how about to make OIE as agent of AI ?

[linxd]

Value of Using Mirth Connect (OIE) as an AI Agent‌
Mirth Connect (now ‌Open Integration Engine, OIE‌) serves as a powerful ‌AI agent‌ in healthcare IT by bridging fragmented systems and enabling intelligent automation. Its core value lies in:

Real-Time Data Orchestration‌: Seamlessly ingests, transforms, and routes clinical data (HL7, FHIR, DICOM) to AI models for analysis.
Closed-Loop Automation‌: Translates AI outputs (e.g., risk predictions, diagnostic insights) into actionable workflows (e.g., alerts, reports).
Interoperability‌: Acts as a universal adapter between legacy systems (EHRs, devices) and modern AI/ML platforms.
Scalability‌: Supports high-throughput data pipelines and hybrid deployments (edge/cloud).

Implementation Path‌

Data Integration Layer‌:

Configure Mirth channels to ingest data from EHRs, devices, or APIs.
Normalize data into AI-friendly formats (e.g., JSON) using built-in transformers or custom JavaScript.

AI Model Integration‌:

Connect to AI services (e.g., TensorFlow, Azure ML) via REST APIs or message queues (Kafka, MQTT).
Apply rules to trigger model inference (e.g., run sepsis prediction when ICU data meets criteria).

Actionable Outputs‌:

Route AI results to downstream systems (e.g., alerts to nurse stations, reports to PACS).
Enable feedback loops by logging clinician responses to refine models.

Optimization‌:

Monitor performance via Mirth dashboards; adjust routing logic or model endpoints dynamically.
Ensure compliance with privacy regulations (e.g., HIPAA) via encryption and audit trails.

Example‌:
Mirth preprocesses ECG data, sends it to an arrhythmia detection model, and pushes critical alerts to cardiologists’ mobiles—all within seconds.

This integration turns OIE into a ‌real-time AI co-pilot‌, enhancing clinical decision-making while preserving existing IT investments.

[nobeckett]

Sounds like a very good idea to me. This would not be a heavy lift for Mirth.
The AI/ML crowd is certainly open-source friendly and this project needs all the support it can get. This might bring Mirth into a entirely new light.

There are some additional use cases in-line with what you've already mentioned, like using Mirth as a DICOM router for sending images to post processing systems, then routing the results of that to downstream systems. Restful API calls are easy to use and will work fine in most situations, but sometimes connecting directly to the database is a better way to go. There are others .. I like this idea alot.

Sign me up!