Four researchers reading results together around a table

Can repeated mental health observations help Practitioners see deterioration earlier?

We are testing whether repeated, time-stamped observations can describe deterioration, persistence and recovery more reliably than isolated assessments. The work defines where those signals hold, where they fail and what evidence is required before clinical use.

Every result keeps its questionnaire, version, language, Consent and source.

Instruments developed for one population or language do not automatically answer every question in another. Heyrafiki preserves the context researchers and Practitioners need to evaluate each result responsibly.

Population context stays visible
Language, age range, responder role and setting remain part of the evidence instead of becoming assumptions.
Changes stay comparable
The Instrument and scoring version that produced each result remain attached to it, preserving the basis for comparison.
Methods can be inspected
Public scoring methods and conformance evidence show how each number was reached before another team builds on it.

Iris Affective Dynamics

What can repeated check-ins tell a Practitioner?

Iris Affective Dynamics compares repeated observations with a Person’s own history. We are testing whether it can describe deterioration, persistence, recovery and unusual departures without assigning clinical meaning.

Read the research Try the Lab

Example check-ins over timeThe shaded area shows the Person’s usual range in this example.Example data

What the view describes01
Repeated check-ins can describe change
Average, variation, step-to-step change and persistence each answer a different question.
What research is testing02
Compare each Person with their own history
With example data, we are testing how the view holds up when check-ins are irregular or incomplete.
Limits03
A trajectory is not a Diagnosis
It cannot establish cause, clinical meaning or outcome. A Practitioner decides how to interpret it.

How we work.

01
Published Instruments, checkable scoring
Every Check uses a published Instrument, its documented scoring method and a direct link to the source literature. Repeating the same answers produces the same result, so changes over time can be compared on a stable basis.
See every Instrument and its Citation
02
Safety outranks the score
When an answer indicates immediate concern, Crisis Support appears before the result. Safety is a direct path to human help, never an interpretation left to a score.
The Crisis Centre
03
Every health claim carries its source
Guidance we publish cites the research behind it, study by study. Where evidence is thin or contested, we say so rather than rounding up.
Read the Articles
04
Public studies keep their evidence together
For each public study, the protocol, sample, baseline, limitations and publication decision belong beside the result.

Three questions worth answering first.

Each question needs a declared method, a relevant baseline and review by the Practitioners, researchers or actuaries responsible for its consequences.

01
Does it hold up for Practitioners?
Whether a drafted Note is accurate enough to approve, and whether the tools save Real hours or only move the work around.
02
Does it help the People using it?
Whether matching produces a Practitioner someone stays with, and whether Care reached People who could not reach it before.
03
Does the system preserve its safety and privacy boundaries?
Whether the privacy separation holds under audit, and whether safety routing catches what it must.

A desk with the evidence chain and claim comparison worked out on paper

Open Benchmark

Reproducing a Claim at an Actuarial Cutoff.

A Claim has more than one clock. Heyrafiki’s open assurance benchmark tests whether an insurance integration can reproduce what was known at a valuation date without pulling later facts backwards through time. Its fixtures, schemas and runner are public.

The public repository keeps the API definition, example Claims, expected results and failure cases beside the runner.

Run the Benchmark Read the Article

What a public result must include.

Heyrafiki connects published Instruments, safety protocols and executable benchmarks to the evidence they produce. Population, clinical and actuarial conclusions require a declared protocol, approved data, baseline methods, limitations and expert review. Results are published against that method, whichever way they point.

01
Protocol beside the result
02
Sample and baseline named
03
Limitations kept visible
04
Publication decision recorded

Join the Rafiki Researchers

How we work.

Published Instruments, checkable scoring

Every Check uses a published Instrument, its documented scoring method and a direct link to the source literature. Repeating the same answers produces the same result, so changes over time can be compared on a stable basis.

See every Instrument and its Citation

Safety outranks the score

When an answer indicates immediate concern, Crisis Support appears before the result. Safety is a direct path to human help, never an interpretation left to a score.

The Crisis Centre

Every health claim carries its source

Guidance we publish cites the research behind it, study by study. Where evidence is thin or contested, we say so rather than rounding up.

Read the Articles

Public studies keep their evidence together

For each public study, the protocol, sample, baseline, limitations and publication decision belong beside the result.

Three questions worth answering first.

Each question needs a declared method, a relevant baseline and review by the Practitioners, researchers or actuaries responsible for its consequences.

Does it hold up for Practitioners?

Whether a drafted Note is accurate enough to approve, and whether the tools save Real hours or only move the work around.

Does it help the People using it?

Whether matching produces a Practitioner someone stays with, and whether Care reached People who could not reach it before.

Does the system preserve its safety and privacy boundaries?

Whether the privacy separation holds under audit, and whether safety routing catches what it must.

Reproducing a Claim at an Actuarial Cutoff.

The public repository keeps the API definition, example Claims, expected results and failure cases beside the runner.

Can repeated mental health observations help Practitioners see deterioration earlier?

Every result keeps its questionnaire, version, language, Consent and source.

What can repeated check-ins tell a Practitioner?

Repeated check-ins can describe change

Compare each Person with their own history

A trajectory is not a Diagnosis

How we work.

Published Instruments, checkable scoring

Safety outranks the score

Every health claim carries its source

Public studies keep their evidence together

Three questions worth answering first.

Does it hold up for Practitioners?

Does it help the People using it?

Does the system preserve its safety and privacy boundaries?

Reproducing a Claim at an Actuarial Cutoff.

What a public result must include.

Can repeated mental health observations help Practitioners see deterioration earlier?

Every result keeps its questionnaire, version, language, Consent and source.

What can repeated check-ins tell a Practitioner?

Repeated check-ins can describe change

Compare each Person with their own history

A trajectory is not a Diagnosis

How we work.

Published Instruments, checkable scoring

Safety outranks the score

Every health claim carries its source

Public studies keep their evidence together

Three questions worth answering first.

Does it hold up for Practitioners?

Does it help the People using it?

Does the system preserve its safety and privacy boundaries?

Reproducing a Claim at an Actuarial Cutoff.

What a public result must include.