ANN Optimization Using Ant Colony Algorithm for Predicting the Valsartan Sustained Release from Polyelectrolyte Complexes Matrix Tablets

The aim of this work is to combined an artificial neural network (ANN) with the Ant Colony Optimization (ACO) to model the prolonged release profile of Valsartan (VAL), an antihypertensive drug formed in matrix tablets based on a mixture blend of carboxymethyl-kappa-carrageenan (CMKC) and chitosan (CTS) forming polyelectrolyte complexes (PECs). CMKCs derivatives were obtained by etherification of KC with various degrees of substitution (DS = 0.8, 1.0 and 1.2) and different formulations were prepared by direct compression method and compared to CTS/KC tablets. The CMKC has taken different concentrations in formulated tablets (60, 80 and 100mg/tablet). The in vitro dissolution test was conducted under simulated gastric and intestinal conditions to achieve drug release for more than 12 h. The ANN-ACO model has been developed by considered the DS, the CMKC amount and the time as inputs of the system, while the VAL release amount has been considered as the output. The results showed that the CTS/CMKCs matrices ensured a prolonged release of VAL as compared to CTS/KC based formulations. Meanwhile, increasing the DS of CMKC derivatives resulted in a slower drug release rates kinetics. ANN-ACO modeling has accurately predicted the kinetic release of VAL from each formulation. This performance was demonstrated by the obtained R
²
=0.999 and RMSE=4,85×10
^-4
. The experimental results indicated that CTS/CMKC has a potential as a hydrophilic matrix that can be employed to prepare prolonged drug release dosage forms.