Research with sources
Natural-language questions on protocols, literature and regulations. Every answer cites the document it comes from.
AgenticHealth
Data stays local. AI for clinical research and quality.
Source-cited research on protocols, literature and regulations, research on pseudonymised cohorts, and draft study and quality documents, with LLMs run at the provider. It is not medical software: it does not support diagnosis or patient-care decisions.
Who it's for
- Hospitals & health authorities
- Private clinics
- Biomedical research centres
- GP networks and group practices
AI for clinical research
AgenticHealth brings agentic AI to healthcare providers and research centres, with open-weight LLMs run locally. The agents search the provider’s documents, prepare drafts and redact personal data, always with cited sources and a person’s approval. AgenticHealth is for clinical research, quality and documentation. It is not medical software: it does not support diagnosis or patient-care decisions.
It builds on experience with Meyer Children’s Hospital and Italy’s CNR. It is one of noze’s four products built on the open-source Admina framework, with IntelliPA, DataGovern and VerdePilot, and follows the OISG paradigm.
Use cases
Research on pseudonymised cohorts
Counts and criteria on pseudonymised research cohorts, in support of approved studies. Real cohorts are enabled after the provider's DPIA; until then, work is done on synthetic data.
Study and quality drafts
Synopses, minutes and procedures in draft, checked and approved by a person.
Redaction for sharing
Personal data redacted in documents to be shared with research partners.
Features
On-premise AI for clinical research and quality
Research with sources
Open-weight LLMs validated by noze ®, run locally. Every answer cites its source.
The provider’s knowledge base
The provider’s protocols, literature and procedures, indexed read-only and with the source’s permissions.
Pseudonymised cohorts
Read-only FHIR R4 data, pseudonymised before the knowledge base. Answers give counts, never lists of patients.
Drafts checked by the Reviewer
Draft synopses, minutes and procedures. The Reviewer checks the text and returns the corrected document.
Audit trail
Every AI interaction and every access to clinical data is recorded.
Compliance in one place
An overview and registers for the GDPR, NIS2 and the AI Act in the same installation, with the Transparency page.
How it works01 02 03 04
Local installation
AgenticHealth is installed on-premise at the provider and the LLMs run locally.
Knowledge base
Protocols, literature, regulations and procedures come in read-only. For approved studies, pseudonymised cohorts too.
Questions with sources
The Assistant answers in natural language and every answer links to the documents it comes from.
A person approves
The agents prepare drafts and redactions. A person checks and approves every proposal.
Pricing
Licences for healthcare facilities and research centres
Licences scaled to facility and volumes, for healthcare organisations and research projects. Quote on request.
Request a demoRelated solutions
