Rivaroxaban, commonly used as an anticoagulant, is known for its low aqueous solubility, which limits its bioavailability and therapeutic effectiveness. A recent study published by researchers from the University of Ha’il has utilized innovative techniques to optimize the solubility of rivaroxaban using mixed solvent systems and advanced machine learning models.
To overcome the solubility challenges associated with rivaroxaban, this research investigated its solubility behavior in various mixed solvent systems—specifically, primary alcohols combined with dichloromethane. The authors aimed to identify optimal conditions for solubility enhancement which is pivotal for ensuring the drug's efficacy. The significance of enhancing solubility lies not only in improving drug absorption but also minimizing the risks associated with administering higher doses of poorly soluble drugs.
Machine learning was employed to model the solubility data, allowing researchers to predict the influences of temperature, solvent type, and the composition ratio of dichloromethane on rivaroxaban solubility. Three distinct machine learning approaches were assessed: AdaBoost Gaussian Process Regression (ADAGPR), AdaBoost Multilayer Perceptron (ADAMLP), and AdaBoost LASSO Regression (ADALASSO). Among these, ADAGPR yielded the best predictive performance with high accuracy metrics, achieving an R² score of 0.96485, compared to 0.96022 for ADAMLP and 0.88712 for ADALASSO.
The researchers found the optimal solubility conditions for rivaroxaban to be at 308.15 K, with the best performance observed at a mass fraction of 0.8190 of dichloromethane and methanol mixture, where they predicted solubility to be approximately 4.2 × 10-3 g/mL. This result highlights the significant interplay between temperature and solvent composition on rivaroxaban solubility, providing clear indications for pharmaceutical scientists working on drug formulation to achieve effective solubility outcomes.
Machine learning, particularly ADAGPR, proved most effective, capturing complex nonlinear patterns which would typically require extensive experimentation to identify. The integration of such sophisticated modeling approaches showcases the advancement toward rapid and reliable pharmaceutical formulations. The authors emphasized: “Advanced machine learning techniques are pivotal for addressing complex solubility prediction challenges within binary solvent systems.”
Significant insights were also gleaned from the relationship between solubility and solvent composition. The study illustrated how optimizing solvent systems using the correct ratios of dichloromethane and methanol can greatly influence the dissolution rate of rivaroxaban, indicating stronger solute-solvent interactions at certain conditions due to the complementary properties of these solvents. “Solubility exhibited a nonlinear response, with optimal solubility observed at specific temperature and mass fraction combinations, beyond which solubility decreased,” explained the authors.
This research holds substantial promise for the pharmaceutical industry, particularly for designing effective drug delivery systems for poorly soluble medications. The findings could pave the way for future work as researchers may seek to expand these models by analyzing additional solvent systems, refining the conditions for achieving maximum solubility.
Overall, this ambitious approach not only enhances our current knowledge of rivaroxaban’s solubility profile but emphasizes how advanced computational techniques can bridge the gap between theoretical predictions and practical applications, demonstrating significant potential to improve therapeutic outcomes for patients requiring anticoagulation therapy.
With the pharmaceutical sector continually striving for innovations, the methodologies and insights derived from this study set the stage for future explorations aimed at optimizing drug formulations and enhancing bioavailability for various pharmaceutical applications.