HugoScore hugoscore.org

Symptom assessment AI

Ada Health

Ada Health is queued as a patient-facing AI symptom assessment and enterprise patient-finding/navigation tool. The CAIHL review should examine patient visibility, possible-cause explanations, care urgency guidance, language access, medical-device boundaries, data governance, partner accountability, and whether enterprise deployments change patient control.

Draft profile Open directory

Public-source research has been drafted; final human publication review and change-log detail are still required.

66 /100 toward patient-directed
Agency posture Mixed, potentially agency-expanding
The question we ask Who does Ada Health serve in this deployment?
Control Institutional or clinician-mediated use with patient impact
Agency read May help care, but must be tested for visibility, consent, correction, and institutional priority drift.
Vendor
Ada Health GmbH
Who it serves
Hybrid patient-facing symptom assessment AI
Primary User
Patients, health systems, payers, pharmaceutical companies, and population health programs
Control Model
Public-facing app and enterprise deployments controlled by vendor and deploying organization
Patient Impact
Symptom assessment, possible-cause reporting, triage guidance, care navigation, report sharing, and patient finding
Profile Status
Draft profile
Last Reviewed
Jun 12, 2026
Review Confidence
Low draft, partial public-source check

Summary judgment · 66% toward patient-directed

Mixed, potentially agency-expanding

Patient-facing symptom assessment may support reflection and navigation, but triage, routing, regulatory, and enterprise deployment effects need close review.

Patient agency

How this tool changes agency

Expands agency when

Symptom reports may support action if uncertainty, urgency, and limits are made clear.

Limits agency when

Needs review of app use, enterprise deployment, opt-out, and alternative care routing.

Patient-facing signals

Who does this AI serve?

Hybrid patient-facing / enterprise

Ada offers a direct symptom checker and enterprise care-utilization, navigation, and patient-finding deployments, including pharmaceutical company use cases.

Can patients tell AI is involved?

Likely yes

Public pages call the symptom checker AI-powered, but deployment-specific disclosure needs review.

Can patients meaningfully choose?

Not researched

Needs review of app use, enterprise deployment, opt-out, and alternative care routing.

Can patients correct or challenge what the AI produces?

Partially documented

Users can delete cases under Ada privacy policy controls, but Ada says business partners may make automated decisions based on assessment outcomes and related GDPR Article 22 rights must be asserted against the partner. In-app correction, report export, and challenge routes still need review.

Does it help patients understand or act?

Potentially

Symptom reports may support action if uncertainty, urgency, and limits are made clear.

Text findings

Who is left out or burdened?

Partially documented

Ada publishes multiple language options and requires users to be at least 16. Accessibility, safety-net use, pediatric/elderly use, digital access, and symptom-literacy support still need review.

What happens to patient data?

Partially documented

Ada Health GmbH is the GDPR controller for the app. Health-data processing is consent-based. Users can delete individual cases, with deletion or irreversible anonymization within one month, though some data may be retained for medical-device safety, legal, or post-market surveillance duties. Ada says it does not share data for third-party commercial interests without explicit consent. US-specific policy details and model-improvement specifics still need review.

Are the clinical boundaries clear?

Partially documented

Ada's privacy policy states Ada Assess is registered as a Class IIa medical device under EU MDR, while Ada notes the app or Assess may not be regulated as a medical device everywhere it is used. Emergency/triage boundaries, clinician review, and deployment-specific regulatory claims still need review.

Who defined what good looks like?

Partially documented

Public materials cite vendor and clinical evidence, including care-utilization outcomes. A full review should verify study independence, deployment context, and whether patient-defined quality or agency measures were included.

Review method

Initial seed classification updated with a 2026-06-12 factual audit of Ada product, regulatory, and privacy pages. No hands-on testing or full CAIHL review.

Draft profile · Low draft, partial public-source check