Why it matters
Without AI, subtle signs such as flashing, early lesions, or fin rot can be missed, delaying treatment, risking spread, triggering false alarms, or compromising accreditation.
From design to safety
Design
Designing the diagnostic workflow
Define data sources, camera coverage, warning thresholds, and records to support early disease detection and compliant health documentation. Map data provenance, retention rules, and audit trails across quarantine tanks and veterinary hospital rooms.
- Underwater camera array
- LED lighting panel
- Multiparameter water quality probe
Build
Building data pipelines
Install cameras, calibrate lighting, integrate sensors, and configure AI processing rules with quarantined records for provenance. Establish secure data storage, versioned AI models, and access controls.
- Underwater camera array
- LED lighting panel
- Multiparameter water quality probe
- UV sterilizer unit
Operation
Real-time disease monitoring
Automated detection of white spot, fin rot, lesions, flashing, and early alerts; generate daily health summaries, outbreak flags, and discipline-specific documentation in the fish health log.
- Underwater camera array
- Computer vision processing server
- Quarantine tank filtration
Maintenance
Maintain detection fidelity
Regular calibration of cameras, revalidation of models, sensor recalibration, and refresh of health records; run weekly synthetic tests, track drift, and archive obsolete data.
- Underwater camera array
- Multiparameter water quality probe
- UV sterilizer unit
Safety
Safety and compliance checks
Alarms require sign-off by aquatic veterinarian and dive safety officer; ensure accreditation documentation remains current, annual reviews completed, and critical incident logs accessible to inspectors.
- Quarantine tank filtration
- Computer vision processing server
Part by part
- 1
Underwater camera array
- Watch for
- Low light, blur, occlusion, fogging lenses, algae growth on glass; lens contrast loss, moisture intrusion, unexpected glare, and reflective surfaces compromising identification
- AI action
- Flag frames to re-focus, trigger lighting refresh, stabilize exposure, and store key frames for review by the veterinary imaging team
- 2
LED lighting panel
- Watch for
- Color shifts, flicker, non-uniform illumination, shadowing of behavior, thermal drift affecting color calibration
- AI action
- Adjust exposure, white balance, and gain; run color calibration cycles and maintain a log of lighting changes tied to image quality metrics
- 3
Multiparameter water quality probe
- Watch for
- Nitrate rise, ammonia spike, pH drift, temperature variance, salinity instability
- AI action
- Raise alarm, suggest partial water changes, log parameter trend, trigger quarantine protocol if thresholds exceed setpoints
- 4
UV sterilizer unit
- Watch for
- UV lamp aging, reduced irradiance, flow blockage, fouling on quartz sleeve
- AI action
- Increase lamp run-time alert, adjust flow or schedule lamp replacement, document maintenance event in asset log
- 5
Quarantine tank filtration
- Watch for
- Pressure drop, filter clog, pump vibration, biofilm accumulation, flow rate deviation
- AI action
- Trigger maintenance request, schedule filter clean or replacement, record filter health in asset ledger
- 6
Computer vision processing server
- Watch for
- Processing latency, model drift, data queue backlog, disk fullness, network latency spikes
- AI action
- Flag processing delay, re-train model, scale compute resources, log AI pipeline status
Worked example
A fin-rot early warning in the quarantine tank
Illustrative scenario with sample data.
An AI system detects subtle fin edge erosion and irregular swimming that precedes clinical signs, prompting early containment and veterinary review.
Signal in
camera_id: CQ-12, timestamp: 2026-08-14T09:15Z, image_quality: good, exposure: 0.42, fish_id: F-317, species: Dwarf Angelfish, context: quarantine tank, temp_C: 26.4, ammonia_ppm: 0.02, nitrate_ppm: 5, light_level: 650 lux, tank_id: QT-04
System out
alert_id: ER-4832, alert_type: early_disease, confidence: 0.82, suggested_action: isolate fish, initiate targeted observation for 48h, record in fish_health_log, request veterinary review
Human sign-off: Aquarium veterinary nurse reviews alert, confirms signs are consistent with early fin rot, approves isolation and observation plan, updates health records, and schedules recheck with the aquatic veterinarian.
How to roll it out
Phase 1
Weeks 1 to 4
Baseline deployment of cameras and sensors, establish data flows, and build initial AI alert rules.
Phase 2
Weeks 5 to 12
Calibrate models to exhibit-specific lighting and housing; validate incident workflows with staff.
Phase 3
Weeks 13 to 20
Stabilize operations, implement escalation paths, and document health records integration.
Where people stay in charge
- Decisions on diagnostic labeling, treatment plans, and animal welfare actions must be signed by a licensed aquatic veterinarian.
- All modifications to alert thresholds or model retraining must be reviewed by the veterinary team and the life support supervisor.
- Critical alarms and human-off actions (isolation, medical treatment) must be approved by the dive safety officer or curator.
Questions owners and crews ask
How does AI help detect early disease signs in quarantine fish?
AI analyzes camera imagery for patterns like flashing, edge erosion, and unusual swimming; it flags likely cases and prompts health records updates, with veterinary sign-off for medical actions.
What data feeds are required to run this system?
High-resolution underwater cameras, lighting calibration data, water quality sensor readings, and a central processing server for analytics; all feeds must be time-synced and access-controlled.
Who signs off on treatment recommendations generated by the AI?
The aquatic veterinarian makes final medical decisions, with sign-off from the life support supervisor for facility compliance and record integrity.
How are privacy, safety, and regulatory needs addressed?
All alerts and actions are logged in the fish health records; escalation and accreditation requirements drive workflows and audits, with periodic external inspections.
Sources and further reading
- EAZA: Animal HealthVeterinary hospital rooms guidance on quarantine, isolation housing, and veterinary care for aquarium animals.
- WOAH: Codes and ManualsWhite spot disease reporting and control measures in the Aquatic Animal Health Code.
- dvm360: Fish MedicineClinical fish medicine on fin edge erosion, water quality disease, and treatment for aquarium fish.
- International Association for Aquatic Animal MedicineProfessional network for aquatic veterinarians, with conferences and resources on fish disease diagnosis.
- Merck Veterinary Manual: Exotic and Laboratory AnimalsClinical reference on early lesions, fin disease, and diagnostic features in fish species.
- EAZA: AccreditationAccreditation standards and documentation expectations for zoos and aquariums, including record currency.
- GOV.UK: Standards of Modern Zoo PracticeStatutory zoo licensing standards covering records of incidents, animal care, and safety obligations.
- EAZA: Animal WelfareWelfare assessment and positive welfare practice guidance for zoo and aquarium animals.
- MBARI: TechnologiesOcean sensors, underwater imaging, and machine learning tools for marine science monitoring.
- HSE: Diving at WorkDiving at Work Regulations guidance on dive team roles, supervision, and diver safety duties.
- Aquatic Animal Life Support Operators (AALSO)Life support system operation, water quality, training, and certification for operators.