P2Model-based neuro-fuzzy control of FiO 2 for intensive care mechanical ventilation HF Kwok, GH Mills, M Mahfouf, DA Linkens Introduction: In our experience, very often, even with a nonrebreathing mask (NRM), high oxygen delivery to patient with the existent materials is insufficient. However, many of these patients need high oxygen therapy for a limited period of time. In this study we report our experience with a new device that serves to increase the concentration of oxygen delivered by a classical nasal cannula or catheter. It is not an oxygen mask.
Aim:To demonstrate how a simple adjunctive system to classical nasal cannula or catheter improves considerably the oxygenation of patients at constant O 2 flow rate.Design: Prospective, observational study.
Method:The double trunk mask (DTM) is a modified tusk mask described by Hnatiuk. It is composed by a normal aerosol mask with 22 mm of diameter lateral holes, 38 cm of long flexible tubing are inserted to each side of the mask. The DTM is just applied to the face of the patients who already receive O 2 through a nasal cannula or catheter. Forty-five consecutive patients, admitted in the ER or ICU, and needing oxygen delivery, are included in our study. The data collected are: PaO 2 , PaCO 2 , breathing rate with a mean flow rate of 3.58 l/min, at t0, t30 min prior to DTM and then 30 min after DTM application.
Results:Nasal The knowledge-based approach to fuzzy logic control of mechanical ventilation on the ICU can be prone to bias in the experts' knowledge and errors resulting from poor communication during rule-base derivation. Therefore, a different approach was explored in the development of a fuzzy controller to control the inspired oxygen fraction (FiO 2 ). The performance of such a controller was compared with the performance of the clinicians.Method: (1) The development of a neuro-fuzzy controller. This was developed by training a neural network to generate an optimal change in the FiO 2 in order to achieve a target arterial oxygen tension (PaO 2 ) on a mathematical model of the gas exchange system (SOPAVent). The neural network learnt the relationship between the blood gases, FiO 2 and PEEP and other ventilator settings. This was done by exposing the neural network to the blood gas results produced by applying a range of FiO 2 and PEEP values to the SOPAVent model. This first neural network was then combined with another neural network which represented a fuzzy logic rule-base. The fuzzy rule-base consists of a set of 'If …, Then …' statements based around combinations of FiO 2 , PEEP and PaO 2 . The fuzzy rule-base was then adjusted by changing the weights of the neuro-controller (which correspond to the 'Then …' part of the fuzzy rules) during neural network training. The neuro-controller output is equivalent to the output from a fuzzy inference system of three inputs (the difference between the actual PaO 2 and the target, the PEEP level and the FiO 2 ).(2) Comparing neuro-fuzzy and clinicians' control. The scenarios were based on the data from th...