4 von 4 Kunden fanden die folgende Rezension hilfreich
G. William Moore
- Veröffentlicht auf Amazon.com
This book is a must-read for biomedical professionals, on-the-job or in-training, with a serious interest in data management. The author, Dr.
Jules J. Berman, studied mathematics at M.I.T., earned a Ph.D. in pathology and an M.D., practiced general pathology for over a decade, and served as Program Director for the Pathology Informatics program at the U. S. National Cancer Institute. He has published over one hundred first-author papers in peer-reviewed journals in this field. He brings together this varied background, plus a breezy, no-nonsense writing style, in a comprehensive but enjoyable book about biomedical informatics.
You can digest the highlights of the book in three hours. First, carefully study the Table of Contents. Then, scan through the tables with bullet headers, which summarize all the basic ideas of the book. Finally, read the annotated bibliography and glossary at the end of the book. The author cites many of the major works in the field, and has insightful opinions about the scope and value of these works.
The book embraces all modern biomedical informatics. There are chapters devoted to the definition of biomedical databases; the availability of biomedical databases; privacy and confidentiality; database standard formats; programming; biomedical nomenclatures; automated translation and de-identification methods; cryptography; clinical trials; and how to apply for a grant. Each chapter begins with the existing history and background of the chapter's subject matter, followed by a complete description of the area, and copious examples. The author includes numerous references to publicly available sources of biomedical data and software, as well as instructions on how to download and run these resources. The author has tested all these resources himself, and provides practical advice on what to do and how to avoid pitfalls.
The descriptions and explanations of U. S. privacy regulations, including the Common Rule and the Health Insurance Portability and Accountability Act (HIPAA), are themselves worth the price of the book. The book includes sections on the use of unconsented patient data; use of proprietary software and standards; conflicts of interest; patents; etiquette of free software usage; and lawsuits. There is even a six-point list of when it is OK to tell a lie to a patient.
Dr. Berman presents instances in which patients' rights have been grossly violated, as for example, the Tuskegee Syphilis Study, and a study performed at the Memorial Sloan Kettering Cancer Center involving the injection of live cancer cells into patients. Dr. Berman also discusses the use of "thought experiments" as a substitute for some clinical trials on patients.
The author believes, and I agree, that every biomedical informatician should be able to do SOME programming. This skill is valuable both for examining databases "on the fly" before handing them over to a professional programmer; and for more effectively supervising the work of professional programmers. The chapter devoted to this subject is entitled, `Just Enough Programming'. The author discusses Perl, which is available free to the public, and which is relatively easy-to-use. You can start programming in Perl after studying it for a few hours. The author shows how to download and install a free Perl interpreter, and discusses how to write simple programs, for such tasks as: indexing an electronic book by page number for all occurrences of disease names; finding a distribution of all palindromes in the human genome database; finding a frequency distribution of octamers in the human genome database; finding diseases that have a chronologic relationship with another disease; collecting 100 histopathologic images, say, of small cell carcinoma of lung. He provides sample Perl scripts, and lists common programming errors and their resolution.
The author has a flair for biomedical informatics history. He discusses such historical events and personages as: the Pyramids of Giza (large, labor-intensive project); the Sumerians (ancient data archivists); Hippocrates (father of medicine; life is short, art is long); Tycho Brahe and Johannes Kepler (data sharing); Isaac Newton (father of physics, beneficiary of Brahe's data); Giovanni Battista Morgagni (pathologic anatomy, 640 published autopsies); Edward Jenner (smallpox clinical trial); Louis Pasteur (rabies clinical trial); Karl Wunderlich (body temperature data); Paul Ehrlich (unnecessary data sharing); and numerous present-day luminaries.
A minor quibble: In a major work about data management, the author can't seem to decide whether `DATA' is singular or plural. My preference, as a student of Latin, is plural, with singular DATUM. Whatever, the author should be consistent in this usage.
The index for this book leaves much to be desired. One of my long-standing gripes about reading non-scientific literary works is that, if I want to re-visit a particular paragraph, I almost have to reread the entire book to find it. Some novelists and English professors seem to take almost sadistic pleasure in enforcing this discipline on their readers and students.
For a work aimed at busy biomedical professionals, I expect a higher standard. When the author develops a significant new idea, he should leave footprints in the index. This practice allows the reader to return quickly to that idea, and to find other occurrences of the same idea elsewhere in the book. There are major, significant ideas that appear in the Table of Contents, which are not included in the index, at least as far as I can tell. These include: directory; download; hash; interpreter; meaning; object code; script; source code; and syntax. Some of these concepts are present under another name, but they cannot be found under these names used by the author in the text. Some of these concepts are present in the glossary, but typically without page numbers pointing to the text. Some concepts are indexed in a strange place, and not cross-indexed. For example, `pseudocode programs' is an index-item pointing to page 96, with no apparent relation to `perl pseudocode', pointing to pages 108-109 and 115-116. I would have more confidence in this index if I could understand why some terms are present and other equally or more important terms are absent.
Another annoying feature of this index is the capricious use of acronyms, such as ACR/NEMA, DAML, HL7, MESH, UMLS, XML, etc., without expanding the acronym in the index, and often without even having the expanded form present at the appropriate location elsewhere in the index. Some acronyms ARE expanded in the index. What motivated the indexer to expand some acronyms, and not expand others?
Except for the index, this book is an unqualified success.