1Assistant Professor, Department of Computer Engineering, L. D. College of Engineering, Ahmedabad, Gujarat, India
2*Associate Professor, Department of Information Technology, Vidush Somany Institute of Technology & Research, Kadi Sarva Vishwavidyalaya, Gujarat, India
3Assistant Professor, Department of Computer Engineering, L. D. College of Engineering, Ahmedabad, Gujarat, India
The recent developments in artificial intelligence and machine learning had a substantial impact on pharmaceutical research and healthcare technologies by allowing the prediction and optimization of the complex processes in biomedicine based on data. Drug delivery behaviour and release characteristics is an outstanding issue that pharmaceutical formulation is yet to solve, because the drug properties, formulation parameters and physiological conditions interact nonlinearly. Conventional experimental methods are time-consuming, costly and not always able to analyse a multidimensional data. Machine learning methods have become useful tools in the recent years to model this type of a complex relationship and enhance predictive power in pharmaceutical systems.
This paper is a comparative analysis of several machine learning algorithms to predictive model the drug delivery systems. Various supervised learning models such as Decision Tree, Random Forest, Support Vector Machine, K-Nearest Neighbour, and Artificial Neural Networks were used to test their ability to predict drug release behaviour and formulation behaviour. The study dataset includes the important variables of the formulation, including polymer composition, solubility of drugs, particle size and environmental factors that influence the rate of drug release. The pre-processing method of data such as normalization, feature selection, and cross-validation was used to increase the reliability of the models.
The experimental findings suggest that ensemble based learning models, especially the random forest algorithm and gradient boosting algorithms, have superior prediction performance over conventional algorithms. The neural network models have also high ability of capturing nonlinear interactions of the formulation variables. The relative analysis shows the strengths and weaknesses of various algorithms in predicting pharmaceutical tasks. The results indicate that by incorporating machine learning algorithms into drug delivery studies, it is possible to save a lot of time conducting experiments, speed up the process of formulations development, and implement smart design of pharmaceuticals. The research offers a computation framework that can guide researchers and pharmaceutical developers in choosing suitable machine learning approaches to predictive drug delivery modeling and optimization.
Keywords: Machine Learning, Drug Delivery Systems, Predictive Modeling, Artificial Neural Networks, Random Forest Algorithm, Pharmaceutical Data Analysis.
How to cite this article: Shah AV, Shah PK, Pandya HB. Comparative Analysis of Machine Learning Algorithms for Predictive Drug Delivery Systems. Int J Drug Deliv Technol. 2026;16(15s): 190-197. DOI: 10.25258/ijddt.16.15s.22
Source of support: Nil.
Conflict of interest: None