A supervised machine-learning analysis of doxorubicinloaded electrospun nanofibers and their anticancer activity capabilities

Mohammadreza Rostami ¹ Maliheh Gharibshahian ² Mehrnaz Mostafavi ³ Ali Sufali ⁴ Mahsa Golmohammadi ⁵ MohammadReza Barati ⁶ Reza Maleki ⁶ Nima Beheshtizadeh ^7*

• ¹ Tehran University of Medical Sciences, Tehran, Tehran, Iran
• ² Semnan University of Medical Sciences, Semnan, Semnan, Iran
• ³ Shahid Beheshti University of Medical Sciences, Tehran, Tehran, Iran
• ⁴ Universal Scientific Education and Research Network, Tehran, Tehran, Iran
• ⁵ Amirkabir University of Technology, Tehran, Tehran, Iran
• ⁶ Iranian Research Organization for Science and Technology, Tehran, Tehran, Iran
• ⁷ Tabriz University of Medical Sciences, Tabriz, Iran

Thanks to the diverse advantages of electrospun nanofibers, multiple drugs have been loaded in these nanoplatforms to be delivered healthily and effectively. Doxorubicin is a drug used in chemotherapy, and its various delivery and efficacy parameters encounter challenges, leading to the seeking of novel delivery methods. Researchers have conducted numerous laboratory investigations on the encapsulation of doxorubicin within nanofiber materials. This study employed a supervised machine-learning analysis to extract the influencing parameters of the input from quantitative data for doxorubicin-loaded electrospun nanofibers. The study also determined the significance coefficient of each parameter that influences the output results and identified the optimum points and intervals for each parameter. Our Support Vector Machine (SVM) analysis findings showed that doxorubicin-loaded electrospun nanofibers could be optimized by employing a machine learning-based investigation on the polymer solution parameters (such as density, solvent, electrical conductivity, and concentration of polymer), electrospinning parameters (such as voltage, flow rate, and distance between the needle tip and collector), and our study parameters, i.e., drug release and anticancer activity, which affect the properties of the drug-loaded nanofibers, such as the average diameter of fiber, anticancer activity, drug release percentage, and encapsulation efficiency. Our findings indicated the importance of factors like distance, polymer density, and polymer concentration, respectively, in optimizing the fabrication of drug-loaded electrospun nanofibers.The smallest diameter, highest encapsulation efficiency, highest drug release percentage, and highest anticancer activity are obtained at a molecular weight between 80 and 474 kDa and a doxorubicin concentration of at least 3.182 wt.% with the polymer density in the range of 1.2-1.52 g/cm3, polymer concentration of 6.618-9 wt.%, and dielectric constant of solvent more than 30. Also, the optimal distance of 14-15 cm, the flow rate of 3.5–5 mL/h, and the voltage in the range of 20–25 kV result in the highest release rate, the highest encapsulation efficiency, and the lowest average diameter for fibers. Therefore, to achieve optimal conditions, these values should be considered. These findings open up new roads for future design and production of drug-loaded polymeric nanofibers with desirable properties and performances by machine learning methods.