Development of Clinical Decision Support Systems using Bayesian Networks

Building a clinical decision-support system is not just a matter of encoding rules: it also means translating clinical expertise, uncertainty, and team-based practice into a model that can be examined and tested. Mario A. Cypko explores that challenge through Bayesian networks, using multidisciplinary treatment decisions for laryngeal cancer as a working case.

This research monograph follows the development of a clinical decision model from its foundations to its validation and use. Its value lies in bringing the clinical and technical sides of the work into the same discussion: how expert knowledge is acquired, represented, checked, and made accessible to decision-makers.

From tumor-board practice to a decision model

The opening chapters introduce clinical decision-support systems, Bayesian-network concepts, and the setting of head-and-neck tumor boards. The laryngeal cancer example gives the technical discussion a concrete clinical context, while keeping the emphasis on how a support system is developed—not on offering treatment advice to patients.

Following the TreLynCa case study

A central section presents TreLynCa, a tumor-board decision model for laryngeal cancer. Cypko discusses teamwork and knowledge acquisition, model development, and validation. The case helps readers see how clinical knowledge and probabilistic modeling meet in a specific multidisciplinary decision process.

Modeling, validation, and clinical interaction

The book also considers guided tools for constructing Bayesian-network models and a graphical interface for verifying patient-specific decisions. These topics make the work relevant to readers interested not only in the model itself, but in how models can be reviewed, used, and quality-managed within a clinical environment.

For readers working across disciplines

Medical informatics researchers, knowledge engineers, clinicians involved in decision-support development, and advanced students will find a focused account of the methods and practical challenges involved. The book’s distinctive contribution is its sustained connection between probabilistic modeling and the realities of multidisciplinary clinical decision-making.