A quantitative size stability metrics for long-acting suspensions and its prediction with machine learning

Area of Science:

• Pharmaceutical Sciences
• Materials Science
• Chemical Engineering

Background:

• Suspension stability is critical for nano- and microsuspensions in long-acting injectables.
• Current methods rely on subjective trial-and-error, lacking quantitative metrics.
• Particle agglomeration and Ostwald ripening are key challenges in formulation development.

Purpose of the Study:

• To propose a quantitative measurement for suspension stability (Suspension Stability Index, SSI).
• To develop predictive models for SSI using D50CV and SpanCV.
• To establish a more objective and comparable method for assessing suspension stability.

Main Methods:

• Quantitative measurement of suspension stability using D50CV and SpanCV over 28 days under accelerated conditions.
• Construction of predictive models for SSI, including Support Vector Classification (SVC).
• Evaluation of model performance using 5-fold cross-validation, achieving 0.82 accuracy and F1 score.

Main Results:

• The Support Vector Classification (SVC) model demonstrated the best performance for predicting SSI.
• Stabilizer type and concentration were identified as critical factors impacting SSI.
• The drug compound showed interaction with multiple parameters, except bead size.

Conclusions:

• The proposed quantitative metric (SSI) offers a more objective approach to assessing suspension stability.
• The developed classification model can serve as a screening tool in formulation development.
• This approach facilitates more comparable investigations of suspension stability for long-acting injectables.