Build an instrument. Test its assumptions.

The question behind the system.

When a person is following an RDN-defined dietary plan, can context-sensitive support help them sustain that behavior? When is support unnecessary, burdensome, or inappropriate? Which observations would let us distinguish these possibilities?

Here, JITAI means adapting the timing or type of support using information available at a decision point. Governance means explicit limits on permissible actions and human responsibility. Auditability means being able to reconstruct what was known, which rule applied, and what happened afterward. These are design goals whose implementation must be verified.

What exists today.

Inspect the component status table and technical overview. Existing governance and recording code is a candidate foundation; the public Maya demo does not execute that code.

A bounded decision-review prototype, a separate protected professional-evaluation workflow, and public synthetic demonstrations make parts of the design inspectable. The Maya demo shows a proposed study record with context, eligibility, assignment, delivery, outcome windows, and missing reports.

The public demo is a local simulation with fixed examples. It does not execute the production decision engine, randomize participants, deliver support, or collect research observations. It cannot establish clinical benefit or validate the underlying system.

Independence, ownership and interests.

This is founder-led independent research and development, currently unaffiliated with an academic institution. Neil James founded Statera Systems Inc., the company that owns and develops KeyRD, and has a commercial interest in its future use. Independent does not mean independent of that interest.

Collaborators would agree on roles, data access, analysis, publication rights and conflict-of-interest management before a study. No institutional endorsement, accepted advisory role, grant award or study approval is implied.

Questions to develop with collaborators.

• Measurement: How should adherence to an individualized prescription be defined? How accurate and burdensome are the proposed measures? What does missingness mean?
• Decision quality: Do eligibility rules, pauses, and clinical boundaries work as intended? Can reviewers reproduce the reason for an action or abstention?
• Feasibility: Can the system reliably deliver and document the intended intervention while participants retain meaningful control?
• Intervention effects: At eligible moments, does support change the next target behavior compared with no prompt? Does any benefit persist across weeks, and for whom?

The proposed KeyRD study: a design to refine together.

KeyRD names the broader research and development project. “The KeyRD study” is a working name for a proposed investigation within it, not an approved or active trial. The protocol, team, oversight, funding and recruitment arrangements remain to be established. Maya illustrates one candidate design.

Population: a proposed starting point is adults under RDN-led care with an active, individualized dietary prescription. Clinical indication, inclusion and exclusion criteria, access needs and study size remain to be agreed with clinical and methods collaborators.

Schedule and availability: Maya illustrates one 11:30 local-time decision each weekday for six weeks: 30 planned opportunities per person. A current plan, sufficient current context and receptivity are required in the example; a pause or clinical concern makes a moment unavailable. Unavailable moments are not the randomized no-prompt group.

Candidate measure: a brief next-lunch self-report by 15:00, using criteria agreed with the RDN and participant before enrollment. A report of whether all specified components were eaten is a candidate behavioral measure, not yet a validated measure of a clinical prescription. Partial completion, food access, recall error and missing reports need separate handling. A photo-assisted record or repeated dietary recall could be considered as a comparison measure; neither is assumed to be a gold standard.

Study design: a later MRT could repeatedly assign permitted support or no prompt at eligible moments. Randomization is a study method for evaluating support; a JITAI does not inherently require randomization. The effect of interest, sample size, within-person correlation, availability and missing-data strategy must be prespecified. The demo’s invented counts are not pilot estimates.

Read the JITAI, MRT and dietary-measurement foundations.

A staged research path.

1. Define and verify. Agree on the population, dietary prescription, decision rules, measurement, and intended scientific use. Test the software and audit records using synthetic cases.
2. Establish measurement and feasibility. Design appropriate studies of measurement validity, burden, usability, delivery reliability, and safety. Set progression criteria before interpreting results.
3. Test adaptive support. Consider a prospective micro-randomized trial that repeatedly assigns a prompt or no prompt at eligible moments. Prespecify the effect of interest, sample size, analysis, and missing-data strategy. Thirty days for one fictional person are not thirty independent participants.

Any study involving people would require appropriate ethics review, consent, data governance, safety and withdrawal procedures, and an agreed protocol before recruitment or data collection. This site does not announce an approved study or enroll participants.

What would count against the idea?

Unreliable adherence measures, excessive burden, inappropriate decisions, poor delivery fidelity, or no meaningful advantage over simpler support would all matter. The objective is to learn whether KeyRD is useful for the research question, including results that do not support further development.

The existing continuity-evaluation plan addresses a separate workflow question. It is not a JITAI efficacy study, and professional judgments from it would not establish an effect on adherence.

A research-led development opportunity.

The technical challenge is to connect an individualized dietary prescription, changing context, bounded intervention decisions, and interpretable adherence measurements in one reproducible research record. The proposed contribution is that connection—not simply another reminder or a conversational interface. Its novelty and value relative to existing tools still need to be established. Existing mHealth infrastructure such as MD2K’s mCerebrum already supports sensing and intervention research. We need to compare reuse of established platforms, a simple scheduled-prompt workflow, and KeyRD’s proposed plan-bound governance and audit record. No comparative advantage has been demonstrated.

A focused feasibility project could verify rule execution and audit reconstruction, test assignment and delivery records, and assess whether a clinically meaningful adherence measure can be collected with acceptable burden. Collaborators would set measurable success thresholds and stopping criteria before starting. Passing those tests would justify further study, not establish effectiveness.

Potential applications include tools for nutrition researchers and RDN-led practices, a standalone offering, licensed components, partnerships, or interfaces with existing research and care systems. These are options to investigate, not available services or confirmed integrations. Any intended use would need suitable verification and evidence. Demand, workflow fit, willingness to pay, implementation cost, and a sustainable route to adoption remain hypotheses to investigate. Statera Systems Inc. provides the company context for this independent development work.

This agenda may inform conversations about early-stage research funding, including SBIR/STTR where an appropriate opportunity and eligibility requirements align. It is not a claim of an application, award, agency endorsement, or confirmed eligibility. A funding proposal would need specific aims, a qualified team, milestones, a budget, and an evidence-based commercialization rationale. For NIH, SBIR permits research partnerships and STTR requires a nonprofit research-institution partner; investigator employment and other requirements must be checked against the selected opportunity. NIH’s SBIR/STTR comparison provides the starting point. This site is an invitation to develop the team and feasibility case, not a funding application.

The collaboration I’m seeking.

RDNs can help define clinically meaningful behavior and boundaries. Behavioral scientists can challenge the intervention logic and burden. Methodologists and statisticians can shape measurement and causal questions. Research engineers can assess reproducibility and reliable execution.

Bring criticism, a sharper question, or a proposed study partnership. No patient records are needed for an initial discussion. KeyRD is developed through Statera Systems Inc.; potential research roles, interests, ownership, and publication arrangements would need to be agreed transparently.