The Eloquence Advanced Communication Solution (ACS) was a re-think of the hospital nurse-call system. Patients press a button on the bedside device and the platform tracks the request through to fulfillment, with status updates back to the patient ("your call has been sent," "help is on the way") that the original beep-and-light systems did not provide.
Requests routed by need: pain calls went to the assigned nurse, hospitality calls were batched to a tech. Patients could ask for specific items directly ("water," "blanket," "pain") instead of paging the whole floor. The system also integrated with bedside language interpretation for non-English-speaking patients, which on its own justified the device for several of our hospital customers.
I worked on Eloquence from 2013 to 2018, first as Lead Developer, then as Project Manager from 2016. I built or co-built most of the system: the Java backend services for call routing, escalation, and reporting; the Android tablet and phone apps; the bedside hardware integration; and a chunk of the JavaFX desktop admin tools. The office was on the top floor of an old foundry building in Ann Arbor with original ironworking machinery still bolted to one wall.
The technical work I'm most proud of
The call-routing engine. Patient requests flow through a rule chain (bed assignment, nurse availability, request priority, escalation timer) and arrive at the correct device within seconds. The protocol is UDP, custom-framed, with our own packet headers. We built on UDP because the hospital networks we deployed onto weren't always TCP-friendly, and urgent calls needed deterministic latency. The protocol was validated by FDA and UL as sufficient for a certified nurse-call system.
An N+1 nested-subquery fix in MySQL. Reporting was running a separate query per row. Refactored the schema and rewrote the queries; latency dropped about 90%.
The AWS migration. When I joined, the platform was on-prem only. By 2017 we had a parallel AWS backend for demos and non-clinical workflows. Clinical nurse-call flow stayed on-prem because of regulatory constraints we couldn't cleanly resolve in the cloud at the time.
FDA Class 2 certification. Design controls, traceable requirements, documented patch and post-market reporting. New to me when I started, second nature by the time I left. I worked with the FDA directly on the certification pipeline. The product was a durable medical device with a defined patch cadence, which is a very different software-delivery rhythm from "merge on green."
UL bedside-hardware testing. Two trips to Chicago to coordinate with Underwriters Laboratories on the nurse-call tablet certification. The hardware had to survive worst-case conditions: fall, fluid spill, EMI, redundant failover, plus the standard 5-pound steel ball dropped from 8 feet onto the center of the screen. I learned more about durable- electronics design in two weeks of UL testing than in the previous five years of writing Java.
The honest framing
ACS never reached broad deployment. The largest installation was a single-floor pilot: 60 beds, 2 nurse stations, 10-15 active nurses. The studies and customer satisfaction were strong. The hospital systems were reluctant to swap out their existing nurse-call infrastructure, and the sales cycle for a durable medical device into a hospital is, on a good day, two years.
The company eventually pivoted toward VidaTalk, an Android-app-only product for nonverbal patients with a shorter sales cycle and a simpler regulatory path. That ended up being the commercial success.
I'm still in touch with most of the Eloquence team. We go climbing or biking. We built something we were proud of.