An Alternative Approach Based on Artificial Neural Networks to Study Controlled Drug Release MARCUS A.A. REIS, RUBE ´ N D. SINISTERRA, JADSON C. BELCHIOR Departamento de Quı ´mica—ICEx, Universidade Federal de Minas Gerais, Pampulha, (31.270-901) Belo Horizonte, MG—Brazil Received 7 February 2003; revised 20 June 2003; accepted 19 August 2003 ABSTRACT: An alternative methodology based on artificial neural networks is proposed to be a complementary tool to other conventional methods to study controlled drug release. Two systems are used to test the approach; namely, hydrocortisone in a biodegradable matrix and rhodium (II) butyrate complexes in a bioceramic matrix. Two well-established mathematical models are used to simulate different release profiles as a function of fundamental properties; namely, diffusion coefficient (D), saturation solubility (C s ), drug loading (A), and the height of the device (h). The models were tested, and the results show that these fundamental properties can be predicted after learning the experimental or model data for controlled drug release systems. The neural network results obtained after the learning stage can be considered to quantitatively predict ideal experimental conditions. Overall, the proposed methodology was shown to be efficient for ideal experiments, with a relative average error of <1% in both tests. This approach can be useful for the experimental analysis to simulate and design efficient controlled drug- release systems. ß 2004 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 93:418–430, 2004 Keywords: neural networks; controlled release; mathematical models INTRODUCTION Considerable attention has been focused on the development of new methodologies to study control- led-release systems mainly because of their great impact on modern therapeutics. Bioceramics or biodegradable polymers are used as matrices to deliver a wide range of drugs in an efficient way. However, analysis may be complicated by the occurrence of multicomponent transport pro- cesses, different kinds of matrices, composition, device geometries, drug loading, saturation solu- bility in the matrix, diffusion, swelling, polymer dissolution, and erosion, 1 and hence theoretical approaches are helpful in determining ideal param- eters e.g. diffusion coefficients. 2 These quantities can be, in principle, predicted by theoretical models to reduce the number of necessary experi- ments. Thus, the models can be considered as useful strategies to simulate the release profile for controlled-release systems. However, as pointed out by Siepmann and Peppas, 3 mathematical approaches covering all the possible chemical and physical processes are not yet available. Nevertheless, efficient models have been pro- posed, for example, by Fu et al. 2 and Higuchi, 4 to circumvent complex problems in drug delivery systems. The advantage of using these mathema- tical models is to allow the experimentalists to determine the drug release for tablets with a specific size, polymer, and unknown diffusion coefficient. In general, most of these models for drug delivery systems are based on the solution of Fick’s second law of diffusion. However, these conventional approaches, as pointed out by Fu et al., 2 have some limitations because all data points are treated with equal weight. Further- more, the least-squares method may not be the 418 JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 2, FEBRUARY 2004 Correspondence to: Jadson C. Belchior (Telephone: 5531 3499 5775; Fax: 5531 3499 5700; E-mail: jadson@carbono.qui.ufmg.br) Journal of Pharmaceutical Sciences, Vol. 93, 418–430 (2004) ß 2004 Wiley-Liss, Inc. and the American Pharmacists Association