Predictive Modeling of Bacterial Inactivation With Hydrogen Peroxide Over a Cobalt Ferrite Catalyst

Area of Science:

• Environmental Science
• Chemical Engineering
• Microbiology

Background:

• Hydrogen peroxide activated by cobalt ferrite generates hydroxyl radicals for bacterial inactivation.
• A continuous-flow packed-bed reactor system was developed for cobalt ferrite-catalyzed disinfection.
• Bacterial inactivation kinetics exhibited high nonlinearity, challenging conventional modeling.

Purpose of the Study:

• To evaluate the predictive accuracy of machine learning algorithms for bacterial disinfection using cobalt ferrite and hydrogen peroxide.
• To compare the performance of gradient boosting (GB) and random forest (RF) models against traditional kinetic models.
• To develop a predictive framework for optimizing disinfection parameters in continuous-flow systems.

Main Methods:

• A continuous-flow packed-bed reactor with cobalt ferrite was used for bacterial disinfection experiments.
• The classical Weibull model was applied to describe inactivation kinetics.
• Ten machine learning algorithms, including GB and RF, were trained and validated using experimental data.
• A combined GB+RF model was developed by averaging predictions.

Main Results:

• The Weibull model accurately described initial inactivation at low H2O2 concentrations but failed at higher concentrations and longer times.
• Gradient boosting (GB) and random forest (RF) algorithms demonstrated superior performance in predicting bacterial inactivation.
• GB showed better generalization, while RF excelled under moderate oxidative stress.
• A combined GB+RF model achieved balanced accuracy with average errors below ±10% at low-to-moderate peroxide concentrations.

Conclusions:

• Conventional kinetic models have limitations in predicting cobalt ferrite-catalyzed peroxide disinfection.
• Machine learning, particularly combined GB+RF models, offers a robust predictive framework for optimizing disinfection.
• The study provides a validated approach for selecting optimal hydrogen peroxide doses and contact times for effective bacterial inactivation.