Why it matters
Without AI scoring, welfare concerns may be missed during daily checks, leading to undetected stress, inconsistent records, and failed accreditation inspections due to gaps in monitoring and action planning.
From design to safety
Design
Define welfare scoring framework
Define five welfare domains (behavior, physiology, environment, social interactions, health indicators), data streams (video, sensors, health records, enrichment logs), and thresholds; map continuous data sources to observable indicators; establish governance for scores, escalation rules, and action plans across exhibits and quarantine facilities; specify retention and access controls.
- welfare assessment toolkit
- RGB video analytics camera
- infrared thermography camera
- RFID tag reader
Build
Integrate data streams and interfaces
Install cameras, thermal cameras, accelerometers, pressure and oxygen sensors, pH and salinity probes, data loggers, edge devices; configure data pipelines (MQTT, REST, SQL), time synchronization (NTP), secure storage (encrypted buckets), and data retention policies for longitudinal welfare scoring and audit trails.
- behavior sensors
- data logger
- quarantine cage door sensor
Operation
Real-time welfare scoring and response
AI computes domain scores, flags anomalies, assigns risk levels, and triggers care plans; dashboards surface animal-level and exhibit-level views with per-animal line charts, enrichment recommendations, and veterinary prompts for intervention.
- ethogram database server
- welfare assessment toolkit
- RGB video analytics camera
Maintenance
Maintain scoring accuracy and validity
Regularly recalibrate thresholds using veterinary input, validate sensor drift, audit data integrity, update action plans; schedule quarterly governance reviews, backfill gaps from archived feeds, and verify regulatory alignment.
- infrared thermography camera
- RFID tag reader
- behavior sensors
Safety
Governance and accountability
Sign-offs by veterinary team and accreditation officer on scores, interventions, and records; maintain audit trails; ensure documentation meets accreditation schemes and site-specific SOPs for welfare management.
- quarantine cage door sensor
- ethogram database server
- data logger
Part by part
- 1
welfare assessment toolkit
- Watch for
- drift in domain scores due to sensor drift or misalignment in behavior interpretation
- AI action
- AI recalibrates domain weights and flags drift to staff for review.
- 2
RGB video analytics camera
- Watch for
- persistent gaps in activity coverage or occluded views affecting behavior detection
- AI action
- AI initiates camera reorientation or additional camera deployment.
- 3
infrared thermography camera
- Watch for
- unusual surface temperature hot spots indicating stress or inflammation
- AI action
- AI flags thermography anomaly and prompts veterinary check.
- 4
RFID tag reader
- Watch for
- mismatch between animal presence and tagged identities during feed or health checks
- AI action
- AI prompts cross-check with manual roster and updates records.
- 5
behavior sensors
- Watch for
- reduced activity or abnormal posture patterns
- AI action
- AI triggers welfare score adjustment and caregiver notification.
- 6
data logger
- Watch for
- missing time-stamped data from any sensor
- AI action
- AI generates data completeness report and alerts system admin.
Worked example
Catching a welfare drift before a transfer
Illustrative scenario with sample data.
An exhibit shows rising stress indicators during a routine transfer window; AI detects early signals and prompts a welfare review before the transfer completes.
Signal in
time:2026-11-07 09:15:00, exhibit:OceanView, animal_id:AJ-12, domain_scores:behavior:0.72, physiology:0.58, environment:0.65, social:0.60, health:0.55, prior_event:transfer_planned, sensor_status:ok, data_latency:2.3s, water_temp:24.6C, salinity:35ppt
System out
alert_type: Welfare Drift; threshold: domain behavior exceeds 0.70; suggested_action: pause transfer, reassess destination, log welfare note; nurse roster updated; action_required_by: aquatic veterinarian; follow_up_time: 24h; enrichment_adjustment: increase foraging challenge
Human sign-off: The curator and aquatic veterinarian review the drift alert, confirm enrichment and transfer plan adjustments, and authorize or modify the transfer schedule, while the accreditation officer files the welfare note for the site file.
How to roll it out
Phase 1
Weeks 1 to 4
Baseline data capture with all sensors; verify data flow and initial score mappings.
Phase 2
Weeks 5 to 8
Impact assessment and threshold tuning; establish standard operating procedures for welfare interventions.
Phase 3
Weeks 9 to 12
Full governance cycle with veterinary sign-off and accreditation documentation templates.
Where people stay in charge
- Decisions on welfare interventions must be signed by the aquatic veterinarian or equivalent qualified authority.
- All welfare data, scores, and actions must be retained in the official life support and accreditation records.
- Any changes to thresholds or domain weights require veterinary committee review and senior curator approval.
Questions owners and crews ask
What is AI animal welfare scoring used for in collection planning?
AI welfare scoring informs planning by highlighting animals needing priority observation, guiding enrichment, and shaping transfer decisions within the collection. It is the primary trigger for documented care plans.
How does the system handle data gaps in welfare indicators?
The system flags missing data, prompts manual reconciliation, and escalates data integrity reviews to the life support and veterinary teams before any action.
Who approves changes to welfare action plans?
The aquatic veterinarian signs off on action plans; the curator reviews alignment with collection goals and accreditation standards.
How are accreditation requirements integrated?
Scores and interventions are logged into the records system, with reports generated for audits and site visits to demonstrate welfare governance.
Sources and further reading
- WAZA: Approach to Animal WelfareExplains the global five domains framework for assessing welfare across zoo and aquarium animals.
- BIAZA: AccreditationDescribes how UK and Irish aquariums are assessed against welfare and conservation standards during inspections.
- GOV.UK: Standards of Modern Zoo PracticeSets out statutory licensing standards for animal care, safety, and records that local procedures must meet.
- International Association for Aquatic Animal MedicineConnects readers with aquatic animal veterinary medicine resources, conferences, and a professional network of aquarium vets.
- MBARI: TechnologiesCovers ocean imaging and sensing technologies, including thermal imaging used in marine science monitoring.
- Species360: MissionDescribes the global animal records network and the data standards used for collection records.
- Merck Veterinary Manual: Exotic and Laboratory AnimalsOffers a clinical reference on signs of illness and physiology in fish and other non-domestic species.
- Aquatic Animal Life Support Operators (AALSO)Covers operator training, certification, and water quality practice for aquarium life support systems.
- EAZA: AccreditationShows the EAZA accreditation programme standards and screening that accreditation officers work against.
- EAZA: Animal HealthProvides zoo and aquarium animal health guidance, including disease, quarantine, and veterinary monitoring practice.