A partial dependence factorial analysis to deal with selection bias in observational studis

This thesis presents a creative and practical approach to dealing with the problem of selection bias. Selection bias may be the most important vexing problem in program evaluation or in any line of research that attempts to assert causality. Some of the greatest minds in economics and statistics have scrutinized the problem of selection bias, with the resulting approaches – Rubin’s Potential Outcome Approach(Rosenbaum and Rubin,1983; Rubin, 1991,2001,2004) or Heckman’s Selection model (Heckman, 1979) – being widely accepted and used as the best fixes. These solutions to the bias that arises in particular from self selection are imperfect, and many researchers, when feasible, reserve their strongest causal inference for data from experimental rather than observational studies. The innovative aspect of this thesis is to propose a data transformation that allows measuring and testing in an automatic and multivariate way the presence of selection bias. The approach involves the construction of a multi-dimensional conditional space of the X matrix in which the bias associated with the treatment assignment has been eliminated. Specifically, we propose the use of a partial dependence analysis of the X-space as a tool for investigating the dependence relationship between a set of observable pre-treatment categorical covariates X and a treatment indicator variable T, in order to obtain a measure of bias according to their dependence structure. The measure of selection bias is then expressed in terms of inertia due to the dependence between X and T that has been eliminated. Given the measure of selection bias, we propose a multivariate test of imbalance in order to check if the detected bias is significant, by using the asymptotical distribution of inertia due to T (Estadella et al. 2005) , and by preserving the multivariate nature of data. Further, we propose the use of a clustering procedure as a tool to find groups of comparable units on which estimate local causal effects, and the use of the multivariate test of imbalance as a stopping rule in choosing the best cluster solution set. The method is non parametric, it does not call for modeling the data, based on some underlying theory or assumption about the selection process, but instead it calls for using the existing variability within the data and letting the data to speak. The idea of proposing this multivariate approach to measure selection bias and test balance comes from the consideration that in applied research all aspects of multivariate balance, not represented in the univariate variable- by-variable summaries, are ignored. The first part contains an introduction to evaluation methods as part of public and private decision process and a review of the literature of evaluation methods. The attention is focused on Rubin Potential Outcome Approach, matching methods, and briefly on Heckman’s Selection Model. The second part focuses on some resulting limitations of conventional methods, with particular attention to the problem of how testing in the correct way balancing. The third part contains the original contribution proposed , a simulation study that allows to check the performance of the method for a given dependence setting and an application to a real data set. Finally, we discuss, conclude and explain our future perspectives.

New insights into methodology and interpretation of osteoarthritis. The study of Frassetto identified skeletal collection.

Osteoarthritis (OA) or degenerative joint disease (DJD) is a pathology which affects the synovial joints and characterised by a focal loss of articular cartilage and subsequent bony reaction of the subcondral and marginal bone. Its etiology is best explained by a multifactorial model including: age, sex, genetic and systemic factors, other predisposing diseases and functional stress. In this study the results of the investigation of a modern identified skeletal collection will be presented. In particular, we will focus on the relationship between the presence of OA at various joints. The joint modifications have been analysed using a new methodology that allows the scoring of different degrees of expression of the features considered. Materials and Methods The sample examined comes from the Sassari identified skeletal collection (part of “Frassetto collections”). The individuals were born between 1828 and 1916 and died between 1918 and 1932. Information about sex and age is known for all the individuals. The occupation is known for 173 males and 125 females. Data concerning the occupation of the individuals indicate a preindustrial and rural society. OA has been diagnosed when eburnation (EB) or loss of morphology (LM) were present, or when at least two of the following: marginal lipping (ML), esostosis (EX) or erosion (ER), were present. For each articular surface affected a “mean score” was calculated, reflecting the “severity” of the alterations. A further “score” was calculated for each joint. In the analysis sexes and age classes were always kept separate. For the statistical analyses non parametric test were used. Results The results show there is an increase of OA with age in all the joints analyzed and in particular around 50 years and 60 years. The shoulder, the hip and the knee are the joints mainly affected with ageing while the ankle is the less affected; the correlation values confirm this result. The lesion which show the major correlation with age is the ML. In our sample males are more frequently and more severely affected by OA than females, particularly at the superior limbs, while hip and knee are similarly affected in the two sexes. Lateralization shows some positive results in particular in the right shoulder of males and in various articular surfaces especially of the superior limb of both males and females; articular surfaces and joints are quite always lateralized to the right. Occupational analyses did not show remarkable results probably because of the homogeneity of the sample; males although performing different activities are quite all employed in stressful works. No highest prevalence of knee and hip OA was found in farm-workers respect to the other males. Discussion and Conclusion In this work we propose a methodology to score the different features, necessary to diagnose OA, that allows the investigation of the severity of joint degeneration. This method is easier than the one proposed by Buikstra and Ubelaker (1994), but in the same time allows a quite detailed recording of the features. Epidemiological results can be interpreted quite simply and they are in accordance with other studies; more difficult is the interpretation of the occupational results because many questions concerning the activities performed by the individuals of the collection during their lifespan cannot be solved. Because of this, caution is suggested in the interpretation of bioarcheological specimens. With this work we hope to contribute to the discussion on the puzzling problem of the etiology of OA. The possibility of studying identified skeletons will add important data to the description of osseous features of OA, enriching the medical documentation, based on different criteria. Even if we are aware that the clinical diagnosis is different from the palaeopathological one we think our work will be useful in clarifying some epidemiological as well as pathological aspects of OA.

Towards Rapid Environmental Assessment and Coastal Forecasting in the Northern Adriatic Sea

A new Coastal Rapid Environmental Assessment (CREA) strategy has been developed and successfully applied to the Northern Adriatic Sea. CREA strategy exploits the recent advent of operational oceanography to establish a CREA system based on an operational regional forecasting system and coastal monitoring networks of opportunity. The methodology wishes to initialize a coastal high resolution model, nested within the regional forecasting system, blending the large scale parent model fields with the available coastal observations to generate the requisite field estimates. CREA modeling system consists of a high resolution, O(800m), Adriatic SHELF model (ASHELF) implemented into the Northern Adriatic basin and nested within the Adriatic Forecasting System (AFS) (Oddo et al. 2006). The observational system is composed by the coastal networks established in the framework of ADRICOSM (ADRiatic sea integrated COastal areaS and river basin Managment system) Pilot Project. An assimilation technique exerts a correction of the initial field provided by AFS on the basis of the available observations. The blending of the two data sets has been carried out through a multi-scale optimal interpolation technique developed by Mariano and Brown (1992). Two CREA weekly exercises have been conducted: the first, at the beginning of May (spring experiment); the second in middle August (summer experiment). The weeks have been chosen looking at the availability of all coastal observations in the initialization day and one week later to validate model results, verifying our predictive skills. ASHELF spin up time has been investigated too, through a dedicated experiment, in order to obtain the maximum forecast accuracy within a minimum time. Energetic evaluations show that for the Northern Adriatic Sea and for the forcing applied, a spin-up period of one week allows ASHELF to generate new circulation features enabled by the increased resolution and its total kinetic energy to establish a new dynamical balance. CREA results, evaluated by mean of standard statistics between ASHELF and coastal CTDs, show improvement deriving from the initialization technique and a good model performance in the coastal areas of the Northern Adriatic basin, characterized by a shallow and wide continental shelf subject to substantial freshwater influence from rivers. Results demonstrate the feasibility of our CREA strategy to support coastal zone management and wish an additional establishment of operational coastal monitoring activities to advance it.

A Design Methodology for Computer Security Testing

The field of "computer security" is often considered something in between Art and Science. This is partly due to the lack of widely agreed and standardized methodologies to evaluate the degree of the security of a system. This dissertation intends to contribute to this area by investigating the most common security testing strategies applied nowadays and by proposing an enhanced methodology that may be effectively applied to different threat scenarios with the same degree of effectiveness. Security testing methodologies are the first step towards standardized security evaluation processes and understanding of how the security threats evolve over time. This dissertation analyzes some of the most used identifying differences and commonalities, useful to compare them and assess their quality. The dissertation then proposes a new enhanced methodology built by keeping the best of every analyzed methodology. The designed methodology is tested over different systems with very effective results, which is the main evidence that it could really be applied in practical cases. Most of the dissertation discusses and proves how the presented testing methodology could be applied to such different systems and even to evade security measures by inverting goals and scopes. Real cases are often hard to find in methodology' documents, in contrary this dissertation wants to show real and practical cases offering technical details about how to apply it. Electronic voting systems are the first field test considered, and Pvote and Scantegrity are the two tested electronic voting systems. The usability and effectiveness of the designed methodology for electronic voting systems is proved thanks to this field cases analysis. Furthermore reputation and anti virus engines have also be analyzed with similar results. The dissertation concludes by presenting some general guidelines to build a coordination-based approach of electronic voting systems to improve the security without decreasing the system modularity.

A numerical methodology for the multi-objective optimization of the DI Diesel engine combustion

DI Diesel engine are widely used both for industrial and automotive applications due to their durability and fuel economy. Nonetheless, increasing environmental concerns force that type of engine to comply with increasingly demanding emission limits, so that, it has become mandatory to develop a robust design methodology of the DI Diesel combustion system focused on reduction of soot and NOx simultaneously while maintaining a reasonable fuel economy. In recent years, genetic algorithms and CFD three-dimensional combustion simulations have been successfully applied to that kind of problem. However, combining GAs optimization with actual CFD three-dimensional combustion simulations can be too onerous since a large number of calculations is usually needed for the genetic algorithm to converge, resulting in a high computational cost and, thus, limiting the suitability of this method for industrial processes. In order to make the optimization process less time-consuming, CFD simulations can be more conveniently used to generate a training set for the learning process of an artificial neural network which, once correctly trained, can be used to forecast the engine outputs as a function of the design parameters during a GA optimization performing a so-called virtual optimization. In the current work, a numerical methodology for the multi-objective virtual optimization of the combustion of an automotive DI Diesel engine, which relies on artificial neural networks and genetic algorithms, was developed.

Seismic sequences analysis for estimation of earthquake source parameters: corner frequency, stress drop, and seismic moment observations

The present study has been carried out with the following objectives: i) To investigate the attributes of source parameters of local and regional earthquakes; ii) To estimate, as accurately as possible, M0, fc, Δσ and their standard errors to infer their relationship with source size; iii) To quantify high-frequency earthquake ground motion and to study the source scaling. This work is based on observational data of micro, small and moderate -earthquakes for three selected seismic sequences, namely Parkfield (CA, USA), Maule (Chile) and Ferrara (Italy). For the Parkfield seismic sequence (CA), a data set of 757 (42 clusters) repeating micro-earthquakes (0 ≤ MW ≤ 2), collected using borehole High Resolution Seismic Network (HRSN), have been analyzed and interpreted. We used the coda methodology to compute spectral ratios to obtain accurate values of fc , Δσ, and M0 for three target clusters (San Francisco, Los Angeles, and Hawaii) of our data. We also performed a general regression on peak ground velocities to obtain reliable seismic spectra of all earthquakes. For the Maule seismic sequence, a data set of 172 aftershocks of the 2010 MW 8.8 earthquake (3.7 ≤ MW ≤ 6.2), recorded by more than 100 temporary broadband stations, have been analyzed and interpreted to quantify high-frequency earthquake ground motion in this subduction zone. We completely calibrated the excitation and attenuation of the ground motion in Central Chile. For the Ferrara sequence, we calculated moment tensor solutions for 20 events from MW 5.63 (the largest main event occurred on May 20 2012), down to MW 3.2 by a 1-D velocity model for the crust beneath the Pianura Padana, using all the geophysical and geological information available for the area. The PADANIA model allowed a numerical study on the characteristics of the ground motion in the thick sediments of the flood plain.

Symbolic Dynamics Analysis: a new methodology for foetal heart rate variability analysis

Cardiotocography (CTG) is a widespread foetal diagnostic methods. However, it lacks of objectivity and reproducibility since its dependence on observer's expertise. To overcome these limitations, more objective methods for CTG interpretation have been proposed. In particular, many developed techniques aim to assess the foetal heart rate variability (FHRV). Among them, some methodologies from nonlinear systems theory have been applied to the study of FHRV. All the techniques have proved to be helpful in specific cases. Nevertheless, none of them is more reliable than the others. Therefore, an in-depth study is necessary. The aim of this work is to deepen the FHRV analysis through the Symbolic Dynamics Analysis (SDA), a nonlinear technique already successfully employed for HRV analysis. Thanks to its simplicity of interpretation, it could be a useful tool for clinicians. We performed a literature study involving about 200 references on HRV and FHRV analysis; approximately 100 works were focused on non-linear techniques. Then, in order to compare linear and non-linear methods, we carried out a multiparametric study. 580 antepartum recordings of healthy fetuses were examined. Signals were processed using an updated software for CTG analysis and a new developed software for generating simulated CTG traces. Finally, statistical tests and regression analyses were carried out for estimating relationships among extracted indexes and other clinical information. Results confirm that none of the employed techniques is more reliable than the others. Moreover, in agreement with the literature, each analysis should take into account two relevant parameters, the foetal status and the week of gestation. Regarding the SDA, results show its promising capabilities in FHRV analysis. It allows recognizing foetal status, gestation week and global variability of FHR signals, even better than other methods. Nevertheless, further studies, which should involve even pathological cases, are necessary to establish its reliability.