Let’s start from a fixed point. Having an open and curious mind is essential to establish a fruitful and “friendly” collaboration with AI, which should not be seen only as a technical tool but as a cognitive partner that “rewards” those who are creative and flexible more than technical specialists.
This was recently reiterated by researchers from Mnesys, the largest Italian and European neuroscience research programme. The investigations also reveal that the most useful forms of human intelligence for taking maximum advantage of interaction with AI are creative, critical and conversational; furthermore, it is being discovered that some personality traits also facilitate human-AI synergy and that the most advantaged are not technical specialists, but people with an open, curious and elastic mind, such as philosophers or researchers. These aspects must obviously be exploited if we move on to professional refresher courses, or what is called continuous education, for ECM healthcare workers.
AI, Prompts and teaching
How much can AI help in this area? Science says that with generative AI it can increase creative performance. But there is also the risk of reducing the diversity of ideas: it is therefore necessary to develop one’s personal creative intelligence, to ensure that AI can be a sort of ‘muse’ that amplifies possible ideas. The human imagines, interprets, creates and can use AI to have more possibilities to choose from: the other form of intelligence that is fundamental in collaboration is in fact the critical one, which must be cultivated together with skills on the use of AI.
Studies on human-AI interaction and prompting have also observed that AI works best when the user is able to formulate good questions, knows how to explore hypotheses and is capable of iterative reasoning, and is therefore equipped with good conversational intelligence. This silent revolution is already rewriting medical history as it happens. And it poses two inevitable questions: are healthcare professionals trained to work in this new ecosystem? And is the ECM system equipped to train them?
At the moment, according to what emerged in a focus on the topic organized by Meduspace as part of the AI Healthcare Summit at AI Week in Milan, there is still a long way to go on this front. And it is essential to define a strategic roadmap to implement technological innovation, taking into account the economic sustainability of the NHS and the urgency of updating the training models of healthcare professionals.
How AI helps the doctor
In the field of prevention, efficient algorithms are now able to detect early signs of disease even before symptoms arise, predicting, for example, Alzheimer’s, COPD and kidney disease years in advance, activating prevention strategies that avert damage and acute events. The same predictive approach applied on a global scale opens up new scenarios in the monitoring and prevention of epidemics and pandemics.
Not only that. In the diagnostic field we are increasingly precise. Software that analyzes brain scans in stroke patients achieves double the accuracy of the clinical average, even identifying the exact moment of the event – crucial information for choosing treatment in critical time windows. In oncology, AI models reduce false positives and negatives in breast cancer diagnosis, reducing radiologists’ workload by up to 80%.
Not because the doctor doesn’t know how to do his job: because AI processes volumes of data that no human being could process in the same unit of time.
Finally, considering that 4.5 billion people in the world do not have access to essential health services, AI can at least partially fill this gap, bringing advanced diagnostic capabilities even where specialists cannot reach. In triage, AI systems can reduce the readmission rate by 30%, easing the pressure on facilities and professionals. All of this, not to mention the revolution that AI is bringing to pharmacological research and decision support, in order to reduce the risk of possible errors.
How will training change?
Hundreds of thousands of healthcare professionals subject to the ECM obligation operate in Italy. A training system of this scale inevitably has long adaptation times: each evolution of the contents takes months, often years, to spread widely into clinical practice.
Meanwhile, artificial intelligence models applied to medicine evolve with update cycles on average between 6 and 12 months. A traditional ECM path, between planning, accreditation and delivery, can take 12-18 months. This means that the system risks training professionals on technologies, tools and paradigms which, at the very moment in which they are transferred on a large scale, are already evolving towards a new operational configuration.
For the future, by making the most of AI, the continuous training system for healthcare professionals will have to overcome its structural limits: we are talking about knowing how to read the output of a diagnostic algorithm, recognize a hallucination of a linguistic model, critically evaluate a recommendation generated by AI. These are not niche skills reserved for hospital “technologists”: they are transversal skills that every clinician, every pharmacist, every healthcare manager will have to master.
