Visualization and validation of (Q)SAR models

Analyzing and modeling relationships between the structure of chemical compounds, their physico-chemical properties, and biological or toxic effects in chemical datasets is a challenging task for scientific researchers in the field of cheminformatics. Therefore, (Q)SAR model validation is essential to ensure future model predictivity on unseen compounds. Proper validation is also one of the requirements of regulatory authorities in order to approve its use in real-world scenarios as an alternative testing method. However, at the same time, the question of how to validate a (Q)SAR model is still under discussion. In this work, we empirically compare a k-fold cross-validation with external test set validation. The introduced workflow allows to apply the built and validated models to large amounts of unseen data, and to compare the performance of the different validation approaches. Our experimental results indicate that cross-validation produces (Q)SAR models with higher predictivity than external test set validation and reduces the variance of the results. Statistical validation is important to evaluate the performance of (Q)SAR models, but does not support the user in better understanding the properties of the model or the underlying correlations. We present the 3D molecular viewer CheS-Mapper (Chemical Space Mapper) that arranges compounds in 3D space, such that their spatial proximity reflects their similarity. The user can indirectly determine similarity, by selecting which features to employ in the process. The tool can use and calculate different kinds of features, like structural fragments as well as quantitative chemical descriptors. Comprehensive functionalities including clustering, alignment of compounds according to their 3D structure, and feature highlighting aid the chemist to better understand patterns and regularities and relate the observations to established scientific knowledge. Even though visualization tools for analyzing (Q)SAR information in small molecule datasets exist, integrated visualization methods that allows for the investigation of model validation results are still lacking. We propose visual validation, as an approach for the graphical inspection of (Q)SAR model validation results. New functionalities in CheS-Mapper 2.0 facilitate the analysis of (Q)SAR information and allow the visual validation of (Q)SAR models. The tool enables the comparison of model predictions to the actual activity in feature space. Our approach reveals if the endpoint is modeled too specific or too generic and highlights common properties of misclassified compounds. Moreover, the researcher can use CheS-Mapper to inspect how the (Q)SAR model predicts activity cliffs. The CheS-Mapper software is freely available at http://ches-mapper.org.

Investigating strategies and indicators for sustainable campus: Comparison and synergies between University of Bologna and Delft University of Technology

Sustainable development is one of the biggest challenges of the twenty fist-century. Various university has begun the debate about the content of this concept and the ways in which to integrate it into their policy, organization and activities. Universities have a special responsibility to take over a leading position by demonstrating best practices that sustain and educate a sustainable society. For that reason universities have the opportunity to create the culture of sustainability for today’s student, and to set their expectations for how the world should be. This thesis aim at analyzing how Delft University of Technology and University of Bologna face the challenge of becoming a sustainable campus. In this context, both universities have been studied and analyzed following the International Sustainable Campus Network (ISCN) methodology that provides a common framework to formalize commitments and goals at campus level. In particular this work has been aimed to highlight which key performance indicators are essential to reach sustainability as a consequence the following aspects has been taken into consideration: energy use, water use, solid waste and recycling, carbon emission. Subsequently, in order to provide a better understanding of the current state of sustainability on University of Bologna and Delft University of Technology, and potential strategies to achieve the stated objective, a SWOT Analysis has been undertaken. Strengths, weaknesses, opportunities and threats have been shown to understand how the two universities can implement a synergy to improve each other. In the direction of framing a “Sustainable SWOT” has been considered the model proposed by People and Planet, so it has been necessary to evaluate important matters as for instance policy, investment, management, education and engagement. Regarding this, it has been fundamental to involve the main sustainability coordinators of the two universities, this has been achieved through a brainstorming session. Partnerships are key to the achievement of sustainability. The creation of a bridge between two universities aims to join forces and to create a new generation of talent. As a result, people can become able to support universities in the exchange of information, ideas, and best practices for achieving sustainable campus operations and integrating sustainability in research and teaching. For this purpose the project "SUCCESS" has been presented, the project aims to create an interactive European campus network that can be considered a strategic key player for sustainable campus innovation in Europe. Specifically, the main key performance indicators have been analyzed and the importance they have for the two universities and their strategic impact have been highlighted. For this reason, a survey was conducted with people who play crucial roles for sustainability within the two universities and they were asked to evaluate the KPIs of the project. This assessment has been relevant because has represented the foundation to develop a strategy to create a true collaboration.

European and Chinese legislation regulating the portrayal of women in television advertising: A comparison

This dissertation has two main purposes. On the one hand, it aims at comparing the gender stereotypes presented in the television commercials in China and in Europe. Considering the cultural, historical and socio-economical differences between these two contexts, it is interesting to examine the gender role models offered and used by the advertising industry in European Union and China in order to see if the gender stereotypes are similar and to evaluate to which extent they reflect, challenge or reinforce the gender roles of the society where they are broadcasted. On the other hand, the objective of this dissertation is to establish the degree of adequateness and effectiveness of the existing regulatory framework through an analysis of the positive and negative aspects of the regulatory acts issued to safeguard a fair representation of genders in the EU Member States and in China.

Deep learning for computer vision: a comparison between convolutional neural networks and hierarchical temporal memories on object recognition tasks

In recent years, Deep Learning techniques have shown to perform well on a large variety of problems both in Computer Vision and Natural Language Processing, reaching and often surpassing the state of the art on many tasks. The rise of deep learning is also revolutionizing the entire field of Machine Learning and Pattern Recognition pushing forward the concepts of automatic feature extraction and unsupervised learning in general. However, despite the strong success both in science and business, deep learning has its own limitations. It is often questioned if such techniques are only some kind of brute-force statistical approaches and if they can only work in the context of High Performance Computing with tons of data. Another important question is whether they are really biologically inspired, as claimed in certain cases, and if they can scale well in terms of "intelligence". The dissertation is focused on trying to answer these key questions in the context of Computer Vision and, in particular, Object Recognition, a task that has been heavily revolutionized by recent advances in the field. Practically speaking, these answers are based on an exhaustive comparison between two, very different, deep learning techniques on the aforementioned task: Convolutional Neural Network (CNN) and Hierarchical Temporal memory (HTM). They stand for two different approaches and points of view within the big hat of deep learning and are the best choices to understand and point out strengths and weaknesses of each of them. CNN is considered one of the most classic and powerful supervised methods used today in machine learning and pattern recognition, especially in object recognition. CNNs are well received and accepted by the scientific community and are already deployed in large corporation like Google and Facebook for solving face recognition and image auto-tagging problems. HTM, on the other hand, is known as a new emerging paradigm and a new meanly-unsupervised method, that is more biologically inspired. It tries to gain more insights from the computational neuroscience community in order to incorporate concepts like time, context and attention during the learning process which are typical of the human brain. In the end, the thesis is supposed to prove that in certain cases, with a lower quantity of data, HTM can outperform CNN.

Dynamic characterisation of four nine-story large-panel R.C. buildings in Bishkek (Kyrgyzstan): A comparison between experimental ambient vibration analysis and numerical finite element modeling

With the outlook of improving seismic vulnerability assessment for the city of Bishkek (Kyrgyzstan), the global dynamic behaviour of four nine-storey r.c. large-panel buildings in elastic regime is studied. The four buildings were built during the Soviet era within a serial production system. Since they all belong to the same series, they have very similar geometries both in plan and in height. Firstly, ambient vibration measurements are performed in the four buildings. The data analysis composed of discrete Fourier transform, modal analysis (frequency domain decomposition) and deconvolution interferometry, yields the modal characteristics and an estimate of the linear impulse response function for the structures of the four buildings. Then, finite element models are set up for all four buildings and the results of the numerical modal analysis are compared with the experimental ones. The numerical models are finally calibrated considering the first three global modes and their results match the experimental ones with an error of less then 20%.

A comparison between 2D and 3D cephalometry on CBCT scans of human skulls

The purpose of this study was to evaluate whether measurements on conventional cephalometric radiographs are comparable with 3D measurements on 3D models of human skulls, derived from cone beam CT (CBCT) data. A CBCT scan and a conventional cephalometric radiograph were made of 40 dry skulls. Standard cephalometric software was used to identify landmarks on both the 2D images and the 3D models. The same operator identified 17 landmarks on the cephalometric radiographs and on the 3D models. All images and 3D models were traced five times with a time-interval of 1 week and the mean value of repeated measurements was used for further statistical analysis. Distances and angles were calculated. Intra-observer reliability was good for all measurements. The reproducibility of the measurements on the conventional cephalometric radiographs was higher compared with the reproducibility of measurements on the 3D models. For a few measurements a clinically relevant difference between measurements on conventional cephalometric radiographs and 3D models was found. Measurements on conventional cephalometric radiographs can differ significantly from measurements on 3D models of the same skull. The authors recommend that 3D tracings for longitudinal research are not used in cases were there are only 2D records from the past.

Sonographic visualization and ultrasound-guided blockade of the greater occipital nerve: a comparison of two selective techniques confirmed by anatomical dissection

BACKGROUND: Local anaesthetic blocks of the greater occipital nerve (GON) are frequently performed in different types of headache, but no selective approaches exist. Our cadaver study compares the sonographic visibility of the nerve and the accuracy and specificity of ultrasound-guided injections at two different sites. METHODS: After sonographic measurements in 10 embalmed cadavers, 20 ultrasound-guided injections of the GON were performed with 0.1 ml of dye at the classical site (superior nuchal line) followed by 20 at a newly described site more proximal (C2, superficial to the obliquus capitis inferior muscle). The spread of dye and coloration of nerve were evaluated by dissection. RESULTS: The median sonographic diameter of the GON was 4.2 x 1.4 mm at the classical and 4.0 x 1.8 mm at the new site. The nerves were found at a median depth of 8 and 17.5 mm, respectively. In 16 of 20 in the classical approach and 20 of 20 in the new approach, the nerve was successfully coloured with the dye. This corresponds to a block success rate of 80% (95% confidence interval: 58-93%) vs 100% (95% confidence interval: 86-100%), which is statistically significant (McNemar's test, P=0.002). CONCLUSIONS: Our findings confirm that the GON can be visualized using ultrasound both at the level of the superior nuchal line and C2. This newly described approach superficial to the obliquus capitis inferior muscle has a higher success rate and should allow a more precise blockade of the nerve.

Comparison of actual surgical outcomes and 3-dimensional surgical simulations

PURPOSE: The advent of imaging software programs has proved to be useful for diagnosis, treatment planning, and outcome measurement, but precision of 3-dimensional (3D) surgical simulation still needs to be tested. This study was conducted to determine whether the virtual surgery performed on 3D models constructed from cone-beam computed tomography (CBCT) can correctly simulate the actual surgical outcome and to validate the ability of this emerging technology to recreate the orthognathic surgery hard tissue movements in 3 translational and 3 rotational planes of space. MATERIALS AND METHODS: Construction of pre- and postsurgery 3D models from CBCTs of 14 patients who had combined maxillary advancement and mandibular setback surgery and 6 patients who had 1-piece maxillary advancement surgery was performed. The postsurgery and virtually simulated surgery 3D models were registered at the cranial base to quantify differences between simulated and actual surgery models. Hotelling t tests were used to assess the differences between simulated and actual surgical outcomes. RESULTS: For all anatomic regions of interest, there was no statistically significant difference between the simulated and the actual surgical models. The right lateral ramus was the only region that showed a statistically significant, but small difference when comparing 2- and 1-jaw surgeries. CONCLUSIONS: Virtual surgical methods were reliably reproduced. Oral surgery residents could benefit from virtual surgical training. Computer simulation has the potential to increase predictability in the operating room.

Comparison of three optical tracking systems in a complex navigation scenario

Three-dimensional rotational X-ray imaging with the SIREMOBIL Iso-C3D (Siemens AG, Medical Solutions, Erlangen, Germany) has become a well-established intra-operative imaging modality. In combination with a tracking system, the Iso-C3D provides inherently registered image volumes ready for direct navigation. This is achieved by means of a pre-calibration procedure. The aim of this study was to investigate the influence of the tracking system used on the overall navigation accuracy of direct Iso-C3D navigation. Three models of tracking system were used in the study: Two Optotrak 3020s, a Polaris P4 and a Polaris Spectra system, with both Polaris systems being in the passive operation mode. The evaluation was carried out at two different sites using two Iso-C3D devices. To measure the navigation accuracy, a number of phantom experiments were conducted using an acrylic phantom equipped with titanium spheres. After scanning, a special pointer was used to pinpoint these markers. The difference between the digitized and navigated positions served as the accuracy measure. Up to 20 phantom scans were performed for each tracking system. The average accuracy measured was 0.86 mm and 0.96 mm for the two Optotrak 3020 systems, 1.15 mm for the Polaris P4, and 1.04 mm for the Polaris Spectra system. For the Polaris systems a higher maximal error was found, but all three systems yielded similar minimal errors. On average, all tracking systems used in this study could deliver similar navigation accuracy. The passive Polaris system showed ? as expected ? higher maximal errors; however, depending on the application constraints, this might be negligible.

Patients with substance use and personality disorders: a comparison of patient characteristics, treatment process, and outcomes in Swiss and U.S. substance use disorder programs

Cross-cultural comparisons may increase our understanding of different models of substance use treatment and help identify consistent associations between patients' characteristics, treatment conditions, and outcomes.

Mesh-based vs. Image-based Statistical Appearance Model of the Human Femur: a Preliminary Comparison Study for the Creation of Finite Element Meshes

Statistical models have been recently introduced in computational orthopaedics to investigate the bone mechanical properties across several populations. A fundamental aspect for the construction of statistical models concerns the establishment of accurate anatomical correspondences among the objects of the training dataset. Various methods have been proposed to solve this problem such as mesh morphing or image registration algorithms. The objective of this study is to compare a mesh-based and an image-based statistical appearance model approaches for the creation of nite element(FE) meshes. A computer tomography (CT) dataset of 157 human left femurs was used for the comparison. For each approach, 30 finite element meshes were generated with the models. The quality of the obtained FE meshes was evaluated in terms of volume, size and shape of the elements. Results showed that the quality of the meshes obtained with the image-based approach was higher than the quality of the mesh-based approach. Future studies are required to evaluate the impact of this finding on the final mechanical simulations.

Evaluative comparison of physician-based vs. patient-based outcomes in posterior lumbar fusion

Posterior lumbar fusion is a frequently performed procedure in spinal surgery. High percentages of good and excellent results are indicated by physicians. On the other hand patient-based outcomes are reported. Little is known about the correlations of these two assessment types. We aimed at their comparison. The analysis included 1013 patients with degenerative spinal disease or spondylolisthesis from an international spine registry, treated with posterior lumbar fusion. All patients were pre/postop assessed by physician-based McNab criteria (‘excellent’, ‘good’, ‘fair’, ‘poor’). Of these patients, 210 (mean age 61 years; 57% females) were in addition assessed by patient-based Oswestry Disability Index (ODI). The remaining 803 patients (mean age 59 years; 56% females) were assessed by patient-based Core Outcome Measure Index (COMI), including Visual Analogue Scale (VAS) for back and leg pain as well as verbal self-rating (‘helped a lot’, ‘helped’, ‘helped only little’, ‘didn’t help’, ‘made things worse’). McNab criteria were compared to the Minimal Clinically Important Difference (MCID) in ODI (12.8), in VAS back (1.2) and leg pain (1.6). We investigated the correlations between McNab criteria and these patient-based outcomes. In the ‘excellent’ group as rated by physicians, the proposed MCID was reached in 83% of patients for ODI, in 69% for VAS back and in 83% for VAS leg pain. All patients said the treatment had ‘helped’ or ‘helped a lot’. In the ‘good’ group 56% (ODI), 66% (back pain) and 86% (leg pain) reached the MCID. 96% of patients perceived the treatment as positive. In the ‘fair’ group 37% (ODI), 55% (back pain) and 63% (leg pain) reached the MCID. 49% had positive treatment considerations. The ‘poor’ group revealed 30% (ODI), 35% (back pain) and 44% (leg pain) of patients with reached MCID. Only 15% rated the treatment as positive. The Spearman correlation coefficients between McNab criteria on the one hand and ODI, back and leg pain as well as patients’ verbal self-rating on the other hand were 0.57, 0.37, 0.36 and 0.46 respectively. The comparison of physician and patient-based outcomes showed the highest correlations between McNab criteria and ODI, somewhat weaker correlations with patients’ self-rating and the weakest correlations with back and leg pain. Based on these findings, physicians’ evaluation of patient outcomes can be considered a valuable part of patient assessment, corresponding very well with patients’ perceptions of success or failure of spinal surgery.

A comparison of the original and simplified Pulmonary Embolism Severity Index

The Pulmonary Embolism Severity Index (PESI) is a validated clinical prognostic model for patients with pulmonary embolism (PE). Recently, a simplified version of the PESI was developed. We sought to compare the prognostic performance of the original and simplified PESI. Using data from 15,531 patients with PE, we compared the proportions of patients classified as low versus higher risk between the original and simplified PESI and estimated 30-day mortality within each risk group. To assess the models' accuracy to predict mortality, we calculated sensitivity, specificity, and predictive values and likelihood ratios for low- versus higher-risk patients. We also compared the models' discriminative power by calculating the area under the receiver-operating characteristic curve. The overall 30-day mortality was 9.3%. The original PESI classified a significantly greater proportion of patients as low-risk than the simplified PESI (40.9% vs. 36.8%; p<0.001). Low-risk patients based on the original and simplified PESI had a mortality of 2.3% and 2.7%, respectively. The original and simplified PESI had similar sensitivities (90% vs. 89%), negative predictive values (98% vs. 97%), and negative likelihood ratios (0.23 vs. 0.28) for predicting mortality. The original PESI had a significantly greater discriminatory power than the simplified PESI (area under the ROC curve 0.78 [95% CI: 0.77-0.79] vs. 0.72 [95% CI: 0.71-0.74]; p<0.001). In conclusion, even though the simplified PESI accurately identified patients at low-risk of adverse outcomes, the original PESI classified a higher proportion of patients as low-risk and had a greater discriminatory power than the simplified PESI.

A strategy combining imaging and laboratory biomarkers in comparison with a simplified clinical score for risk stratification of patients with acute pulmonary embolism

This study aimed to assess the performance of two prognostic models-the European Society of Cardiology (ESC) model and the simplified Pulmonary Embolism Severity Index (sPESI)-in predicting short-term mortality in patients with pulmonary embolism (PE).

How valuable are multiple treatment comparison methods in evidence-based health-care evaluation?

Objectives To compare the use of pair-wise meta-analysis methods to multiple treatment comparison (MTC) methods for evidence-based health-care evaluation to estimate the effectiveness and cost-effectiveness of alternative health-care interventions based on the available evidence. Methods Pair-wise meta-analysis and more complex evidence syntheses, incorporating an MTC component, are applied to three examples: 1) clinical effectiveness of interventions for preventing strokes in people with atrial fibrillation; 2) clinical and cost-effectiveness of using drug-eluting stents in percutaneous coronary intervention in patients with coronary artery disease; and 3) clinical and cost-effectiveness of using neuraminidase inhibitors in the treatment of influenza. We compare the two synthesis approaches with respect to the assumptions made, empirical estimates produced, and conclusions drawn. Results The difference between point estimates of effectiveness produced by the pair-wise and MTC approaches was generally unpredictable—sometimes agreeing closely whereas in other instances differing considerably. In all three examples, the MTC approach allowed the inclusion of randomized controlled trial evidence ignored in the pair-wise meta-analysis approach. This generally increased the precision of the effectiveness estimates from the MTC model. Conclusions The MTC approach to synthesis allows the evidence base on clinical effectiveness to be treated as a coherent whole, include more data, and sometimes relax the assumptions made in the pair-wise approaches. However, MTC models are necessarily more complex than those developed for pair-wise meta-analysis and thus could be seen as less transparent. Therefore, it is important that model details and the assumptions made are carefully reported alongside the results.