Clinician-supervised early intervention
ECI prepares structured observations and draft intervention plans from home video, so early-intervention clinicians can focus on the decisions only they can make.
Pre-revenue / Validation planned
No plan released before approval.
01/ The constraint
The research question
Can ECI reduce clinician time per family without compromising review quality or appropriate escalation?
Early-intervention systems cannot add qualified clinicians on demand. ECI is designed to move preparation away from scarce licensed time, while leaving interpretation and accountability with the clinician.
02/ One case, end to end
Prototype screens, sequenced for illustrationSynthetic demonstration. No patient data.
Step 1 of 8 / Home evidence
The family adds their concern and priority alongside the clip: “Single words often, longer phrases less consistent,” and “How do we support communication without stopping Spanish?”
The video stays the source of truth. Every later observation links back to a moment in it.
Advances the demonstration only. It never grants approval.
Step 2 of 8 / AI preparation
Seven items drafted, zero decided. Each is linked to its evidence and none can reach the family.
Advances the demonstration only. It never grants approval.
Step 3 of 8 / Proposed plan
Proposals, not guidance. Every item waits for a clinician decision.
Advances the demonstration only. It never grants approval.
Step 4 of 8 / Clinician review
They watch the clip, read each drafted item and choose: accept, edit, reject or escalate.
Use the tabs above to follow each outcome. Here the clinician overrides one item.
Advances the demonstration only. It never grants approval.
Step 5 of 8 / Clinician correction
The override is recorded against the item as structured supervision data.
Advances the demonstration only. It never grants approval.
Step 6 of 8 / Approval
All seven items are decided. The clinician signs, and the plan names them as the accountable authority for every item in it.
Advances the demonstration only. It never grants approval.
Step 7 of 8 / Family receives guidance
The signed plan lists each accepted item and marks the overridden one. Nothing drafted by AI appears without a clinician decision beside it.
Advances the demonstration only. It never grants approval.
Step 8 of 8 / Follow-up evidence
When a follow-up clip arrives the case comes back to the clinician marked Follow-up received. Draft items are decided again, and a rejected item records its reason.
Advances the demonstration only. It never grants approval.
Step 1 of 8 / Home evidence
The family adds their concern and priority alongside the clip: “Single words often, longer phrases less consistent,” and “How do we support communication without stopping Spanish?”
The video stays the source of truth. Every later observation links back to a moment in it.
Advances the demonstration only. It never grants approval.
Step 2 of 8 / AI preparation
Seven items drafted, zero decided. Each is linked to its evidence and none can reach the family.
Advances the demonstration only. It never grants approval.
Step 3 of 8 / Proposed plan
Proposals, not guidance. Every item waits for a clinician decision.
Advances the demonstration only. It never grants approval.
Step 4 of 8 / Clinician review
They watch the clip, read each drafted item and choose. Here the clinician sees a concern that needs more than a home-video review.
Advances the demonstration only. It never grants approval.
Step 5 of 5 / Escalation
Escalation is the clinician's decision. It is recorded with a reason, for example “Diagnostic uncertainty, specialist required.”
No plan is released while the case is escalated.
Advances the demonstration only. It never grants approval.
Step 1 of 4 / Home evidence
The system reports no child identified in the video, and assessment is limited to insufficient data only.
Advances the demonstration only. It never grants approval.
Step 2 of 4 / AI response
It makes no observation about the child and proposes no plan. The system can refuse to proceed.
Advances the demonstration only. It never grants approval.
Step 3 of 4 / Clinician review
They can accept the flag, edit it, reject it or escalate the case. Nothing moves toward a plan.
Advances the demonstration only. It never grants approval.
Step 4 of 4 / Request another video
The request states what to capture. In the prototype: “Nothing is drafted until it arrives.”
No plan exists yet, so none is shown.
Advances the demonstration only. It never grants approval.
03/ First deployment
04/ Next proof
The question is not whether AI can draft. The question is whether ECI can safely increase clinical throughput while keeping accountability with the clinician.
Planned feasibility validation. No results claimed.
What will be measured
The goal is not AI efficiency.It is more children reached early enough for intervention to matter.
05/ What compounds
Structured clinician edits, overrides, escalation decisions and insufficient-evidence decisions can create proprietary supervision data tied to real developmental planning.
Could become an asset for
Potential, not established. Defensibility depends on structured capture, appropriate data rights and demonstrated product improvement.
06/ Physician-trained. Parent. Builder.
Dr. Chioma Agha-Ayorinde
“I understand both sides of this system: the clinical constraints, and the reality of trying to carry a child's plan into everyday life.”
Discuss the clinical workflow and the next validation.