Artificial Neural Network Prediction of the Patterns of Deposition of Polydisperse Aerosols Within Human Lungs Javed Nazir, David J. Barlow, M. Jayne Lawrence, Ian Shrubb Journal of Pharmaceutical Sciences Volume 94, Issue 9, Pages 1986-1997 (September 2005) DOI: 10.1002/jps.20413 Copyright © 2005 Wiley‐Liss, Inc. Terms and Conditions
Figure 1 Log-normal particle size distribution simulated using monodisperse aerosols. Journal of Pharmaceutical Sciences 2005 94, 1986-1997DOI: (10.1002/jps.20413) Copyright © 2005 Wiley‐Liss, Inc. Terms and Conditions
Figure 2 MLP network architecture used to predict lung deposition fraction of inhaled particles. Journal of Pharmaceutical Sciences 2005 94, 1986-1997DOI: (10.1002/jps.20413) Copyright © 2005 Wiley‐Liss, Inc. Terms and Conditions
Figure 3 Comparison of predicted and measured deposition patterns of unit density spheres within the human respiratory tract during oral breathing (under ‘physical work’ conditions). (A) Laryngeal deposition fraction. (B) Bronchial deposition fraction. (C) Alveolar deposition fraction. (D) Total lung deposition fraction. The breathing conditions are: 1500mL tidal volume, 4s breathing cycle period and 750mL/s inspiratory flow rate. The filled triangles (together with a continuous line to guide the eye) represent the experimental data from Heyder et al.,7 the empty circles, the MLP training data and the filled circles, the MLP predictions for test/cross-validation data. Journal of Pharmaceutical Sciences 2005 94, 1986-1997DOI: (10.1002/jps.20413) Copyright © 2005 Wiley‐Liss, Inc. Terms and Conditions
Figure 4 Comparison of predicted and measured total deposition patterns of unit density spheres within the human respiratory tract during oral breathing. The breathing conditions are: panel A (TV=1000mL, T=16 s, Q=125mL/s), panel B (TV=2000mL, T=16 s, Q=250mL/s) and panel C (TV=1500mL, T=12 s, Q=250mL/s). The filled triangles (together with a continuous line to guide the eye) represent the experimental data from Heyder et al.,7 the empty circles, the MLP training data and the filled circles, the MLP predictions for test/cross-validation data. Journal of Pharmaceutical Sciences 2005 94, 1986-1997DOI: (10.1002/jps.20413) Copyright © 2005 Wiley‐Liss, Inc. Terms and Conditions
Figure 5 Comparison of ANN prediction on total lung dosage with estimates taken from Finlay et al.8 and observed data taken from Newman et al.10 Journal of Pharmaceutical Sciences 2005 94, 1986-1997DOI: (10.1002/jps.20413) Copyright © 2005 Wiley‐Liss, Inc. Terms and Conditions
Figure 6 Correlation of the amount of salbutamol found in urine 30min post-administration from six nebulisers (data taken from Silkstone et al.9) with ANN predictions of total lung deposition. The R2 value obtained is 0.91, and error bars indicate standard errors on the in vivo data (n=9). Labels (a to f) represent the following nebulisers respectively: NLS2-C, Microneb III (2.5mL), Pari LC, Microneb III (4mL), Microneb III (6mL) and Sonix 2000. Journal of Pharmaceutical Sciences 2005 94, 1986-1997DOI: (10.1002/jps.20413) Copyright © 2005 Wiley‐Liss, Inc. Terms and Conditions
Figure 7 Effect of variations in both polydispersity and MMAD on number fractions of polydisperse aerosols deposited in the lungs during intermediate breathing conditions, that is, 500mL tidal volume, 4s breathing cycle period and 250mL/s inspiratory flow rate. Panels A and B show deposition in the tracheobronchial and alveolar regions respectively. The empty circle, filled triangles and filled diamond represent aerosols of MMAD=1 µm, MMAD=3 µm and MMAD=5 µm respectively. Journal of Pharmaceutical Sciences 2005 94, 1986-1997DOI: (10.1002/jps.20413) Copyright © 2005 Wiley‐Liss, Inc. Terms and Conditions
Figure 8 Effect of variations in both polydispersity and MMAD on mass fractions of polydisperse aerosols deposited in the lungs during an intermediate breathing conditions, that is, 500mL tidal volume, 4s breathing cycle period and 250mL/s inspiratory flow rate. Panels A and B show deposition in the tracheobronchial and alveolar regions respectively. The empty circle, filled triangles and filled diamond represent aerosols of MMAD=1 µm, MMAD=3 µm and MMAD=5 µm respectively. Journal of Pharmaceutical Sciences 2005 94, 1986-1997DOI: (10.1002/jps.20413) Copyright © 2005 Wiley‐Liss, Inc. Terms and Conditions