MUMBAI, India, Oct. 26 -- Intellectual Property India has published a patent application (202611107407 A) filed by Noida Institute Of Engineering And Technology Pharmacy Institute on September 07, 2026, for Physics-Informed Neural Ode With Bayesian Optimization For Efficient, Uncertainty-Aware Sustained-Release Formulation Design.
Inventors include Dr. Priyanka Bansal; and Mr. Anmol Dogra.
The application for the patent was published on October 23, 2026, under issue no. 43/2026.
Abstract: The invention introduces PINODE-MOBO, a computer-implemented framework for designing sustained-release pharmaceutical formulations under mixture- composition constraints. It combines a physics-informed neural ordinary differential equation surrogate to model cumulative release, a Bayesian deep-kernel Gaussian process for calibrated uncertainty, and a multi-objective Bayesian optimization loop using q-Noisy Expected Hypervolume Improvement for batch candidate selection. Mixture constraints are enforced via softmax reparameterization, avoiding penalty functions, while physics-informed regularization embeds Fickian, Higuchi, and Korsmeyer-Peppas kinetics. On a metronidazole sustained-release film benchmark with four formulation factors and pharmacopoeial targets across five days, the framework achieves root mean square errors of 0.94–1.12%, outperforming a prior FDA-Scheffé-NSGA-III pipeline (errors 1.20–6.73%) while reducing experiments from 20 to 12. A meta-learned prior enables cross-drug transfer with as few
Disclaimer: Curated by HT Syndication.
