PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of ccforumBioMed CentralBiomed Central Web Sitesearchsubmit a manuscriptregisterthis articleCritical CareJournal Front Page
 
Crit Care. 2010; 14(Suppl 1): P209.
Published online 2010 March 1. doi:  10.1186/cc8441
PMCID: PMC2934064

Towards partially automated ventilation: adapting decision-making according to medical preferences

Introduction

A novel methodology is proposed to adapt decision-making strategies into our fuzzy-based expert system, AUTOPILOT-BT [1]. The special features of this approach are: knowledge from clinical experts can be extracted in a setup simulating daily ICU routine; an automated process serves to obtain the required information and to create new fuzzy sets; and the fuzzy controller from AUTOPILOT-BT employs the newly derived fuzzy rules. Thus, the knowledge base can easily be modified and the resulting mechanical ventilation therapies may be adapted to individual preferences of the clinician.

Methods

The methodology consists of: (i) Acquisition of decision-making strategies from single or groups of anesthesiologists. This can either be done with a questionnaire or with a PC-based program simulating the doctors every day situation in diagnosis. (ii) Definition of fuzzy membership functions based on the acquired knowledge (fuzzification of the input). (iii) Construction of fuzzy inference rules and defuzzification or calculation of change in ventilator settings (controller action). This approach allows implementing clinician-dependent decision-making that reflects individual preferences. Thus guidelines from EBM can be used but as well ICU specifics can be realized, for example different levels of acceptable hypercapnia in patients with acute respiratory distress syndrome (ARDS).

Results

Exemplarily Figure Figure11 shows, for healthy and ARDS lungs, the difference between two fuzzy sets for our paCO2 controller given from two clinicians (C10 and C51) with different expertise in mechanical ventilation. With the newly designed fuzzy sets, our AUTOPILOT-BT reacts according to the clinicians' preferences, but still minimizes the time in which the patient is not ventilated within the specified limits.

Figure 1
Normal ventilation for healthy vs ARDS patients - fuzzy sets for paCO2 given from two clinicians: (left) C10, (right) C51.

Conclusions

The system automatically implements the know-how of medical experts in ventilation management if the clinicians are willing to interact with the query system. The resulting strategy is mainly influenced by the expertise, experience and demands of the clinician. Thus the AUTOPILOT-BT, has the potential to select established guidelines or to adapt the system to modified ventilation therapies. Further clinical trials will test the actual clinical efficiency of different controllers.

References

  • Lozano S, Tech Health Care. 2008. pp. 1–11. [PubMed]

Articles from Critical Care are provided here courtesy of BioMed Central