Abstract
Substituting spacer by another in noninvasive ventilation (NIV) involves many variables, e.g. total emitted dose (TED), mass median aerodynamic diameter (MMAD), type of spacer, total lung deposition and total systemic absorption, which must be adjusted to ensure patient optimum therapy. Data mining based on artificial neural networks and genetic algorithms were used to model in vitro inhalation process, predict and optimize bioavailability from inhaled doses delivered by metered dose inhaler (MDI) using different spacers in NIV. Modeling of data indicated that in vitro performance of MDI-spacer systems was dependent mainly on fine particle dose (FPD), fine particle fraction (FPF), MMAD and to lesser extent on spacer type. Ex vivo model indicated that amount of salbutamol collected on facemask filter was directly affected by FPF. In vivo model (24hQ) depended directly on spacer type, FPF and TED. Female patients showed higher 0.5hQ and 24hQ values than males. AeroChamber VC spacer demonstrated higher TED and 24hQ in vivo values. Results indicated suitability of MDI-spacer systems in achieving appropriate in vitro inhalation performance. The possibility of modeling and predicting both ex vivo and in vivo capabilities of MDI-spacer systems from knowledge of in vitro attributes enabled detailed focus on important variables required to deliver safe and accurate doses of salbutamol to ventilated patients.