Artificial intelligence is no longer only a laboratory topic. An expert opinion paper by Italian specialists, published in ESMO Open, brings order to the debate around breast cancer and separates two very different ways of applying it: tools that talk directly to the patient and tools that support the professional in daily work. The main conclusion is organisational rather than technological: on their own, these solutions are not enough; they need to be accompanied by change-management and continuous process-improvement methods.
What was known
Across the health system, artificial intelligence is presented as a tool able to improve quality of care, ease the burden on professionals and make care more efficient, while also improving the patient experience. An open question remained, however: how to bring it into clinical practice so that it is broadly accessible and does not remain a set of isolated experiences. In breast cancer, the literature describes possibilities, but a framework grouping them by function and by recipient was missing.
What this work adds
This is an opinion article by a multidisciplinary team of Italian opinion leaders, not a patient study or a systematic review with its own data. Its value lies in the map it proposes: two broad areas of application. The first gathers solutions designed for direct interaction with the patient, such as conversational agents that provide information and organisational support along the care journey. The second groups clinical-support tools and algorithms for professionals, aimed at reducing workload and optimising operational flows. The authors also note that, when implemented alongside lean and change-management strategies, these solutions can streamline care pathways and improve the satisfaction and interaction of patients and professionals.
Main results
As an opinion document, it provides no efficacy figures, sample sizes or measured clinical outcomes. It does set out three core statements: that the two areas of application follow different logics and require different evaluation; that benefit depends largely on how they are implemented, not only on the technical quality of the algorithm; and that ethical, regulatory, methodological and educational challenges must be resolved before these tools can be widely adopted in clinical practice.
What it means
For a patient, this paper clarifies what to expect and what not to expect. A conversational agent can guide her, remind her of steps or answer administrative questions along her treatment journey; it does not replace clinical judgement or make decisions about her disease. For the professional, support tools can free time from repetitive tasks, but their real usefulness is measured across the whole workflow, not in an isolated demonstration. The authors insist that staff training and a clear regulatory framework are conditions, not optional extras.
A view from the breast unit
The message carries a useful idea for any unit: adopting artificial intelligence is above all a process-redesign project. Before choosing a tool, it is worth defining which step of the pathway is to be relieved —patient information, appointment management, consultation preparation— and who is accountable for what the tool says. The authors’ experience suggests that without that prior work on organisation and training, the technology adds little and may breed distrust.
Reference: Toss A, Bellardita L, Botticelli A, Caputo R, Criscitiello C, De Angelis C, et al. Advancing breast cancer clinical management through technological innovation: the potential role of artificial intelligence and integrated care models. ESMO Open, 2026. https://doi.org/10.1016/j.esmoop.2026.108351
