AI Integration for IDEXX Neo Preventive Care

Traditional preventive care in IDEXX Neo relies on static, rule-based schedules for reminders (e.g., "annual wellness exam for dogs over 7"). AI integration transforms this by connecting to Neo's Patient, Client, and Service History data objects via its API. An AI model analyzes hundreds of signals—including breed-specific disease predispositions, local parasite prevalence from lab data, vaccination titers, previous compliance history, and even client communication responsiveness—to calculate a dynamic, individualized risk score and predict the optimal time for a next service.

The implementation typically involves a secure, event-driven architecture: a background service polls or receives webhooks from Neo for eligible patients. For each patient, it retrieves the relevant clinical and client data, passes it to a risk-scoring model, and returns a recommendation payload. This payload can trigger several automated workflows within Neo: creating a Preventive Care Campaign for a cohort of high-risk patients, generating a personalized draft Reminder with a specific, evidence-based call-to-action, or even flagging the patient record for the veterinarian's review during their next scheduled visit. The goal is to shift from "your pet is due" to "based on Fluffy's breed, weight trend, and local leptospirosis cases, we recommend a wellness panel and lepto booster next month."

Rollout requires careful governance. Initial models should be configured for vet-in-the-loop approval, where AI-generated campaign lists or reminder text are presented in a dedicated queue within Neo for clinical staff review before sending. This builds trust and ensures medical oversight. Over time, as confidence grows, low-risk, high-compliance reminders (like routine heartworm prevention in endemic areas) can be fully automated, while complex medical recommendations remain assisted. This approach turns IDEXX Neo from a reminder system into an intelligent care coordination hub, increasing client engagement and practice revenue by intervening at the right time for the right patient.

The integration architecture centers on the Patient and Preventive Care Plan objects within IDEXX Neo. A background service, typically deployed as a secure container, polls Neo's API for patients due for preventive care reviews. For each patient, it extracts a consolidated data payload including: species, breed, age, weight history, previous diagnostic results, medication history, lifestyle flags (e.g., indoor/outdoor), and geographic location. This data is sent to a hosted AI model—which could be a fine-tuned open-source LLM or a managed service like OpenAI—that outputs a personalized risk score and a recommended service schedule (e.g., "Dental prophylaxis recommended in 4 months due to breed predisposition and mild gingivitis noted 6 months ago").

The generated recommendations are posted back to Neo via API, creating or updating Preventive Care Plan records and triggering Neo's native reminder engine. The key integration points are:

• Data Ingestion Hook: A scheduled job using Neo's GET /patients and related clinical APIs.
• Risk Model Service: A secure endpoint that receives patient JSON, runs the model, and returns structured recommendations.
• Plan Update Action: Neo's POST /preventive-care-plans API to create actionable items linked to the patient record.
• Audit Log: All model inferences, input data, and user overrides are written to a separate audit database for compliance and model retraining.

This flow transforms preventive care from a simple calendar-based alert ("Fido is due for his annual") into a dynamic, evidence-based recommendation system.

Rollout is typically phased, starting with a pilot species (e.g., canine patients) and a single high-value service like dental care. Governance is critical: all AI-generated plans are flagged as "Draft - Veterinarian Review Required" in Neo, requiring a clinician's sign-off before reminders are sent to clients. This human-in-the-loop step ensures medical oversight, builds staff trust, and creates a feedback loop where vet overrides are used to refine the model. The system's impact is measured by tracking the percentage of AI-generated plans accepted without modification and the change in client compliance rates for targeted preventive services over time.

A production integration for IDEXX Neo preventive care campaigns is built on a human-in-the-loop architecture. The AI acts as a copilot, analyzing patient records—including past visits, lab results, medications, and lifestyle data from the Neo Patient and MedicalRecord objects—to generate a draft risk assessment and service recommendation. This draft is presented within the Neo interface as a suggestion for the veterinarian or practice manager to review, modify, and approve before any client communication is triggered. All AI-generated content and recommendations are logged to a dedicated AIAuditLog object, creating a clear lineage from suggestion to action.

Rollout follows a phased, risk-managed path. Phase 1 (Pilot) targets a single, high-value service like dental prophylaxis or senior wellness panels for a subset of low-risk patients. The AI's draft recommendations are reviewed 100% of the time, and feedback is collected directly in Neo to fine-tune the model. Phase 2 (Expansion) broadens to additional preventive services and patient cohorts, introducing automated draft generation for approved workflows while maintaining mandatory review for new or complex cases. Phase 3 (Scale) enables conditional auto-approval for high-confidence, routine recommendations (e.g., annual vaccines for healthy adult pets with complete histories), freeing clinical staff to focus on nuanced cases.

Governance is enforced at the API and data layer. Access to the AI inference service is scoped via API keys and role-based permissions within Neo, ensuring only authorized users can generate or approve recommendations. Patient data sent for analysis is pseudonymized where possible, and all prompts are engineered to avoid generating novel medical advice, instead grounding suggestions in the practice's established protocols and the patient's own historical data. This controlled approach ensures the integration enhances efficiency without compromising the veterinarian's role as the final clinical decision-maker.