Issue link: https://www.e-digitaleditions.com/i/1481708
18 Pharmaceutical Technology ® Trends in Formulation 2022 eBook PharmTech.com Formul ation and drug delivery spite the advantages offered by DoE, this tool could be unavailing if the data set generated from the ex- perimental design is not fittingly coupled with the equitable statistical methodology [e.g., analysis of variance (ANOVA)] and/or machine learning appli- cation (e.g., multiple linear regression model). Ma- chine learning is a statistical offshoot of artificial intel ligence (AI) which is broad ly said to im itate intel ligence of human behav ior. It is a process of using the existing data to learn (i.e., leverage data and employ mathematical models to make predic- tions, improve performance, and conduct effective model-informed decisions). Ever y ML application deploys a model based on the data (text, numbers, sound, images, etc.), but not all ML applications are suitable for every kind of data. And of course, 'more the merrier' aligns with higher number of sample sets or data sets required to effectively run a model and improve the predictabilit y of the model. Cur- rently, application of ML in pharmaceutical organi- zations has been practiced and effectively utilized at various level of drug discovery and clinical trials stages. However, its application is highly obscure and poorly understood within the formulation de- velopment and process optimization spaces. By ex- ploiting and utilizing the predictive abilit y of ML, formulation scientists might be able to effectively harness, optimize the multistep multifactorial pro- cess, and streamline the experimental design with the aid of existing datasets." Dr. Rathod is not a lone or work ing in isolation, he points out, "With the growing interests, some researchers, scientists, and process engineers have demon st rated t he appl icat ion of M L v ia va r ious M L a lgor it h m s such a s super v ised a nd u n super- vised learning. ML algorithms have been deployed to opt i m i ze t he d r ug del iver y s ystem s, [such a s nanoparticulate systems-artificial neural network (1), lipid-surfactant systems-Least absolute shrink- age selector operator (2), 3D pr intabi lit y of medi- cines- nearest neighbors, support vector machines, random forests, neural networks (3), prediction of physical stabilit y of formulations (eight machine learning approaches were compared) (4), and test- ing and optimizing in-vitro performance (dissolu- tion, disintegration time, etc.) (5), and various oth- ers]. Continuous improvement in pharmaceutical applications can be made by tactical use of histori- cal data in ML. Overall, ML can help in providing a direction to design formulation and make ML nav- igated formulation developments ef fectively and rapidly." References 1. H. Asadi et al., Journal of Microencapsulation, 28 (5) (Taylor and Francis, 2011). 2. V. Patel a nd V. Rat hod, "Appl icat ion of Compu- t at ion a l Mac h i ne L ea r n i ng Tool s to Compa re Pred ict ive Per for ma nce on Cr it ica l Qua l it y At- tributes of Spray Dried Lipid-based Formulation," AAPS ePoster Library October 2020. 3. M. Elbadawi, International Journal of Pharmaceuti- cals (November 2020). 4. R. Han, Journal of Control Release, October 2019. 5. J. Bourquin, European Jour nal of Phar maceutical Science (October 1998). ■ Drug Solutions Podcast: Overcoming Poor Solubility Through Accelerated Dissolution Pharmaceutical Technology presents the Drug Solutions podcast, where the editors chat with industry experts from across the pharmaceutical and biopharmaceutical supply chain. In this episode, Fernando Muzzio, distinguished pro- fessor, Chemical and Biochemical Engineering, Rutgers University, provides greater insight into a novel approach to improving drug solubility—continuous melt coating. Fernando talks about the technique in some detail, a little about some of the results that have been achieved in research, and provides information about the current patent filing and what companies can do to start working with this new approach. Click here to listen to the podcast.