AI and Disease Diagnosis: from Imagination... to Reality

Artificial intelligence (AI) and diagnostic medicine are topics linked by the concept of ‘prediction’. Since good prediction enables accurate diagnosis, AI should serve to implement mathematical ‘predictive models’ that can identify a disease early. At present, AI has been applied in several areas of diagnostic medicine with positive results (e.g., in the field of lung or breast cancer identification and to support the diagnosis of Covid-19). If not adequately controlled and regulated, however, the use of AI in the field of medicine, and diagnostics in particular, can be a source of risks: for example, the use of AI systems without rigorous scientific validation, the lack of control over the data processed by expert systems, possible violations of user privacy, discrimination induced by algorithm programming, and illusory and misleading expectations for health care professionals and patients resulting from the misuse of technologies. In recent years, the field of AI has been the subject of a series of interventions by the European legislature, aimed at outlining a common reference regulation through a progressive alignment of existing disciplines in member states. In the context of a varied and evolving scenario, the research aims to analyze the possible uses of AI systems in the field of diagnostic medicine, focusing on potential risks, European legislation inspired by a precautionary logic and its impact on the Italian legal system.