In this deliberately provocative position paper, we claim that ten years into Linked Data there are still (too?) many unresolved challenges towards arriving at a truly machine-readable and decentralized Web of data. We take a deeper look at the biomedical domain - currently, one of the most promising "adopters" of Linked Data - if we believe the ever-present "LOD cloud" diagram. Herein, we try to highlight and exemplify key technical and non-technical challenges to the success of LOD, and we outline potential solution strategies. We hope that this paper will serve as a discussion basis for a fresh start towards more actionable, truly decentralized Linked Data, and as a call to the community to join forces. / Series: Working Papers on Information Systems, Information Business and Operations

We claim that ten years into Linked Data there are still many unresolved challenges towards arriving at a truly machine-readable and decentralized Web of data. With a focus on the the biomedical domain, currently, one of the most promising adopters of Linked Data, we highlight and exemplify key technical and non-technical challenges to the success of Linked Data, and we outline potential solution strategies.

The condition of the fetus during labour is inferred from the continuous plot of fetal heart rate and uterine contractions (cardiotocogram, CTG). This can be _ difficult to interpret which results in both unnecessary intervention and a failure to intervene when necessary causing potentially preventable neurological damage and mortality. Conventional computing approaches have not been successful in addressing these problems. This is perhaps because the correct interpretation of fetal condition requires physiological knowledge, considerable practical experience and knowledge of the specific patient. The work described in this thesis is concerned with the investigation of artificial intelligence techniques to assist in the interpretation of fetal condition and advise on labour management. A fundamental investigation examined the performance of five types of scalp electrodes for obtaining the fetal electrocardiogram (ECG), from which heart rate is derived, and examined the factors which hamper fetal ECG data acquisition. New methods were developed to classify the important features from the CTG and included an investigation using neural networks. Other CTG features were classified using novel numerical algorithms developed closely with experts. An expert system, guided by a database of rules obtained from experts, was used to process and interpret changes in the CTG features by taking account of patient specific information. This hybrid approach was adopted to improve performance and reliability. After two internal evaluations had found the system obtained a performance comparable with local experts, an extensive external validation was undertaken. This study involved 17 experts from 16 leading centres within the UK. Each expert and the system reviewed 50 cases twice, at least one month apart which contained those considered most difficult to interpret selected from a database of 2400 high risk labours. A novel method was developed to present all the relevant clinical information in a way which approximated the clinical situation. The reviewers scored each 15 minutes of recording according to the concern they had for the fetus and the management they considered appropriate. In this respect, this is the first reported study to examine the performance of expert obstetricians in the management of labour. A new method was derived to measure the agreement between the scores obtained and is applicable to other areas where it is required to measure the similarity between time related sequences. This study found that the experts agreed well and were consistent in their management of the cases. The system was indistinguishable from the experts, except it was more consistent, even when used by an engineer with little knowledge of labour management. This study has shown that expertise in fetal monitoring is achievable in which case the current evidence suggests that this is not being adequately transferred to clinicians. The challenge remains to formulate a method to effectively transfer knowledge to the labour ward and thereby address the real and practical problems which face fetal monitoring today. This study demonstrates that intelligent systems could provide the vehicle to achieve this. I dedicate this work to the memory of my father, Bradley Kenneth Keith with a hope that he always believed it possible. I know he would have had some interesting comments to make and I sadly miss the opportunity of discussing them with him. I also dedicate this work to my mother for always being there, and to my wife Michelle for her unwavering support, patience and most of all her encouragement throughout this work.

abstract: Learning from high dimensional biomedical data attracts lots of attention recently. High dimensional biomedical data often suffer from the curse of dimensionality and have imbalanced class distributions. Both of these features of biomedical data, high dimensionality and imbalanced class distributions, are challenging for traditional machine learning methods and may affect the model performance. In this thesis, I focus on developing learning methods for the high-dimensional imbalanced biomedical data. In the first part, a sparse canonical correlation analysis (CCA) method is presented. The penalty terms is used to control the sparsity of the projection matrices of CCA. The sparse CCA method is then applied to find patterns among biomedical data sets and labels, or to find patterns among different data sources. In the second part, I discuss several learning problems for imbalanced biomedical data. Note that traditional learning systems are often biased when the biomedical data are imbalanced. Therefore, traditional evaluations such as accuracy may be inappropriate for such cases. I then discuss several alternative evaluation criteria to evaluate the learning performance. For imbalanced binary classification problems, I use the undersampling based classifiers ensemble (UEM) strategy to obtain accurate models for both classes of samples. A small sphere and large margin (SSLM) approach is also presented to detect rare abnormal samples from a large number of subjects. In addition, I apply multiple feature selection and clustering methods to deal with high-dimensional data and data with highly correlated features. Experiments on high-dimensional imbalanced biomedical data are presented which illustrate the effectiveness and efficiency of my methods. / Dissertation/Thesis / M.S. Computer Science 2013

<p>The main objective of the reported study was to investigate how biomedical ontologies, logically structured representations of various aspects of the biomedical reality, can help researchers in analyzing experimental data. The dissertation reports two attempts to construct tools for the analysis of high-throughput experimental results using explicit domain knowledge representations. Furthermore, integrative efforts made by the community of Open Biomedical Ontologies (OBO), in which the author has participated, are reported, and a framework for consistently connecting the Gene Ontology (GO) with the Taxonomy of Species is proposed and discussed.</p>

The main objective of the reported study was to investigate how biomedical ontologies, logically structured representations of various aspects of the biomedical reality, can help researchers in analyzing experimental data. The dissertation reports two attempts to construct tools for the analysis of high-throughput experimental results using explicit domain knowledge representations. Furthermore, integrative efforts made by the community of Open Biomedical Ontologies (OBO), in which the author has participated, are reported, and a framework for consistently connecting the Gene Ontology (GO) with the Taxonomy of Species is proposed and discussed.

>Magister Scientiae - MSc / Headline reports on data breaches worldwide have resulted in heightened concerns about information security vulnerability. In Africa, South Africa is ranked among the top ‘at-risk’ countries with information security vulnerabilities and is the most the most cybercrime-targeted country. Globally, such cyber vulnerability incidents greatly affect the education sector, due, in part, to the fact that it holds more Personal Identifiable Information (PII) than other sectors. PII refers to (but is not limited to) ID numbers, financial account numbers, and biomedical research data. In response to rising threats, South Africa has implemented a regulation called the Protection of Personal Information Act (POPIA), similar to the European Union General Data Protection Regulation (GDPR), which seeks to mitigate cybercrime and information security vulnerabilities. The extent to which African institutions, especially in South Africa, have embraced and responded to these two information security regulations remains vague, making it a crucial matter for biomedical researchers. This study aimed to assess whether the participating universities have proper and reliable information security practices, measures and management in place and whether they fall in line with both national (POPIA) and international (GDPR) regulations. In order to achieve this aim, the study undertook a qualitative exploratory analysis of information security management across three universities in South Africa. A Technology, Organizational, and Environmental (TOE) model was employed to investigate factors that may influence effective information security measures. A Purposeful sampling method was employed to interview participants from each university. From the technological standpoint, Bring Your Own Device (BYOD) policy, whereby on average, a student owns and connects between three to four internet-enabled devices to the network, has created difficulties for IT teams, particularly in the areas of authentication, explosive growth in bandwidth, and access control to security university servers. In order to develop robust solutions to mitigate these concerns, and which are not perceived by users as overly prohibitive, executive management should acknowledge that security and privacy issues are a universal problem and not solely an IT problem and equip the IT teams with the necessary tools and mechanisms to allow them to overcome commonplace challenges. At an organisational level, information security awareness training of all users within the university setting was identified as a key factor in protecting the integrity, confidentiality, and availability of information in highly networked environments. Furthermore, the University’s information security mission must not simply be a link on a website, it should be constantly re-enforced by informing users during, and after, the awareness training. In terms of environmental factors, specifically the GDPR and POPIA legislations, one of the most practical and cost-effective ways universities can achieve data compliance requirements is to help staff (both teaching and non-teaching), students, and other employees understand the business value of all information. Users which are more aware of sensitivity of data, risks to the data, and their responsibilities when handling, storing, processing, and distributing data during their day to day activities will behave in a manner that would makes compliance easier at the institutional level. Results obtained in this study helped to elucidate the current status, issues, and challenges which universities are facing in the area of information security management and compliance, particularly in the South African context. Findings from this study point to organizational factors being the most critical when compared to the technological and environmental contexts examined. Furthermore, several proposed information security policies were developed with a view to assist biomedical practitioners within the institutional setting in protecting sensitive biomedical data.

>Magister Scientiae - MSc / Headline reports on data breaches worldwide have resulted in heightened concerns about information security vulnerability. In Africa, South Africa is ranked among the top ‘at-risk’ countries with information security vulnerabilities and is the most cybercrime-targeted country. Globally, such cyber vulnerability incidents greatly affect the education sector, due, in part, to the fact that it holds more Personal Identifiable Information (PII) than other sectors. PII refers to (but is not limited to) ID numbers, financial account numbers, and biomedical research data.

Philosophiae Doctor - PhD / Electrochemiluminiscence, (ECL), is an electrochemically-induced process that leads to the generation of measurable luminescent signals at the electrode surface. The luminescent signals occur when electrochemically generated intermediates undergo a highly exergonic reaction to produce an electronically excited state that then emits light. Immobilization of the ECL luminophore on an electrode surface provides enhancement of ECL intensity. This work presents results of the feasibility study focused on the application of novel luminophores for electrochemiluminescence (ECL) sensors. The thesis mainly focuses on studying the ECL of polyanilinorutheniumbipyridyl- imidazo phenanthroline and carboxydifluoroboradiazaindancence luminophores. The influence of the synthetic methods on the electrochemical, structural and photophysical properties of poly(8-anilino-lnaphthalene sulphonic acid) (P ANSA) synthesized by electropolymerization (PANSA) and chemical polymerization (PANSA) were studied. Cyclic voltammetry (CV) data revealed that the electrogenerated PANSA contains species of mixed redox states; with evidence of the presence of penigraniline, emeraldine and leucoemeraldine forms of PANSA. In contrast, the CV of PANSA indicated that it is predominantly in the emeraldine form with a reduction potential at approximately + 0.2 V. The presence of emeraldine moiety in PANSA was confirmed from UV-Vis spectroscopy data. A band gap energy value of 2.5 eV was calculated for the emeraldine in PANSA from the UV data. Comparative study of the charge transfer coefficient, DCT, of the two types of PANSA indicated moderate charge propagation in PANSA (DCT = 1.68 ± 0.1 x 10-8 cm2 s-') which was order of magnitude lower than for PANSA (DCT = 1.68 ± 0.3 x 10-7 cm2 s-'). The differences in the structural properties of the two polymers were reflected in their IR spectra, with evidence of C=C and C=N stretching vibrations observed at 2030, 2158 and 2486 cm-I in PANSA, which are absent in PANSA. The mode of synthesis had a modest impact on the photophysics of the polymers, for example PANSA exhibited a luminescent lifetime of9.00 ± 0.05 ns compared with 11.5 ± 0.07 ns for PANSA. However, time resolved emission anisotropy studies gave a rotational correlation time, p, of 13.8 ± 2.47 ns for PANSA compared to 0.633 ± 0.03 ns for its chemically generated analogue. This suggests a much shorter chain length in the PANSA molecule and higher cross-linking or aggregation in PANSA that can limit incorporation of ruthenium complex on the polymer backbone. As a result, electrochemiluminescent films have been formed by electrodepositing polyaniline, PANI, films in the presence of [Ru(bpY)2PIC]2+; bpy is 2,2'-bipyridyl and PIC is (2,2'-bipyridyl)-2( 4- carboxylphenyl) imidazo [4,5 ][ 1,10] phenanthroline in this work. The homogeneous charge transport diffusion coefficient, DCT, for the Ru2+/3+couple within the PANI film is 2.6 ± 0.9 x 10-10 cm2s-l. The large DCT facilitates a fast regeneration of Ru3+and, coupled to a relatively rigid microenvironment, results in a high ECL intensity in the presence oftripropylamine as co-reactant compared to [Ru(bpY)3f+. Significantly, despite the conducting nature of the polymer backbone, the [Ru(bpy)2PICH2]2+ loaded PANI has the highest efficiency, 1.00%, yet reported for a surface confined ruthenium complex. PANI-Ru complex showed to have many properties that make it an ideal luminophore for sensitive and selective analytical methods; however, it would be useful to have other ECL labels that can span a wide range of wavelengths so that simultaneous determination of several analytes in a single sample can be investigated. In this case, the photophysics, electrochemical and electrochemiluminescent properties of a novel 1,3,5,7 -tetramethyl-8-[ (2-fluorophenyl)-6-methoxy-l ,5-naphthyridine-3-carboxy ]-4,4'difluoroboradiazaindace-ne BODIPY -COOH, dye were demonstrated in this work. The photophysics studies revealed that BODIPY -COOH is highly luminescent: exhibiting sharp absorbance bands, intense emission bands and high emission quantum yield. The quantum yield proved to be solvent dependent and was determined to be 0.88 ± 0.02 and 0.60 ± 0.04 in dimethylsulphoxide (DMSO) and acetonitrile (MeCN), respectively. Electrochemiluminescence (ECL) of BODIPYCOOH in solution was generated in the presence of either benzoyl peroxide (BPO) or hydrogen peroxide. The ECL turn-on potential in the presence of BPO was observed at potentials that are greater than - 1.5 V, and when H202 was used the ECL turn-on potential was significantly fine-tuned to less negative potential of - 0.4 V. Electrochemiluminescent thin films of BODIPY -COOH on Pt electrodes exhibited luminescence properties similar to those of the free dye in solution. However, the solution based approach ECL has its own limitations such as loss of signal due to the diffusion of the ECL reagent out of the detection zone. To overcome loss of signal effects, the introduction of cysteamine and cysteine linkers to the BODIPY dye were employed. It was seen that self-quenching was not sufficient to interfere significantly with the film ECL emission properties and thus the BODIPY thin film can be used in ECL applications. Interestingly, the BODIPY film exhibited the strongest luminescence in water and this is potentially useful in ECL application in biological media.

No description available.