Anesthesia is an essential and routine practice in modern aquaculture. From handling and transport to artificial reproduction, controlled sedation is key to minimizing stress in fish and ensuring their welfare. In recent years, the industry has shifted towards using natural anesthetics due to their food safety and more favorable physiological effects. However, one challenge remains: how to determine the perfect dose for each species?
An innovative study published in Frontiers in Veterinary Science by researchers at Recep Tayyip Erdogan University explores the potential of nutmeg oil (Myristica fragrans) and, for the first time, utilizes machine learning to predict its effects.
Key findings
- The study demonstrates that Artificial Neural Networks (ANNs) can predict the induction times, recovery times, and hematological responses of fish to nutmeg oil with high accuracy.
- Sensitivity to the anesthetic varies significantly between species (common carp, Danube sturgeon, and rainbow trout), requiring species-specific predictive models.
- White blood cell (WBC) parameters were the most predictable across all three species, standing out as a reliable indicator of immune and stress status during anesthesia.
- AI modeling provides a tool for determining effective and safe doses (induction <3 min, recovery <5 min) without the need for additional invasive testing, thereby improving animal welfare.
The challenge of tailored anesthesia
Determining the optimal anesthetic concentration is a delicate balance. An insufficient dose fails to calm the fish, prolonging handling stress, while an excessive dose can be toxic and cause mortalities, leading to economic losses. Furthermore, the effective concentration is not universal; it varies drastically depending on the species, size, stress level of the fish, and even water quality.
Traditionally, establishing these doses has required numerous in vivo tests, observing behavioral and physiological responses. This new study proposes a revolutionary approach: using artificial intelligence to create predictive models that define safe and effective doses without the need for additional animal experimentation.
Nutmeg oil and neural networks: An innovative combination
The focus of this study is nutmeg oil, a natural compound whose anesthetic properties are primarily due to myristicin. To evaluate its efficacy, researchers used data from previous studies on three commercially significant freshwater species:
- Common carp (Cyprinus carpio)
- Danube sturgeon (Acipenser gueldenstaedtii)
- Rainbow trout (Oncorhynchus mykiss)
The core of the research was the development of Artificial Neural Networks (ANNs), a type of machine learning model that mimics the functioning of the human brain to identify complex relationships within data.
How did the predictive model work?
The scientists designed species-specific ANN models. The process was as follows:
Stay Always Informed
Join our communities to instantly receive the most important news, reports, and analysis from the aquaculture industry.
- Input: The sole input variable was the concentration of nutmeg oil.
- Output: The model was trained to predict six key parameters:
- Induction Time (IT): The time it takes for the fish to reach a deep state of anesthesia.
- Recovery Time (RT): The time it takes to return to normal behavior.
- Hematological Parameters: White blood cell (WBC) count, red blood cell (RBC) count, hemoglobin (HGB), and hematocrit (HCT), which are rapid indicators of the fish’s health and stress status.
Data from 18 fish per species were used to “train” the neural network, and data from another 12 fish were used to “test” and validate its predictive capabilities.
High-precision predictions and interspecies differences
The results were compelling. The ANN models successfully predicted both behavioral responses and hematological changes in all three species with very high accuracy (R² values generally exceeding 0.92).
Species-specific sensitivity
The study confirmed that each species responds differently to nutmeg oil, which was reflected in the architecture of the ANN models required for each one.
- Rainbow trout: The model for this species was the most successful in predicting induction time (IT), showing low and stable error rates.
- Common carp: This model also demonstrated high accuracy in predicting induction and recovery times.
- Danube sturgeon: Although the predictions were good, the model for this species showed slightly lower stability, which could indicate a particular physiological sensitivity to the anesthetic.
Hematology as a reflection of stress
Blood parameters are crucial biomarkers. The study found that:
- White Blood Cells (WBC): This was the most easily and consistently predicted parameter across all three species. WBCs are immune cells, and their count is a key indicator of stress and the fish’s health status. The ability to model their response is invaluable for monitoring animal welfare.
- Red Blood Cells (RBC), Hemoglobin (HGB), and Hematocrit (HCT): These components, responsible for oxygen transport, were also successfully modeled. Predicting their behavior makes it possible to distinguish whether changes in oxygenation are due to the anesthetic’s effect or to secondary hypoxia, a risk during sedation.
Implications for the future of aquaculture
This study is not just an academic exercise; it opens the door to significant practical applications. The ability to non-invasively predict the optimal anesthetic dose can transform aquaculture operations.
Producers could use these models as decision-support tools to determine the nutmeg oil concentration needed to achieve desired induction and recovery times (e.g., under 3 minutes to induce and under 5 to recover). This ensures both operational efficiency and the welfare of the fish during handling, sorting, or vaccinations.
In conclusion, the integration of artificial intelligence into experimental aquaculture provides a solid scientific foundation for developing safer, more ethical, and personalized anesthetic protocols for each species. This approach not only refines best practices but also drives the industry toward more sustainable and technologically advanced management.
Reference (open access)
Minaz, M., Alparslan, C., & Er, A. (2025). Using machine learning to predict anesthetic dose in fish: A case study using nutmeg oil. Frontiers in Veterinary Science, 12, 1652115. https://doi.org/10.3389/fvets.2025.1652115
Editor at the digital magazine AquaHoy. He holds a degree in Aquaculture Biology from the National University of Santa (UNS) and a Master’s degree in Science and Innovation Management from the Polytechnic University of Valencia, with postgraduate diplomas in Business Innovation and Innovation Management. He possesses extensive experience in the aquaculture and fisheries sector, having led the Fisheries Innovation Unit of the National Program for Innovation in Fisheries and Aquaculture (PNIPA). He has served as a senior consultant in technology watch, an innovation project formulator and advisor, and a lecturer at UNS. He is a member of the Peruvian College of Biologists and was recognized by the World Aquaculture Society (WAS) in 2016 for his contribution to aquaculture.
