A control method based on recognition of time sequential data transition-application to blood pressure control

The development of an automatic system which, as a means of treatment, can control the blood pressure of a patient by a continuous injection of vaso-acting drugs is in great demand in the medical care field. However, the practical application of such a system has been postponed due to safety considerations. This paper attempts to realize the safe control of the physiological system, and a control method based on the recognition of the transition of the time-series signal has been developed. The idea is applied to the automatic control of blood pressure. For the recognition of the time-series signal transition, a dedicated IF-THEN representation TMSTR (time-sequential multiple state transition representation) is devised which includes the concepts of “state” and “time.” A method is proposed in which more than one “state” is defined for the time-sequential data, and the transitions among states are represented by IF-THEN rules. By this scheme, the time sequential data transition is recognized. Using the constructed representation, the physician's knowledge of the blood pressure control is represented and transplanted to the computer. Using a 16-bit microcomputer, a closed-loop control system is constructed. As a result of animal experiments, a blood pressure control close to that of the physician is realized, indicating the effectiveness of the proposed method in the control of the physiological system.