Predictive Modeling of Channel Catfish Under Varying Temperatures: Quality Dynamics and Warning Thresholds

Area of Science:

• Food Science
• Microbiology
• Data Science

Background:

• Channel catfish quality degrades during storage, impacting safety and shelf-life.
• Accurate prediction of quality changes is crucial for effective cold-chain management.
• Microbial growth (Total Viable Count - TVC) and biochemical changes (Total Volatile Basic Nitrogen - TVB-N) are key indicators of spoilage.

Purpose of the Study:

• To develop mathematical models and artificial neural networks for predicting channel catfish quality during storage.
• To establish early warning thresholds for quality degradation.
• To provide tools for optimizing cold-chain distribution.

Main Methods:

• Modified Gompertz and Belehradek equations were used to model microbial growth (TVC).
• Arrhenius equation combined with primary models predicted Total Volatile Basic Nitrogen (TVB-N) changes.
• Back Propagation Neural Networks (BPNN) and Radial Basis Function Neural Networks (RBFNN) were employed for predictive modeling.

Main Results:

• Secondary kinetic models showed good agreement with experimental data (within ±20% error, ±15% for 2-6 days).
• Artificial neural network models achieved high accuracy (R² > 0.9) in predicting quality changes.
• Early warning thresholds were determined based on kinetic model slopes.

Conclusions:

• Developed predictive models offer a scientific basis for channel catfish quality monitoring.
• The models support effective cold-chain distribution strategies for channel catfish.
• This research enhances food safety and reduces post-harvest losses in the catfish industry.