الفهرس | Only 14 pages are availabe for public view |
Abstract The thesis contains several and remarks illustrative examples as an application of our results. The thesis consists of six chapters: Chapter one. Presents the problem statement and its important, motivation behind our approach, and the organization of the thesis. Chapter two. Introduces the most common and traditional data mining predictive modeling schemes in both statistics and computer science communities with data management ,quality, visualizations, outliers identifications, feature selections, supervised and un-supervised learning (generalized linear models, neural networks, support vector machines, Bayesian belief networks, and the proper statistical quality measures techniques in literature).Chapter three. Introduces the brief background about bioinformatics, biomedicine, clinical and pharmaceutical challenge problems, the methodology that attacks such problems, the importance of genomic data in specifying the new style of patient treatment. Chapter four. Presents the most recent innovative data mining new frameworks, such as. extreme learning machines, functional networks, ensemble learning, typeI adaptive neuro-fuzzy systems to overcome the most common challenge drawbacks in the traditional schemes. Chapter five. Presents numerous of bioinformatics with medical and healthcare case studies-real-life challenge problems and how it can be solved using both traditional and new innovative predictive data mining modeling schemes, such as, (i) Protein structure prediction and amino acid sequences with relative solvent accessibility to measure the degree to which each residue is embedded in its solvent environment; (ii) forecasting the translation initiation sites TISS); (iii) prediction of B-cell epitopes; and (iv) solving multi-category disease classification problems. Chapter six. Includes the drawn conclusions and the future open problems. |