Information analysis and validation of intelligent monitoring systems in intensive care units

The validation of intelligent systems is an important task to perform. Typically, the results of the validation analysis are used to verify whether or not the system satisfies the initial design requirements, and to acquire new knowledge and/or refine the knowledge already acquired. In practice, the validation of intelligent systems usually requires the application of several different techniques (e.g. retrospective, prospective and quantitative). In this paper, the authors present the methodology devised to validate PATRICIA: an intelligent monitoring system designed to advise clinicians on the management of patients who are dependent on mechanical ventilation. The application of this methodology requires that appropriate validation paradigms are selected, depending on both the application domain and the characteristics of the intelligent system. The article also presents and discusses validation results.