A SELF-TUNING EFFECT OF MEMBERSHIP FUNCTIONS IN A FUZZY-LOGIC-BASED CARDIAC PACING SYSTEM

This paper describes a self-tuning method of membership functions in a fuzzy-logic-based cardiac pacing system and validates its feasibility in a double sensor system which has minute ventilation and oxygen sat uration level as its guides for the rate regulation. Though the agreem ent between the pacing rates (fuzzy rates) calculated with three lingu istic variables for each parameter and the target rates were not satis factory, it was improved significantly by tuning the membership functi ons. Almost the same evaluated values with those obtained by using six linguistic variables for each parameter were obtained. Time required for the self-tuning process was about 40 s (386CPU, 20 MHz) which was fast enough for the system. The smaller number of linguistic labels re sults in a smaller number of rules, which is beneficial in implantable cardiac pacemakers with limited memory capacity. A fuzzy-logic-based cardiac pacing system is promising for the realization of custom-made cardiac pacemakers.