-density ERP datasets.

Auditory-visual multisensory interactions in humans: timing, topography, directionality, and sources.

Current models of brain organization include multisensory interactions at early processing stages and within low-level, including primary, cortices. Embracing this model with regard to auditory-visual (AV) interactions in humans remains problematic. Controversy surrounds the application of an additive model to the analysis of event-related potentials (ERPs), and conventional ERP analysis methods have yielded discordant latencies of effects and permitted limited neurophysiologic interpretability. While hemodynamic imaging and transcranial magnetic stimulation studies provide general support for the above model, the precise timing, superadditive/subadditive directionality, topographic stability, and sources remain unresolved. We recorded ERPs in humans to attended, but task-irrelevant stimuli that did not require an overt motor response, thereby circumventing paradigmatic caveats. We applied novel ERP signal analysis methods to provide details concerning the likely bases of AV interactions. First, nonlinear interactions occur at 60-95 ms after stimulus and are the consequence of topographic, rather than pure strength, modulations in the ERP. AV stimuli engage distinct configurations of intracranial generators, rather than simply modulating the amplitude of unisensory responses. Second, source estimations (and statistical analyses thereof) identified primary visual, primary auditory, and posterior superior temporal regions as mediating these effects. Finally, scalar values of current densities in all of these regions exhibited functionally coupled, subadditive nonlinear effects, a pattern increasingly consistent with the mounting evidence in nonhuman primates. In these ways, we demonstrate how neurophysiologic bases of multisensory interactions can be noninvasively identified in humans, allowing for a synthesis across imaging methods on the one hand and species on the other.