INTRODUCTION: Mechanical ventilation in postoperative cardiac surgery can bring some respiratory complications for the patient. To minimize this risk is necessary for proper and rapid mechanical ventilator. The difficult for that is the large number of variables for adjusting the mechanical ventilator and getting all these variables. As the period of mechanical ventilation, usually not exceed 12 hours, this time must be optimized so that the patient may be in spontaneous ventilation as soon as possible. OBJECTIVES: This paper proposes the use of Bayesian Networks to assist the professional in the making decision, speeding up the patients care. MATERIALS AND METHODS: The development of the RB used a database composed of 137 clinical cases. The evaluation was performed by the instruments operational validity measurement, contingency tables and ROC curves. RESULTS: RB presented an adequate performance for the election of the ventilatory modes and parameters. CONCLUSION: The results applying RB were similar to those suggested by the literature, showing consistency between computational and human reasoning.
Artificial intelligence; Bayesian network; Mechanical ventilation; Cardiac surgery
