Abstract
The functions of proteins in living organisms are related to their 3-D structure, which is known to be ultimately determined by their linear sequence of amino acids that together form these macromolecules. It is, therefore, of great importance to be able to understand and predict how the protein 3D-structure arises from a particular linear sequence of amino acids. In this paper we report the application of Machine Learning methods to predict, with high values of accuracy, the secondary structure of proteins, namely alpha-helices and beta-sheets, which are intermediate levels of the local structure.

Abstract
"Traditional" clustering, in broad sense, aims at organizing objects into groups (clusters) whose members are "similar" among them and are "dissimilar" to objects belonging to the other groups. In contrast, in conceptual clustering the underlying structure of the data together with the description language which is available to the learner is what drives cluster formation, thus providing intelligible descriptions of the clusters, facilitating their interpretation. We present a novel conceptual clustering system for multi-relational data, based on the popular k?-?medoids algorithm. Although clustering is, generally, not straightforward to evaluate, experimental results on several applications show promising results. Clusters generated without class information agree very well with the true class labels of cluster's members. Moreover, it was possible to obtain intelligible and meaningful descriptions of the clusters. © 2012 Springer-Verlag Berlin Heidelberg.

Abstract
Knowledge discovery techniques try to extract patterns and concepts from raw data, and clustering certainly is one of the most popular processes in this research field. However, nowadays data is being produced in streaming fashion and distributed locations, turning most classical methods obsolete. This thesis addresses two different clustering problems in ubiquitous and streaming scenarios, presenting evidence of the advantages produced by applying distributed and streaming machine learning algorithms, and proposing new ones to solve the addressed problems.

Abstract
Although IT governance practices (like ITIL, which recommends on the use of Audit Trails (AT) for proper service level management) are being introduced in many Hospitals to cope with increasing levels of information quality and safety requirements, the standard maturity levels of hospital IT departments is still not enough to reach the level of frequent use of AT. This paper aims to address the issues related to the existence and quality of AT in Hospital Information Systems (HIS). A total of 5 hospitals participated in the study (4 CIO interviews and 7 AT from 4 hospitals). Very few AT are known to exist in these hospitals (about 1 per hospital in an estimate of 21 existing IS). The existing AT has poor quality (garbled, incomplete and misformatted information). CIOs should be much more concerned with the existence, quality and maintenance of AT. Recommendations include server clock synchronization, using advanced AT visualization tools, using also real-time auditing tools, and periodically audit the access to all data of a random patient.

Abstract
Currently, in radiotherapy (RT) departments, there are different manufacturers and stand-alone information systems (IS) for single-purpose applications. These systems have most of the data distributed through different IS. The DICOM-RT extension has six objects that provide a standardized way of transferring the information circulating in the external beam RT. The aim of this study is to assess expert's opinion about DICOM-RT and IS interoperability in the RT context, through the characterization of Portuguese RT facilities, in terms of equipment and IS with the identification of existing interoperability problems. This study is cross-sectional, and the preliminary results presented in this paper are relative to the period May-July 2011. All Portuguese RT departments (i.e. 20) were invited to cooperate in the survey; the response rate was 40% (n=8), while 10% (n=2) of the institutions did not authorize the RT departments to participate. The preliminary results show that the RT departments have some equipment and IS from different vendors contributing for heterogeneity of RT workflows. The experts somehow attribute importance to interoperability, but have low knowledge about their own IS integrations, and DICOM-RT. Compliance with DICOM-RT is recommended when acquiring new RT IS to optimise the interoperability.

Abstract
Mammographic databases play an important role in the development of algorithms aiming to improve Computer-Aided Detection and Diagnosis systems (CAD). However, these often do not take into consideration all the requirements needed for a proper study, previously discussed at the Biomedical Image Processing Meeting in 1993. Case selection and annotation requirements are the most commonly referenced in literature, when describing a database used for the development of such algorithms. This work aims to assess the compliance and suitability of case selection and annotation requirements in the publicly available mammographic databases for development and optimization of CADs. A literature review has been made, applying proper selection criteria related to the research question. In the literature, we found citations to 3 publicly available mammographic databases and ten having restricted access. Through the analysis of the results attained, we noticed that none of the two requirements previously described is on its way to be fully complied in mammographic databases. We can conclude that researchers need a database that fulfils all the mentioned requirements in order to develop efficacious and effective CAD systems. We also believe that the requirements, discussed in 1993, need to be reviewed and updated. New paradigms and ideas to increase algorithms' performance are needed in order to improve CAD schemes.