Related Experiment Video
Updated: Aug 7, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.9K
Anomaly Detection in Biological Early Warning Systems Using Unsupervised Machine Learning
Aleksandr N Grekov1,2, Aleksey A Kabanov2, Elena V Vyshkvarkova1
1Institute of Natural and Technical Systems, 299011 Sevastopol, Russia.
Sensors (Basel, Switzerland)
|March 11, 2023
Summary
Bivalve mollusks like Unio pictorum can be used in automated systems to detect aquatic pollution in real-time. Machine learning methods, particularly Isolation Forest, efficiently identify anomalies in mollusk behavior, signaling pollution events.
Area of Science:
- Environmental Science
- Ecotoxicology
- Biomonitoring
Background:
- Aquatic pollution poses a significant threat to ecosystems.
- Real-time detection of pollution emergencies is crucial for environmental protection.
- Bivalve mollusks show potential as sensitive bioindicators in monitoring systems.
Purpose of the Study:
- To develop an automated monitoring system using bivalve mollusk behavior.
- To evaluate machine learning techniques for anomaly detection in biomonitoring data.
- To assess the efficiency of different algorithms in identifying pollution-related behavioral changes.
Main Methods:
- Utilized the behavior of Unio pictorum as bioindicators.
- Collected experimental data using an automated system from the Chernaya River.
- Applied unsupervised machine learning: elliptic envelope, Isolation Forest (iForest), one-class SVM, and Local Outlier Factor (LOF).
Main Results:
- Elliptic envelope, iForest, and LOF achieved an F1 score of 1 for anomaly detection with optimized hyperparameters.
- These methods successfully identified anomalies in mollusk activity data without false alarms.
- The iForest method demonstrated the highest efficiency in terms of anomaly detection time.
Conclusions:
- Bivalve mollusks are effective bioindicators for automated aquatic pollution monitoring.
- Machine learning algorithms can reliably detect emergency pollution situations through mollusk behavior.
- The developed system offers a promising tool for early warning of aquatic environmental hazards.
