Skip to main content
  • Poster presentation
  • Published:

Knowledge acquisition to design a fuzzy system for disease-specific automatic control of mechanical ventilation


A closed-loop system for automated control of mechanical ventilation, Autopilot-BT, will be enhanced [1]. It must be able to adapt to diverse disease patterns. The Autopilot-BT is based on fuzzy logic, which can model complex systems using expert knowledge. The expert knowledge was acquired by a specifically designed questionnaire (Figure 1a).

Figure 1
figure 1

(a) Questionnaire for ARDS. (b) Membership functions for "healthy" and "ARDS".


Exemplarily we will focus on the respiratory rate (RR) controller, responsible for the arterial partial pressure of carbon dioxide/end-tidal carbon dioxide pressure (etCO2) control. The etCO2 values are classified into seven different fuzzy sets ranging from 'extreme hyperventilation' to 'extreme hypoventilation'. For different diseases such as chronic obstructive pulmonary disease or acute respiratory distress syndrome (ARDS), every clinician assigns given etCO2 values to a ventilation status. By averaging over all assignments of the clinicians, new targets and limits for each disease are obtained. Afterwards the new target and limit areas were implemented in a new fuzzy system controlling the RR.


Sixty-one of the anaesthesiologists filled the questionnaire completely, two did not answer. Figure 1b exemplarily shows the different classifications of etCO2 (membership functions) in 'healthy' and ARDS derived from the questionnaire. The membership areas of 'normal state', 'moderate', 'strong' and 'extreme hypo-ventilation' in the ARDS fuzzy system are shifted to the right. Also the basis of the target area 'normal state' ranges in the ARDS system from 30 to 60 mmHg etCO2. One of the limits in the fuzzy system therefore shifts more to the hypoventilated area and the system tolerates etCO2 values up to 60 mmHg as the normal state range.


Disease-specific expert knowledge derived from the questionnaire greatly modifies the performance of the RR controller. The developed disease-specific adaptive controller provides better mechanical ventilation support to patients.


  1. Lozano S, et al: Technol Health Care.

Download references

Author information

Authors and Affiliations


Rights and permissions

Reprints and permissions

About this article

Cite this article

Gottlieb, D., Lozano, S., Guttmann, J. et al. Knowledge acquisition to design a fuzzy system for disease-specific automatic control of mechanical ventilation. Crit Care 12 (Suppl 2), P306 (2008).

Download citation

  • Published:

  • DOI: