Computerassisted Medical Decision Making

This textbook offers a comprehensive analysis of medical decision making under uncertainty by combining Test Information Theory with Expected Utility Theory. The book shows how the parameters of Bayes’ theorem can be combined with a value function of health states to arrive at informed test and treatment decisions. The authors distinguish between risk-neutral, risk-averse and prudent decision makers and demonstrate the effects of risk preferences on physicians’ decisions. They analyze individual tests, multiple tests and endogenous tests where the test outcome is chosen by the decision maker. Moreover, the topic is examined in the context of health economics by introducing a trade-off between enjoying health and consuming other goods, so that the extent of treatment and thus the potential improvement in the patient’s health becomes endogenous. Finally, non-expected utility models of choice under risk and uncertainty (i.e. ambiguity) are presented. While these models can explain observed test and treatment decisions, they are not suitable for normative analyses aimed at providing guidance on medical decision making.

This textbook offers a comprehensive theory of medical decision making under uncertainty, combining informative test theory with the expected utility hypothesis. The book shows how the parameters of Bayes’ theorem can be combined with a value function of health states to arrive at informed test and treatment decisions.

This textbook introduces students to the basic concepts and methods used to measure and compare the expected benefits, risks and cost of preventive and therapeutic medical interventions. It provides an easily accessible overview of comparative effectiveness and its practical applications to daily medical decisions. The book includes learning objectives for each topic, definitions of key terms and topic summaries. Each chapter is written by a highly regarded academic and extensive reference is made throughout to other sources of literature where the interested reader can find further details.

This book aims to provide threshold models to help physicians to make optimal diagnostic, therapeutic and predictive decisions. Readers will not only find theoretical information but also practical examples illustrating how these decisions should be made.

In recent years, AI/ML tools have become more prevalent in the fields of medical imaging and imaging informatics, where systems are already outperforming physicians in a range of domains, such as in the classification of retinal fundus images in ophthalmology, chest X-rays in radiology, and skin cancer detection in dermatology, among many others. It has recently emerged as one of the fastest growing research areas given the evolution of techniques in radiology, molecular imaging, anatomical imaging, and functional imaging for detection, segmentation, diagnosis, annotation, summarization, and prediction. The ongoing innovations in this exciting and promising field play a powerful role in influencing the lives of millions through health, safety, education, and other opportunities intended to be shared across all segments of society. To achieve further progress, this Special Issue (SI) invited both research and review-type manuscripts to showcase ongoing research progress and development based on applications of AI/ML (especially DL techniques) in medical imaging to influence human health and healthcare systems in the diagnostic decision-making process. The SI published fourteen articles after a rigorous peer-review process across the spectrum of medical imaging modalities and the diversity of specialties depending on imaging techniques from radiology, dermatology, pathology, colonoscopy, endoscopy, etc.