Antiviral drugs preformulation: thermal energy effects by multilayer perceptron network approach and Hirshfeld surface properties

Relevant drugs, atazanavir sulfate (ATA) and ritonavir (RIT), must be studied for their physicochemical properties to determine the critical material attributes that affect product technology. Interactions of the surrounding molecules calculated by the Hirshfeld surface not only direct the internal arrangement of the crystal structure but also provide some insights into the behavior of both molecules. The non-isothermal Vyazovkin method and multilayer perceptron neural networks (MLP) from thermal analysis data were used for the kinetic studies. The thermal behavior of ATA indicates melting at 190.7 °C (ΔH_m=71.5 J g⁻¹) with decomposition (E_a = 101.8 kJ mol⁻¹; α = 10%). For RIT indicates melting at 119.2 °C (ΔH_m=66.5 J g⁻¹) with decomposition (E_a = 100 K J mol⁻¹; α = 10%). The kinetic model contribution AM2 describes the thermal decomposition of ATA and RIT. The Hirshfeld surface properties corroborated these results. The MLP model, under a wide range of heating conditions, proved to be adequate with a low residual error for adjustment. Hirshfeld surface and calorimetric parameters were applied to define the ATA and RIT critical material attributes for the R&D of pharmaceutical products.