Related Experiment Video
Updated: Jan 15, 2026
07:29
A Standardized Protocol for Preference Testing to Assess Fish Welfare
Published on: February 22, 2020
7.4K
AI-Driven Monitoring for Fish Welfare in Aquaponics: A Predictive Approach
Jorge Saúl Fandiño Pelayo1, Luis Sebastián Mendoza Castellanos2, Rocío Cazes Ortega1
1Facultad de Ciencias Naturales e Ingeniería, Unidades Tecnológicas de Santander (UTS), Bucaramanga 680005, Colombia.
Sensors (Basel, Switzerland)
|October 16, 2025
Summary
This study developed an AI system using environmental sensors to monitor aquaponic systems, improving fish welfare by detecting adverse conditions early. The random forest model achieved 99% accuracy in predicting fish health status.
Area of Science:
- Aquaculture
- Artificial Intelligence
- Environmental Monitoring
Background:
- Aquaponic systems require continuous monitoring for optimal fish health and productivity.
- Early detection of adverse water conditions is crucial for preventing fish stress and mortality.
Purpose of the Study:
- To develop an AI-powered predictive system for intelligent monitoring of aquaponic systems.
- To improve fish welfare through early detection of adverse water conditions using environmental sensing.
Main Methods:
- Integrated low-cost digital sensors to measure pH, dissolved oxygen, and temperature.
- Evaluated four supervised machine learning models: LDA, SVM, NN, and RF.
- Collected 1823 instances from a red tilapia aquaponic setup over eight months.
Main Results:
- The random forest model achieved the highest classification accuracy at 99%.
- Neural networks (98%) and SVMs (97%) also demonstrated high performance.
- LDA achieved 82% accuracy, with all models validated using cross-validation.
Conclusions:
- Sensor-based predictive models reliably detect early signs of fish stress or mortality.
- The developed AI system supports intelligent environmental monitoring and automation in sustainable aquaponics.
