Recently, I have reviewed a book, and my book review was published in World Scientific website: http://www.worldscientific.com/worldscibooks/10.1142/7923
The book information is below:
A Gentle Introduction to Support Vector Machines in Biomedicine Volume 2: Case Studies and Benchmarks
Alexander Statnikov, Constantin F Aliferis, Douglas P Hardin, and Isabelle Guyon
World Scientific, 2013; Hard Cover, Edition 1, $49.29, pp. 201.
Here is my review about the book:
“A Gentle Introduction to Support Vector Machines in Biomedicine, Volume 2: Case Studies and Benchmarks” branches out to readers to illustrate the use of a support vector machine (SVM) algorithm—various biomedicine applications—that require advanced machine learning approaches. As its title implies, the book is intended to “gently” introduce SVM to the field of biomedicine. While the authors built upon the essential SVM principles, theory, and algorithms in the first volume of the two-volume book series, the second volume focuses on the practical use and depth of SVM in biomedicine applications and its comparison with other machine learning approaches.
Support vector machine is a high performance machine learning algorithm that allows one to capture complex multivariate relationships in the data. It has been successfully applied to various biomedicine applications, spanning from microarray gene expression of cancer patients data to the development of biomarkers using genetic, clinical, and/or imaging data. Most of the SVM books in the market often detail the underlying theory of this elegant learning algorithm, but those books do not elaborate on all of the necessary details of how the formal methods can be applied in practical settings. Statnikov and his colleagues bridge this gap in the second volume of their two-book series. All 16 chapters of the book are brief and contain precise information for expert and non-expert readers. Each chapter adheres to the same format, with a summary for each chapter’s end. Furthermore, high quality figures (some in color) help readers to easily follow the practical fundamentals of the applications.
“A Gentle Introduction to Support Vector Machines in Biomedicine, Volume 2: Case Studies and Benchmarks” is a useful introductory book that clearly introduces readers to case studies and the practical use of SVM methods in presented cases. The book has 16 chapters that are well organized into five corresponding categories: preliminaries (part I), genomic data (part II), text data (part III), clinical data (part IV), and data-independent (part V). In order to make this book self-contained, core concepts are introduced in the beginning of the book, with introductory chapters in part I. In part II of the book (chapters 3-to-7), the authors shows the use of SVM for building diagnostic classifiers for breast cancers, based on microarray based genomic data and a comparison of SVM with random forest and kernel ridge regression classifiers. The use of SVMs in genetic applications is an excellent choice for an application in biomedicine because microarray usage in drug discovery, biomarker determination, pharmacology, disease sub-categorization, and development of prognostic tests are expanding. Therefore, the use of SVM is often needed for a good understanding of genetic data analysis in various diseases. In part III, SVM modeling is applied to the text-based data to determine whether biomedical journal articles satisfy evidence-based medicine or not. The authors identify web sites that reveal misleading and unproven cancer treatments, predict future article citations, and classifying essence of article citations (chapters 8-to-12). Part IV consists of two chapters (13 and 14) that demonstrate applications of predicting clinical laboratory values, modeling clinical judgment, and guideline compliance in the diagnosis of melanoma. Since extensive and unnecessary laboratory test ordering is a global problem in almost all clinical centers, the use of SVM for this application is well grounded. Part V (chapters 15 and 16) concludes the book, as authors address and highlight the importance of variable selection in simulation experiments.
Although most of the topics in this book are fundamentally sound, the practical applications of the SVM with text data (part III) are somewhat over-emphasized. Even though SVMs are becoming more and more popular in a wide variety of biological and clinical applications, the authors chose to dedicate five long chapters to the application of SVM use (citation and document retrieval/categorization analysis)—unnecessarily long for an introductory book—instead of including different application types in the biomedicine field, where SVM has been shown successful. Moreover, the variable selection method (Part V) could have focused more on current state-of-the-art methods in the literature. And instead of conducting simulation experiments, the authors could have presented real clinical or genomic applications in the biomedicine field.
In conclusion, the reviewed book is the second volume of the two-volume series, intended to gradually introduce SVM to the field of biomedicine. The authors make a worthy contribution to the informatics in the biomedicine field by introducing various biomedicine applications for which SVM can be used successfully. The potentials of SVM in biomedicine are addressed in a well-written, rigorous, and engaging manner; therefore, readers gain a comprehensive perspective on the use of SVM in biomedicine applications that will add to their own bank of knowledge.
Ulas Bagci, PhD.
Center for Infectious Disease Imaging (CIDI),
National Institutes of Health (NIH), Bethesda, Maryland, 20892 USA.