Jason Cahill


Presentation: 1:15-2:30 p.m., Kennedy Union Ballroom



Download Project (874 KB)


Mechanical ventilation, as a resource for critical care, is a balancing act. Everyday physicians, nurses, and respiratory therapists rely on this life saving intervention to support patients who are too weak or ill to breathe on their own. Unfortunately, structural and physiological damage can easily occur as a result of aggressive or long-term ventilator use. Because of the cardiopulmonary system’s tremendous complexity as well as the innate variability in parameters due to disease, individuality, and time, most critical care ventilators require continual adjustment to avoid these side effects, essentially making the physician the controller. This project proposes a radical step forward in design, a three-part control method that will bring the patient into the loop in an unprecedented way. First, a dynamic inversion controller based on a 148-state model of the cardiopulmonary system. Second, a neural network-based adaptive controller capable of reducing real time deviations between the base controller and the patient. Finally, a gradient based concurrent learning algorithm that optimizes the parameters of the base cardiopulmonary model in real-time, thereby further reducing error associated with long term variations. The complete controller will regulate the patient’s respiration in real time utilizing vital information from existing bedside monitors.

Publication Date


Project Designation

Graduate Research

Primary Advisor

Raul Ordonez

Primary Advisor's Department

Electrical and Computer Engineering


Stander Symposium, School of Engineering

Institutional Learning Goals

Scholarship; Practical Wisdom

Intelligent Adaptive Control System for Combating Ventilator Induced Lung Injury