The future of healthcare: the contribution of oncologist Arsela Prelaj
Artificial intelligence is no longer confined to research laboratories: it is entering the ward and redefining the way therapies are chosen. This is the powerful message that emerges from the first contribution to the training course The Future of Healthcare, promoted by Johnson & Johnson and Microsoft Italy with Fondazione Mondo Digitale ETS.
The protagonist of the first contribution is Arsela Prelaj, an oncologist at the National Cancer Institute of Milan and head of an AI laboratory in collaboration with the Politecnico di Milano, who talks about how new technologies are radically transforming clinical practice, especially in the field of oncology.
Beyond screening: AI that predicts therapy
While “first-generation” systems focused on screening and diagnosis, from imaging for lung cancer to digital pathology, the new frontier is the personalisation of treatment.
Research currently under review in Nature Medicine shows how the use of artificial intelligence systems can increase the probability of correctly identifying the best treatment by 37% in patients who are candidates for immunotherapy, compared to human clinical assessment alone. The impact is twofold:
- optimising therapeutic efficacy
- reducing unnecessary toxicity
- supporting even less specialised hospitals in complex clinical decisions.
In this scenario, AI does not replace the doctor but supports them in the most critical steps. “It is essential that these systems are explainable and reliable, because the patient asks you: how did we arrive at this decision?”.
From Digital Twin to agentic systems
Dr Prelaj's vision extends beyond the specific use of the algorithm.
- Digital Twin: a “digital twin” of the patient, continuously fed with clinical data to dynamically adapt the therapeutic strategy.
- Agentic systems: developments in generative AI that act as true “digital colleagues”, capable of integrating guidelines, clinical data and context to suggest motivated and explainable actions.
‘Your tool system is not just a predictive system, but [...] in some ways seeks to provide not a prediction, but an action, so it becomes what we call a “teammate” or, in English jargon, an AI colleague.’ In this paradigm, technology becomes part of the team, but decision-making responsibility remains human.
Democratising data to innovate research
Another crucial issue is the use of synthetic data, which allows predictive models to be validated in different contexts and supports research into rare diseases, where the number of real patients is often limited. This opens up a perspective that intertwines clinical innovation, data ethics and equitable access to research.
‘With synthetic data, data can be more easily shared and, in some ways, democratised. This can help because, on the one hand, by adding more data, you increase performance, and on the other, you try to validate your models in more contexts.’
A path to understanding and managing change
‘It is important for us all to raise awareness and truly understand what the future of healthcare will look like.’
Arsela Prelaj's contribution inaugurates the series of expert voices collected in the project The Future of Care, which addresses the relationship between artificial intelligence, clinical practice and new professional skills. The complete training course is available free of charge on demand on the FMD Academy.
