Artificial neural network for modeling formulation and drug permeation of topical patches containing diclofenac sodium

In this work, topical matrix patches of diclofenac sodium (DS) were formulated by the solvent casting method using different ratios of chitosan (CTS) and kappa carrageenan (KC). Propylene glycol and tween 80 were used as a plasticizer and permeation enhancer, respectively. The drug matrix film was cast on a polyvinyl alcohol backing membrane. All the patches were evaluated for their physicochemical characteristics (thickness, folding endurance, flatness, drug content, tensile strength, bioadhesion, moisture content, and moisture uptake), along with their in vitro release and in vitro skin permeation studies. Franz diffusion cells were used to conduct the in vitro permeation studies. The artificial neural network (ANN) model was applied to simultaneously predict the DS release and the ex vitro skin permeation kinetics. The formulated patches showed good physicochemical properties. Out of all the studied patches, F6 presented sustained permeation in 32 h and was selected as the best formulation. The ANN model accurately predicted both the kinetic release and the skin permeability of DS from each formulation. This performance was demonstrated by the obtained R2 = 0.9994 and R2 = 0.9798 for release and permeation kinetics modeling, respectively, with root mean square error (RMSE) = 3.46 × 10−5.