Assessing medical devices embedding artificial intelligence
As digital technology unfolds and CNEDiMTS assesses more and more connected medical devices, the question of access to reimbursement for those that will use artificial intelligence will arise. In order to streamline the examination of these dossiers - and thus allow patients rapid access to innovation - HAS publishes today a draft analysis grid of self-learning algorithms. This initiative will help identify this new and growing field. It is subject to public consultation until 15 January 2020 to gather suggestions from all stakeholders.
Artificial intelligence is shaking up the world of healthcare. For patients and health care professionals alike, the digital provides for a new way of working, treating, being treated and looking after one’s health. But a technological innovation is not necessarily a clinical innovation.
In this context, the National Committee for the Evaluation of Medical Devices and Health Technologies (CNEDiMTS) - a HAS specialist committee - has decided to supplement its evaluation tools to prepare for requests to evaluate medical devices with embedded self-learning algorithms. To establish the benefits of their reimbursement by health insurance, and as for any other medical device, CNEDiMTS must rule on the benefits that these technologies bring to the patient or public health. The criteria for assessing the benefits of innovations embedding machine learning processes therefore remain clinical endpoints, the proposed analysis grid constituting a complementary descriptive panel for the technologies concerned.
Contributing to rapid access to innovation
To conduct its evaluations, CNEDiMTS uses the data provided by industry. In order to guide them in the constitution of their dossiers, it provides them with several documents developed by the committee and available on the HAS website, including a guide for submitting dossiers.
Currently, this guide is enriched with a new draft section dedicated to medical devices with one or more self-learning algorithms. Presented in grid form, this draft encompasses 36 items covering 8 key areas, such as the algorithm learning process, the data involved in this learning process, or those that are involved in the medical device's decision. It should be noted, however, that none of the items relate to compliance with CE marking-related general requirements for safety and performance - prerequisites for evaluation by CNEDiMTS - or compliance with personal data protection requirements, which is not a CNEDiMTS’s remit.
For HAS, the use of this grid during evaluations also represents an opportunity for patients: once completed, the opinions of the committee, all of which are made public, will comprise the technical answers to the evaluated items.
A public consultation for a collective approach
For this new evaluation field, HAS considers that a collective approach is necessary: this draft grid is therefore subject to public consultation until 15 January 2020. The aim is to gather the opinions and suggestions of all stakeholders involved in the development or use of medical devices integrating learning algorithms: industrialists, patient associations, national professional colleges, but also developers of IT solutions, researchers, Interdisciplinary Artificial Intelligence Institutes, etc. HAS wishes to appreciate the legibility of this draft grid and the relevance of each of the items identified. All contributions will be analysed in order to arrive at the final version of the analysis grid that will be used in the context of future requests by industrial health insurance providers for their medical devices. This grid should be definitively adopted and implemented in April 2020
The National Committee for the Evaluation of Medical Devices and Health Technologies (CNEDiMTS) is the committee of the French National Authority for Health (Haute Autorité de Santé - HAS), which evaluates in particular medical devices (MDs) for reimbursement by Health Insurance. The evaluated MDs are those with CE marking intended for individual use. The CNEDiMTS evaluation criteria are regulatory criteria - those of Article L.165-2 of the Social Security Code - applicable regardless of the type of MD. The aim is to give an account of the benefits of the MD for the patient and for public health (expected clinical benefit): individual dimension with regard to the therapeutic, diagnostic or disability compensation effect and collective dimension, with regard in particular to the impact in terms of improving the health status of a population, mortality, morbidity, quality of life, response to a therapeutic need, diagnostic or disability compensation. Additionally, CNEDiMTS expresses itself on the role of this MD in the arsenal available in France (expected clinical added value).