This edited volume is a most welcome addition to the literature of medical artificial intelligence for many reasons. This book provides a historical perspective not readily attainable by most means. For the physician, scientist, or psychologist interested in applying the vast power of modern computers to medical application, the text provides a useful and usable source of reference material for expert systems. An entire decade of key work is presented with significant scope and with introductory comments to allow a careful and interested reader to be conversant with most of the techniques, concepts, and pitfalls. Obviously, not all of the papers critical to the evolution of this area are presented; that would be an impossible task. Those that are presented focus on a particular problem and present to the reader a wealth of information on which to judge or evaluate the more recent developments.

The introductory chapter by the editors, titled “Medical Artificial Intelligence Programs,” explains in clear terms the active working definitions so necessary for communication and understanding of this new area. The papers themselves frequently reflect the problems of specification and definition so vital to the success of a project. Providing correct definitions helps to evaluate older concepts with a perspective, rather than just randomly. All of the major techniques are presented: rule based, knowledge based, and statistical. No one is presented an an ideal solution; they are just examples. Since this area develops and evolves, it is important to not put value judgments on the systems but to evaluate each on its own merits and in comparison to others.

Some authors have provided a retrospective evaluation of their effort, putting their discussions in context and providing a self-critique. This is a healthy concept; it reflects the thinking and rationale behind the program while affording the reader an education.

Chapters 11 (“Intelligent Computer-Aided Instruction for Medical Diagnosis” by W. J. Clancey, E. H. Shortliffe, and B. G. Buchanan), 13 (“Knowledge Organization and Distribution for Medical Diagnosis” by F. Gomez and B. Chandrasekaran), 14 (“Causal Understanding of Patient Illness in Medical Diagnosis” by R. S. Patil, P. Szolovits, and W. B. Schwartz), and 16 (“Explaining and Justifying Expert Consulting Programs” by W. R. Swartout) are classics and present just the right concepts in precise and exact language. These represent the best of necessary elements for an effective and useful system. Their message is essential for anyone considering a serious venture into medical artificial intelligence.

The last chapter, “Anticipating the Second Decade” by the editors, identifies new problems based on the experience for the trial systems. It is properly cautious in neither predicting nor forecasting wide ranging applications and use. Rather, it takes a conservative view and neatly dictates the design constraints necessary to achieve such a usable system. This is a significant effort and again demonstrates the stature and quality of the editors.

The volume itself is well designed and as consistent as possible considering the varied sources. The book is an excellent selection of necessary papers. The collation of the individual bibliographies into a single list at once provides a necessary and accessible reference list. With computer science departments expanding to include artificial intelligence in their curriculum and the interest in the field by physicians and other health care workers, this book is readily adapted as a text for a focused course. It meets specific and general needs by its scope and content. It is a most welcome volume to the library.