Use cases

Any clinical pathway where a result determines whether action is required.

The engine is disease-agnostic. What changes between pathways is the protocol — the markers, the thresholds, the cadence, the actions. That is content, and content is authored, not engineered.

Initial use case — AML

The most demanding first validation environment.

Acute myeloid leukaemia brings serial molecular monitoring across multiple assays, laboratories, units and reporting conventions — exactly where clinical latency does the most harm. A rising NPM1 MRD is a signal that a relapse is beginning, and it can sit unread in an inbox looking like everything else.

We started here deliberately. If the platform holds in AML, it extends. If we had started somewhere easy, we would have proved nothing.

app.echo-pathways.com / patients
NPM1 MRD on a log scale with treatment phases overlaid, showing a molecular relapse trend.
NPM1 MRD · log scale · treatment phases overlaid · relapse trend detected

Serial molecular monitoring

Trend detection on a log scale, across assays that don't report the same way.

Specimen comparability

Is this result actually comparable to the last one — same assay, same specimen, same scale?

Missing-result detection

The repeat that was due 12 weeks ago and never came.

Expansion

Where it goes next.

Validating now

AML — the flagship

The complex, high-value first use case: serial molecular surveillance, proven end to end, in the environment that punishes delay hardest.

Next

Other blood cancers

CML, ALL, MDS, myeloma, lymphoma, post-transplant surveillance and molecular relapse monitoring — the same shape of problem, different markers.

Next

Standard pathology-driven care

Neutropenia, chemotherapy toxicity, renal and liver monitoring, repeat bloods and overdue follow-up. Higher volume, lower complexity, same engine.

The test for a new pathway is simple.

Does a result determine whether an action is required? Can that judgement be written as a rule a clinician will sign their name to? If yes, it's a protocol. If the judgement can't be written down, it isn't ours to automate — and we won't pretend otherwise.

Have a pathway where results go unactioned?

That's the conversation we want to have.