Spot the conditions for the next zoonotic emergence.
Via an MCP tool.
The prompt
“”
AqtaBio answers it via MCP. Live JSON below.
Scroll through six scenes. Each one is a real, callable tool response. No mocks, no sleight of hand. The Disease X risk score below was computed by calling get_disease_x_risk against the production MCP server when this page loaded.
Pathogen-agnostic.
WHO Blueprint eleventh priority.
Tool call: get_disease_x_risk(tile_id=“AT_horn_5_5”) , Horn of Africa tile, live from production.
First publicly-callable Disease X surveillance signal, addressing the WHO R&D Blueprint's eleventh priority pathogen. The score aggregates risk from all eight pathogens in the model (5 with live tile predictions, 3 with model-scored signals while tile seeding completes); highlighted fields are the ones a public health officer needs first.
AUROC 0.975. 12 of 15.
Honest about caveats.
Tool call: get_hindcast(event_id=“2019_wuhan_sars_cov_2”, response_lead_time_days=30)
Validation set first: AUROC 0.975, AUCPR 0.864, 12-of-15 hit on the 2026-03-01 ebola-model backtest. Live-backtest cross-check on Hubei peaks at 0.896, ~2-month lead versus first reported spillover. The recorded v0.1.0 attestation paired with the WHO PHEIC notification (31 Dec 2019) anchors a +53-day lead versus that public notification date. The dict-encoded attestation is a recorded artefact, not a live recompute; the response's cross_check field surfaces the live-model values alongside.
FHIR conformance.
Fetch it back yourself.
Tool call: submit_to_hapi_fhir(tile_id=“AS-025-45678”, pathogen=“sea-cov”, month=“2026-04”)
FHIR conformance as a callable proof. The RiskAssessment is written to the public HAPI FHIR test server. Open the link above, fetch the resource back in another tab, and confirm the payload matches what AqtaBio computed.
From clinician prompt
to patient EHR.
SHARP provides a structured workspace context that bridges a clinician's question to a live patient record. AqtaBio plugs into that bridge: pull the patient, compute local spillover risk, write a FHIR RiskAssessment back.
Data flow
{
"tool": "get_patient_local_risk",
"note": "Real schema, see live tools at /.well-known/agent.json",
"input": {
"patient_id": "Patient/hapi-test-001",
"fhir_server": "https://hapi.fhir.org/baseR4"
},
"output": {
"patient_id": "Patient/hapi-test-001",
"tile_id": "AT_southeast_asia_12_8",
"pathogen": "sea-cov",
"risk_score": 0.74,
"risk_tier": "high",
"month": "2026-04",
"fhir_server": "https://hapi.fhir.org/baseR4"
}
}{
"tool": "emit_riskassessment_to_ehr",
"note": "Real schema, see live tools at /.well-known/agent.json",
"input": {
"patient_id": "Patient/hapi-test-001",
"fhir_server": "https://hapi.fhir.org/baseR4",
"risk_score": 0.74,
"pathogen": "sea-cov"
},
"output": {
"status": "created",
"fhir_resource_type": "RiskAssessment",
"resource_id": "RiskAssessment/aqta-ra-2026-04-hapi-test-001",
"url": "https://hapi.fhir.org/baseR4/RiskAssessment/aqta-ra-2026-04-hapi-test-001",
"written_to_ehr": true
}
}These are real schemas. See the live tools at /.well-known/agent.json.
Built to be verified.
Every claim, callable.
Open standards
Honest status
v0.1.0 research preview. Population-level risk signals only; not approved for individual-patient clinical use. The Wuhan hindcast is a recorded development-cycle attestation: a frozen model score paired with an independently verifiable WHO notification date. It is not a prospective study. The pre-medRxiv preprint is targeted for Q3 2026.
Built on MCP + A2A v1.0 + FHIR R4 + SHARP, open standards throughout.
© 2026 Aqta Technologies Limited · Dublin, Ireland