Mechanical ventilation is a life-saving therapy for patient treatments in Intensive Care Units (ICUs). The management of mechanical ventilation is a very challenging task. It has long been recognised that a computer-based bedside decision support system is r desirable for optimal ventilator management in ICUs. In this thesis, a closed-loop adaptive model-based ventilator management decision support system is developed. A previously developed ventilated patient mathematical model is further improved and extended with respect to the model parameter estimation and the simulation of the patients as their clinical states evolve. A hybrid modelling strategy is implemented by combining mathematical modelling and data-driven modelling techniques. With the availability of rich data in ICU and the improvements made in the model parameter estimation, the model is able to represent patient state evolution and provide accurate blood gas and tidal volume predictions. An adaptive decision support system is, thereafter developed based on the patient model using an optimisation approach and the system is evaluated via a series of closed-loop simulations. Results show that the srstem can generate good ventilator setting advice subject to the patient state changes and competing ventilator management targets. In addition, a future ventilator management tool, named Electrical Impedance Tomography CElT), is investigated in this thesis in relation to its data processing and feature extraction. The integration of EIT into the current decision support system represents a very promising research direction for the optimal ventilator management decision support.