Impact AI collectively appeals to all concerned actors to protect an AI in terms of ethical and responsible health.
We are the sum of reflection and action impact AI. We bring together actors who work with artificial intelligence: big companies, digital companies, start-up consulting companies, schools. We are committed to developing an ethical and responsible artificial intelligence.
The health crisis highlights the contribution and potential of artificial intelligence in health. Predictive models and the widespread use of data in this sector create new opportunities and also raises numerous medical, legal, political, economic, philosophical, ethical and technical questions.
This question led us to publish in 2021: “AI in Health: In Causes and Feelings”, an application was created with the help of about fifteen AI and / or health experts. Since then, we’ve set up a specific working group to take a closer look at the two issues identified in this document: gender-related algorithmic bias and gray data.
Algorithmic bias can affect medical choices and strategies
AIs are developed and used in healthcare to predict, diagnose, treat or monitor patients. Multiple biases in health affect the perception of symptoms by humans themselves, but also the interpretation of clinical symptoms by health professionals. If AIs are not inherently biased (sexist, racist, etc.), they use data that may be biased. AIs use this data to express potential bias. This is why special attention and specific action should be proposed to limit them as much as possible in order to promote fair health AI for all.
The gray data controller remains in the blind spot
Today, connected objects, voice assistants, platforms and applications explode the available data, especially for everything related to health. These data are very useful for the development of increasingly personalized drugs that allow everyone to be an actor in their own health. A landscape is emerging where data from medical records, behavioral data and wellness data coexist. Personal data is already regulated by GDPR and the structure that specifies health data is defined very precisely by CNIL, however, the information and data that may result from their use may be uncertain and obscure to individuals. This gray area, where data from multiple sources is used and subject to secondary use for treatment purposes, should give rise to recommendations for better control of uses based on this information.
The French Presidency of the European Union: An Opportunity to Express a Common Voice?
France is positioning itself as a leading and essential European player in the field of AI and health data, especially with the development of a health platform for health data hub, health data storage, sharing and use.
The European Commission has released its draft regulations aimed at creating a European Health Data Space (EHDS). The Commission aims to offer all citizens of the Union access to their personal health data by 2030 (the objective is formulated in the first “Digital Decade Path” program). The main one is put forward for the purpose of secondary reuse of data. Substantially accelerate the production of health services by integrating artificial intelligence.
The European Commission will also be responsible for regulating health software publishers (electronic health records – EHR systems), regulating data interoperability, promoting e-prescriptions and telemedicine through personal data or those within the scope of GDPR (consent) or an agreement) but also ” Estimated “data. The Commission can control all producers of health data in this context (medical data, data from health intended software or wellness applications), all users and hosts).
(French) Health Data Hub (HDH) has applied to be a European platform and has met the requirements of the Commission, so HDH could soon become a reference platform in Europe.
Also, the various regulations created at the European level will have a direct and indirect effect on algorithmic bias and gray data as well as healthcare AI players.
This acquired experience therefore needs to act as a springboard to endure all these issues and to hear a common voice in the complex and sensitive points of AI in the field of health. Also, with this forum, we bring together private and public actors, researchers, caregivers, patient and care associations, hospitals, training structures, lawyers, regulators, civil society actors and all those who are concerned to join our working group to make recommendations. Invite Algorithmic bias and the use of health-related data to address so-called gray area problems. To be effective, data control must be understood by all actors. In fact, in order to facilitate and promote its application in software and other health solutions, implementation methods need to be perfectly clear. We need to be careful not to add new complexities to an already highly regulated market, otherwise we run the risk of stifling innovation.
If the current French Presidency of the European Union forms a unique opportunity, outside of this time frame, we feel it is essential to build long-term institutions for dialogue and work, because only if it is together will it be possible to think of an ethic and AI for health. Specific exercises, at the center of which, is a key concept, that an AI is useful for human health, collectively and individually. The impending entry of the AI Act gives everyone the opportunity to prepare for the impending entry of new human guarantee consent for algorithmic processing. This, done five years by Impact AI with EthikIA, is part of the positive control logic to support innovation and control its ethical issues.
Impact AI with La Villa Numeris and Ethik IA