Computational Modeling and Machine Learning Evaluation of Factors Impacting Regional Nasal Dosage Delivery for Micro-Sized Peptide-Based Therapeutics

Nasal peptide delivery offers advantages over systemic administration [۱], [۲], but lacks acomprehensive understanding of crucial parameters for regional dosage deposition in the upperrespiratory system. This study evaluated Peptide-based therapeutics (PTPs) deposition in a healthy adultmale (۳۷ years old). Nasal geometries were constructed from CT-scan images, treating each side of thenostril as a distinct nasal passage model. Computational Flow Dynamics (CFD) simulated PTPtrajectories using the Lagrangian tracking approach, considering forces like drag, Saffman’s lift,thermophoretic, and Brownian forces [۳]. The steady-laminar and steady-kω-SST models incorporatedthree flow rates (۸.۷, ۱۵, and ۳۰ L/min) and two flow regimes (laminar and turbulent). Depositionanalysis in four nasal regions (vestibule, nasal valve, anterior turbinate, and nasopharynx) wasconducted for varying PTP diameter (۱–۱۰۰ μm), spray cone angle (۳۲º, ۷۹º), and injection speed (۲,۱۹.۲ m/s). The anterior turbinate emerged as a favorable site for local and systemic nasal drug delivery[۴]. In this study, low injection speed and spray cone angle play pivotal roles in maximizing anteriorturbinate deposition. Utilizing ۲۴ distinct inhalation and PTP delivery scenarios generated throughnumerical simulations, machine learning models underwent training with five-fold cross-validation topredict the delivered dose, eliminating the need for future partial differential equation solvers. Therandom forest and gradient boosting models yielded R۲ scores of ۰.۹۱ and ۰.۹۰. The deposition locationin the nasal cavity, the diameter of PTP, and injection velocity emerged as the most crucial factorsinfluencing the delivered dose.

کلیدواژه ها:

Intranasal delivery ، Therapeutic peptides ، CFD ، Machine Learning