Run-to-Run control is often used in the semiconductor manufacturing industry when process measurements are not available in situ and changes in the manipulated variables can only be made between process runs. It is pointed out in this paper that many similarities exist between controlling semiconductor manufacturing and designing an effective long-term drug therapy. Based upon these similarities, a run-to-run control algorithm is designed and its validity illustrated with a case study on controlling blood coagulation. The simulation compares the effectiveness of an adaptive run-to-run controller and the anticoagulation control algorithms currently being used by physicians for both short-term and long-term changes in the patient's response to medication. The simulation shows that the adaptive run-to-run control algorithm outperforms the conventional control algorithm by keeping the patient within the therapeutic range during both normal drifts in drug tolerance and large disturbances in the patient's response to the medication.
Proceedings of the 2002 American Control Conference, Anchorage, pp. 4044-4049 (2002)