How we will be protected against the risks of Artificial Intelligence in healthcare (and for disability)

It is said in the industrial world that America innovates, China copies and Europe regulates. This cliché seems to be fully confirmed even when talking about artificial intelligence (AI). In September 2022 the US government through the GAO (US Government Accountability Office), the US government department that deals with accountability (from oxford dictionary; responsibility, reliability) together with the National Library of Medicine has published a study on the use and challenges of artificial intelligence in health, from which the activities of large US companies started, until President Biden's Executive Orderwhich established the first rules for the development of AI.

The European AI Act

Europe has come to regulate, with an AI ACT of more than 430 pages, very detailed, but it will not enter the legislation of the member countries for two years. It is necessary to answer many doubts from today in order to encourage the use of a tool now considered mature in many fields, primarily for health.

The draft bill

With this philosophy he arrives in Italy «Outline of the bill containing provisions and delegation to the government regarding artificial intelligence» published on 23 May 2024. Healthcare begins in Article 7: «Use of artificial intelligence in the health and disability fields». Disability is present from the beginning, we talk not only about care, not only about the National Health Service but about social and health services and the use of AI for the realization of the life project of people with disabilities.

The key principles that must be guaranteed are: protection of personal data, equity, right to information for citizens, accessibility, autonomy, security and social inclusion. AI can never replace the clinician:«Artificial intelligence systems in the healthcare sector provide support in the processes of prevention, diagnosis, treatment and therapeutic choice, without prejudice to the decision, which is always left to the medical profession». In the end, the protection of the citizen through the obligation to continuously update and verify applications

Electronic health record

One last key point, the electronic health record, where the AI
is essential. We doctors often have too much data, patients arrive with bags full of tests and tests, reports, folders in folders. The luckiest of us have young assistants who review the folders, but this is reserved for very few VIPs, both doctors and patients.

The AI ​​in the Electronic Health Record can make an evaluation of the documents: we don't care with data, but with the story we can create using the information. The data are of little use to the doctor if taken all together, they are useful if they can be integrated to meet the needs of the doctor and the patient in that specific moment, in that meeting and to plan the future diagnostic-therapeutic path. This is why AI in health is not just central, it is indispensable.

The National Platform

A solution was also needed to the problems that had arisen after the request for clarification on the tender for AI for local medicinea request from the Privacy Guarantor which had effectively forced Agenas (national agency) to suspend phase III of the creation of the Artificial Intelligence Platform to support primary healthcare and the topic is addressed in the law.

It was necessary quickly fill legislative gaps which could have slowed down the use of AI which must be rapid, so as not to lose many opportunities also for Italian SMEs (Small Medium Enterprises), but regulated. The European Union's AI ACT is comprehensive and complex, but requires a long process. You have to move quickly, guaranteeing citizens' rights and protecting companies that invest to develop projects. In health, the result seems to have been achieved.

*Vice president and head of institutional relations AiSDeT, Italian association of digital health and telemedicine