The care of critically ill patients requiring mechanical ventilation remains beset by the combined effects of critical illness and of the mechanical ventilation of the lung. Such effects are compounded by the lack of knowledge of the rate and time at which 'weaning' from the machine should occur. This project aims at developing an adaptive decision support system to assist ICU staff in the optimisation of ventilation and weaning processes. To help achieve this, an adaptive hybrid model which describes the patient-ventilator interaction during ventilation as well as weaning phases will be elicited. In addition to knowledge gathered through data relating to blood gases and lung expansions, the project aims at exploiting a revolutionary technique developed at sheffield, called Electrical Impedance Tomography (EIT) which consists of measuring, in a non-invasive fashion, the degree of expansion or collapse of the lungs and the effect of the ventilation strategy upon these. Two important aspects of this project relate to the inclusion of the EIT measurement technique to improve the monitoring of the patient's respiratory demands and to the use of granular computing for the hybrid model represented by the neural-fuzzy layer. The elicitation of such a model will form the basis for the design and development of an adaptive decision support system for optimal therapeutic advice on ventilator settings and weaning operation. On-line and off-line validation of the system in a series of ICU trials are envisaged.