AI's Role in Healthcare Under Scrutiny

A recent article in Nature Medicine questions whether artificial intelligence (AI) truly improves healthcare, acknowledging that while some AI tools have demonstrated benefits, many applications are too recent or insufficiently evaluated to confirm their impact on patient health. The article emphasizes the need for more rigorous testing and evaluation of AI systems in medical settings.

Diverse Applications of AI in Health

The article notes that AI in health is not a homogeneous entity, encompassing diverse applications such as administrative tools, medical image analysis, and decision-making support. It compares questioning AI's contribution to asking whether lasers improve surgery, emphasizing that effectiveness depends on proper use and evidence.

Validated AI Tools Improve Patient Care

Studies have shown that validated AI tools can improve patient care. A pragmatic randomized cluster trial involving 60,893 hospital stays found that an early warning system based on nursing observations reduced the risk of death and hospital length of stay. Other systems have accelerated stroke management, reducing the time between patient arrival and intervention by an average of 86.7 minutes and improving reperfusion rates.

Regulatory Framework for AI in France and Europe

In France, Article L. 4001-3 of the Public Health Code requires healthcare professionals using AI-based medical devices to inform patients and ensure the algorithm's functioning is explainable. In Europe, medical AI software may fall under both the medical devices regulation and the AI Act, with high-risk systems subject to specific obligations. A transitional period until August 2028 is planned for systems integrated into regulated products.

AI in Cardiology and Public Health

In cardiology, AI enhances the detection of intermittent symptoms by analyzing data from portable ECG devices. This is particularly relevant in France, where cardiovascular, neurological, and vascular diseases caused 1.2 million hospitalizations and 140,000 deaths in 2022. The law No. 2026-668 of July 27, 2026, aims to develop a national strategy for combating these diseases, emphasizing prevention and care pathways.

Global Efforts to Integrate AI in Healthcare

The European Society of Cardiology highlights AI's potential for ECG analysis and rhythm disorder detection, advocating for continued research. The search for reliable clinical integration of AI tools has mobilized the sector, exemplified by the "Redefining Healthcare in the Age of AI" conference in Paris, which brought together over 600 participants, including 28 international experts and two Nobel Prize winners.

Call for Improved AI Tools and Transparency

Healthtech actors are urged to improve tools, document performance, and recognize limitations, with the primary goal of enhancing patient health and supporting professionals. The article concludes that the key question is not whether to trust AI generally, but to determine, tool by tool and use by use, what works for patients.