Home Cosmetic Science Talk Formulating General Quantitative Structure-Activity Relationship (QSAR) Modeling for Cosmetic Product Formulation

  • Quantitative Structure-Activity Relationship (QSAR) Modeling for Cosmetic Product Formulation

    Posted by spadirect on November 30, 2020 at 4:48 pm

    Have you used Quantitative Structure-Activity Relationship (QSAR) modeling in order to predict which ingredients and ingredient combinations would most likely achieve target physico-chemical and sensorial attributes for your product in development; and thereby potentially speed up the formulation process?

    In other words, have you used QSAR software to input candidate ingredients’ chemical characteristics; i.e., viscosity, refractive index, coefficient of friction, etc., to determine with greater speed and accuracy the most promising ingredient candidates and ingredient blend ratios in order to achieve your target product’s desired sensorial profile?

    What are some software tools available to help cosmetic chemists run QSAR modeling to inform ingredient selection and optimize blending ratios before hitting the bench to begin actual lab formulation work?

    Any experiences, comments and discussion about QSAR modeling for use in cosmetic product formulation would be greatly appreciated. Thanks!

    Pharma replied 3 years, 7 months ago 3 Members · 2 Replies
  • 2 Replies
  • OldPerry

    November 30, 2020 at 6:16 pm

    No, I have not used the software. I have used Design of Experiment software for formulating which has similar aims. The idea is that you can plug in some information about ingredients and the computer will spit out an appropriate formula.  It was never too helpful.

    The biggest problem with this approach to formulating is that there is not a simple connection between a measured variable and consumer preference / acceptability. 

    For example, how could you answer the question “what should the viscosity of a shampoo be?”  Some people might want a thick formula, others want a thinner formula. There is no numerically optimal level. 

    The same is with things like coefficient of friction. While you can get a number measurement in the lab, there is no good way to correlate it with consumer preference. Some consumers will think a formula is too greasy, others will think the same formula is not slippery enough. 

    And to complicate things more, there is no way to correlate a lab measurement with the performance of an ingredient over days or weeks. How would a consumer know whether their condition is improved because of a treatment or it just improved because of time.

    Until you can get a numerical lab measurement that predicts consumer preference & performance, software algorithms like these won’t be much help in formulating.

  • Pharma

    December 1, 2020 at 8:54 am
    QSAR is as suitable for calculating/estimating cosmetic formulations as quantum mechanics is for making vanilla ice cream.
    If you want to spend money in software which might help consider getting familiar with the HLD principle and then, if you’re filthy rich, get a COSMO-RS software. It’s still not very helpful but certainly better suited than QSAR.
    Else, properly design your experiments to speed up development time. Cosmetic ‘science’ is still an experimental science with a lot of hands on work required because there simply is no suitable program. If you do basic research on a simple system or a new ingredient, sure programs help (but not QSAR which is used in development of new active pharmaceutical entities) but no program can predict the fine differences needed in a, say, 20-30 compounds long LOI to meet customer expectations.

Log in to reply.