Artificial intelligence lands in family doctors' offices: 'Mia' helps with diagnosis and prescriptions
Funded by the Pnrr, the platform will be tested by the first 1,500 white-collar workers: they will be able to ask for help with diagnoses, treatment pathways, drugs to be prescribed or prevention activities
It is called 'Mia' and will enter the offices of the first 1500 family doctors in the next few days. Mia is an acronym that stands for Medicine and Artificial Intelligence, and will be more than a virtual assistant, it will be a true scientific helper for white coats, able to support them in many of their daily decisions: perhaps to arrive at a diagnosis when the symptoms are unclear and they are not so certain, or to suggest a treatment pathway, an examination to be carried out or a drug to be prescribed to a patient who has been suffering from a chronic illness for years, or even to recommend useful prevention activities for patients who are over a certain age and are better off having an extra check-up or screening.
After having made its entrance into hospitals - more and more diagnostic equipment is using it to improve its performance - artificial intelligence, thanks to a project financed by the NRP, is about to enter the surgeries of family doctors, including those who work or will work in the new community centres in the area. Agenas, the National Agency for Regional Health Services, which is also the Agency for Digital Health, after having closed the tender worth EUR 8.3 million for the commissioning and management of the first artificial intelligence platform for primary care - a tender won by the company Bv tech - has, together with the regions, hired the first 1,500 family doctors who, after a training course that will start soon, will test 'Mia' in their work during 2026. For now, we are starting with the first group of doctors, but the idea is to bring 'Mia' to a much larger number of white coats within two to three years.
But how does 'Mia' work? It is a platform that can be queried from a PC - also in the form of an app - by means of a user-friendly conversational interface with a technological architecture based on artificial intelligence (Rag: Retrieval Augmented Generation). The doctor will be able to ask questions to 'Mia', which in turn will draw from a 'validated and certified' knowledge base: these are information sources and official documents such as guidelines, protocols, regulations, scientific literature, diagnostic and therapeutic care paths, nomenclators, etc. Among other things, the project envisages the activation of the certification pathway of the instrument as a medical device, a step to further guarantee reliability and adherence to sector-specific standards. The arrival of 'Mia' in family doctors' surgeries coincides with a number of international initiatives in which artificial intelligence is increasingly present in the creation of content in the health sector, as demonstrated by the arrival of the new 'ChatGpt Health' version just announced by OpenAi (see article opposite), which will soon be open to all web users.
The project to create 'Mia' comprises three phases: the first phase of analysis, design and implementation of the IT tool ended last 31 December. A second phase of experimentation and maintenance, to be completed by 31 December this year, which starts in these days, will consist of the progressive implementation by the 1,500 experimenting doctors of the three use cases of the platform, namely 'basic diagnostic assessment', 'support for chronicity management' and 'support for prevention and health promotion'. During this year, a training plan will also be implemented to enable doctors to use the tool and increase their knowledge of artificial intelligence. Finally, the last phase envisaged is the management and maintenance of the Platform, starting from January 2027, which envisages a gradual extension to the other family doctors working for the SSN: after the first 1500, it will be extended to another 3 thousand white coats in 2027, then another 7500 and finally another 15 thousand.


