The Characterization of Dosage Sequences of Self-Dispensing Multi-Dose Dry Powder Inhalers by Spectral Analysis and ARIMA

ABSTRACT

Recently many so called self-dispensing multi-dose dry powder inhalers were designed. Calculation of the average weight, the coefficient of variation of the dosing-sequences are often used to indicate the (in)accuracy of the system. These parameters supply no information with respect to the cause of the inaccuracy. No statistical methods are in use to locate the cause of the inaccuracy. Two techniques, spectral analysis and ARIMA-modelling were used to determine the characteristics of the sequences of two (model) dry powder inhalers. Spectral analysis detects periodic behaviour in sequences, while ARIMA unravels the way a dose interacts with the next one. It was shown that the sequence of the first inhaler is a moving average process, indicating sticking powder in the dosing mechanism itself. The second inhaler showed negative autocorrelation, indicating insufficient supply of powder to the dosing mechanism. Spectral analysis and ARIMA-modelling provide information not obtainable with the present evaluation techniques, enabling the designer to focus his efforts in improving the system. Their use can be implemented in the evaluation of self-dispensing multi-dose dry powder inhalers.