A Measurement System Design Technique for Improving Performances and Reliability of Smart and Fault-Tolerant Biomedical Systems

Abstract

The present Chapter intends to provide a practical guide for designers in planning smart measurement systems to be used in critical applications like medical ones. The authors propose an original approach to the design of measurement instrumentation with high performances. The procedure allows the designer to characterize the best measurement uncertainty that the system must have. The main purpose is to project a system with appropriate performances, so that suitable accuracy and reliability levels are guaranteed for the resulting measurements. The used approach starts from the consideration that typically the measured data are used during the processing stage in order to make decisions. In example, medical diagnoses are based on measurements which are put in comparison with reference limits. Therefore the measurement uncertainty can affect the reliability of the final results so to be source of mistaken decisions. Consequently high values of measurement uncertainty may be cause of unreliable data and inaccurate diagnoses. In the Chapter, a statistical model is used in order to characterize the functional relationship between the measurement uncertainty and the probability to make mistaken decisions because of the same uncertainty. Consequently the designer can characterize the best uncertainty value for the measurement system to be projected. So suitable reliability can be guaranteed during the decision-making stage by assuring a tolerable probability of mistaken decision. Furthermore the Chapter describes the architecture used in order to design smart and patient-adaptive biomedical systems. In detail the use of specific memory devices is shown. Information concerning the metrological characteristics of system and the patient data are so made available. In detail, information on measurement uncertainty and calibration curve is stored in a first memory device in order to estimate the reliability of measurement results. Whereas a further writable and readable storage device stores private and medical data of the specific examined subject. Such memory is a personal data-logger replaced for each patient and updated with the passing of time according to the current clinical conditions of the subject. In this way the computing algorithm fits the patient by means of the available information so to qualify the final diagnosis. In fact the available data allow the system to adapt and configure itself according to the patient features and to his health state so to get fault-tolerant diagnoses. In this way it is possible to project a biomedical system which is updateable and configurable according to the specific subject. Experimental results concerning the project of an ECG measurement system are added.