The DANTE project

Project DANTE

The national DANTE project – an acronym for “Standardized Early Access Data Extracted through Automated Language Processing for Evaluation” – has now been launched.

Led by the Artificial Intelligence & Cancer Consortium in partnership with AstraZeneca, Amgen, and MSD, DANTE was selected as one of the France 2030 winners in the call for expressions of interest titled “Evaluation of New Tools, New Applications, and New Methodological Approaches in Clinical Research.” It is also one of the national projects under the National AI and Health Data Strategy, which aims to improve the sharing and evaluation of the use of health data and AI systems.

The Rationale: From Promise to Practice

His rationale is simple: to move AI-driven clinical data extraction solutions from the promise stage to the implementation stage, building on a national opportunity: early access programs.

Early access programs allow patients who have reached a therapeutic dead end to benefit, on an exceptional and temporary basis, from drugs considered innovative, before they become available under standard regulations. This is an opportunity for France, for patients, and for the teams caring for these patients.

This opportunity comes with a responsibility: to document, with supporting data, what actually happens after these innovations are first made available.

Five Complementary Values

DANTE embodies a number of complementary values.

A medical value

This data is often among the first real-world data available on the use of an innovative treatment for its indicated use. It provides a better understanding of the treatment’s role in the therapeutic strategy, allows for monitoring of tolerability, helps anticipate the occurrence of adverse effects, and improves the expected benefit for patients.

A Scientific value

At present, we do not yet have a robust, comparable, and reproducible measure of the quality of data collected by AI.

DANTE will compare three databases for the same patients:

  • the database created through the usual manual data entry;
  • the database generated by automated AI extraction from intellectual property documents;
  • a baseline consisting of independent ARCs.

The objective is to measure the accuracy, errors, omissions, and consistency of the data based on a predefined set of 23 variables and 147 patients.

An Epidemiological and Regulatory Significance

The rate of reporting early access monitoring data remains insufficient. The HAS reiterates a target of 90%, whereas the current rate is approximately 65% at best. At this level, the data remain too incomplete to be useful: this limits our understanding of the adoption of innovations, the assessment of their medical impact, and their contribution to healthcare decisions. DANTE aims to create the conditions for more reliable and comprehensive data collection.

A land value

All too often, hospital staff still experience “double data entry,” the proliferation of platforms, delays, follow-ups, and administrative burdens as additional constraints on top of patient care. Yet much of this data already exists in electronic health records (EHRs). DANTE’s goal is to test whether properly integrated, tracked, and evaluated AI can extract this data from EHRs without adding to the daily workload of doctors, pharmacists, nurses, clinical technologists, clinical research coordinators, and care teams.

An economic value

The funding received by institutions for collecting early-access data is tied to the completeness of the data submitted. Improving this rate, therefore, is not just about producing better data; it is also about better recognizing the work done by the centers and contributing to the sustainability of health data warehouses beyond their launch phase.

Beyond early access programs, the issue is strategic: the implementation of these AI tools should make it possible to generate up-to-date, accurate, and reusable data that complements medical and administrative data. This “high-definition” data is essential to modern oncology.

With DANTE, that’s exactly what we’re setting out to do.

Early access programs mobilized

The project will be based on real-world early-access programs data provided by AstraZeneca, Amgen, and MSD, respectively.

The first early-access program (AstraZeneca) has been launched

It evaluates IMFINZI (durvalumab), in combination with FLOT chemotherapy as neoadjuvant and adjuvant therapy, followed by IMFINZI monotherapy as adjuvant therapy, in adult patients with resectable gastric or gastroesophageal junction adenocarcinoma.

HAS page is available here

Who can join ?

If you have access to this innovation as part of the DANTE project demonstration and would like to join the project to showcase your expertise and the maturity of your solutions, participation is open:

  • to healthcare facilities – whether they are university hospitals (CHU), regional university hospitals (CHRU), general hospitals (CH), cancer care centers (CLCC), private facilities, or other prescribing facilities—provided they have an operational AI solution;
  • to centers that work with an SME or a specialized service provider;
  • to networks – whether public or private – or to SMEs with network-deployed AI solutions, meaning those capable of operating in one or more locations where their solution is already installed, deployed, or can be activated quickly.

Learning as You Go

The procedures for onboarding partners are intentionally simple. Applications will be processed on a first-come, first-served basis: the first complete application received will be the first one processed.

This is therefore not a competition, nor is it a ranking of solutions: DANTE aims to provide a snapshot of the ecosystem’s maturity, which will be shared with regulatory authorities, in order to collectively lift all stakeholders to a higher level.

Two documents to be submitted

If the application is complete, admissible, and ready to proceed, the Consortium may enter into a contract with the candidate. To do so, in practical terms, interested parties must submit two items: a presentation package and a signed letter of intent.

1. The Partnership Proposal

It will include a proposal introducing the candidate, the relevant center(s), the AI solution being used, its level of maturity, available data sources, experience with health data and the GDPR, as well as a realistic timeline for implementation. The objective is to enable the Network to verify that the applicant is qualified to participate, that the solution is operational, and that the DANTE framework can be adhered to. The format of this proposal is flexible but must not exceed 10 pages.

2. The Letter of Intent

It must be signed by the authorized representative(s) of the applicant parties. Its purpose is to confirm that the applicant has reviewed the DANTE partnership agreement and agrees to the terms of the agreement. This letter helps ensure the prompt processing of applications as part of a rolling admissions process. The sample letter is available below.

Where to start ?

To help you prepare this application and letter of intent under the best possible conditions, we have provided a series of documents below. To make it easier to read and get straight to the key points, we recommend that you start by reviewing the partnership specification document, which outlines the general framework of DANTE, the conditions for participation, the operational requirements, the proposed contractual structure, and the business model.

If you feel up to the challenge of this project, please read the study protocol and the partnership agreement. These documents outline the terms of reference, but in scientific and legal formats, respectively. Finally, please feel free to use the appointment scheduling option provided below if you would like additional information about any of these documents.

Join us !

Documents to Review

The documents are available (in french only):

Connect with the DANTE team

To answer any questions you may have and help ensure that you submit complete applications, one-on-one appointments can be scheduled at the following link: appointment scheduling link.

Note: These meetings are intended to help applicants understand the terms of reference and prepare a complete application. They are therefore reserved for those who actually wish to apply under the proposed framework.

Requests related to consulting services, general sales offers, or matters outside the scope of DANTE cannot be processed under this framework.