Machine Learning-Based Framework Contributes to Sustainable Development in Pharmaceutical Research

Strategic Insights -

A novel approach to quantification of active pharmaceutical ingredients in ophthalmic preparations has been developed, addressing a significant challenge in sustainability research. The study, led by researchers from King Abdulaziz University, employs machine learning-enhanced UV-spectrophotometric chemometric models to concurrently quantify latanoprost, netarsudil, benzalkonium chloride, and two related compounds. This breakthrough method has far-reaching implications for the pharmaceutical industry, contributing meaningfully to several United Nations Sustainable Development Goals (UN-SDGs).

Key Takeaways:

  • The study presents a novel and sustainable approach to quantifying active pharmaceutical ingredients in ophthalmic preparations using machine learning-enhanced UV-spectrophotometric chemometric models.
  • A 25-mixture calibration set was created using a strategic multi-level, multi-factor experimental design, with the D-optimal design generated by MATLAB's candexch algorithm providing a robust validation set.
  • The optimized MCR-ALS model outperforms in predictive ability, with recovery percentages of 98-102% and low root mean square errors of calibration and prediction.
  • The Greenness Index Spider Charts and the Green Solvents Selection Tool were applied to replace hazardous solvents, contributing to a more eco-friendly profile.
  • The method's environmental and societal benefits were further validated using the Need, Quality, Sustainability (NQS) index.
  • The research concluded that the machine learning-based framework contributes meaningfully to ten UN-SDGs, highlighting its value for future-oriented pharmaceutical research.
  • The study's findings have the potential to enhance the sustainability of pharmaceutical research and development, aligning with the goals of the pharmaceutical industry and regulatory agencies.

Statistics:

  • 25-mixture calibration set was created using a strategic multi-level, multi-factor experimental design.
  • Recovery percentages: 98-102%
  • Low root mean square errors of calibration and prediction
  • Mean square error of prediction
  • Relative root mean square error within acceptable limits
  • Limits of detection for pharmaceutical analysis: 0.01-0.1 mg/L
  • Seven advanced evaluation tools were employed to assess the method's greenness, blueness, violetness, and whiteness.

Sources:

  • "D-optimal candexch algorithm-enhanced machine learning UV-spectrophotometry for five-analyte determination in novel anti-glaucoma formulations and ocular fluids: four-color sustainability framework with NQS assessment and UN-SDG integration." BMC Chemistry, 2025;19(1):198.
  • King Abdulaziz University
  • BMC Chemistry can be contacted at: Bmc, Campus, 4 Crinan St, London N1 9XW, England.
  • Lateefa A. Al-Khateeb, Dept. of Chemistry, Faculty of Science, King Abdulaziz University, P.O. Box 80203, 21589, Jeddah, Saudi Arabia.