Reverse-engineering dialysis-machine protocols into a real-time monitoring system

Dialysis centers in Taiwan often operate under a harsh and under-digitized reality. Different machine brands behave differently, the data is rarely integrated, and nurses may be responsible for four to eight beds while manually recording critical values every fifteen minutes.

The real problem was not the abstract statement that medical data matters. It was that there was no workable system to extract machine data reliably and turn it into something operational.

I chose to start with immersion instead of architecture diagrams. I spent three months inside dialysis centers to understand the nursing rhythm, machine differences, and operational pressure before defining the product. The core move was reverse-engineering the dialysis-machine protocol and building an IoT bridge that could sit between the device and the monitoring layer.

Once that data layer existed, high-frequency machine signals could be moved into a dashboard and used for monitoring, alerts, and semi-automated operations.

The hard part was that there was no clean documentation and no standard interface to rely on. I had to understand multiple machine brands and decode the data-transfer logic directly. Once the feed was stable, the system could ingest a large volume of data per second, which made real-time monitoring a viable direction.

At the product level, we turned the raw stream into a dashboard that reduced dependence on manual note-taking and fragmented human judgment. That moved the workflow from pure manual operation toward a system-assisted one.

QuantumSmile mattered not only because the technical barrier was high, but because it converted a real clinical pain point into an operational system:

• Data that used to be manually recorded on a schedule moved into a continuous monitoring flow.
• It proved that I could work inside a device-heavy environment and still complete reverse engineering and data integration.
• It attracted serious investment and partnership interest around the product direction.

This is a classic deep-tech founder chapter: on-site understanding, technical breakthrough, and product judgment all had to happen together.

This project made one lesson very clear to me: trust in medical environments is not built with slides. It is built by showing up on-site and solving the hardest data problem in the room. Without real immersion, even good technology stays abstract.

It also reinforced that a deep-tech breakthrough is only the opening move. To become a durable business, the partnership structure, commercial boundaries, and organizational alignment matter just as much as the technical win itself.