[Artificial intelligence and palliative care: opportunities and limitations.]

The so-called artificial intelligence tools applied to palliative care (machine learning, natural language processing) have great potential to support clinicians in improving decision-making processes and in identifying those who are at high risk of mortality or at greater risk of inappropriate treatment and/or non-positive outcomes. The improvement of prognostic abilities may help to avoid that some choices of patients with serious diseases are taken only in the last days of life, in the face of treatment options not previously discussed in an adequate and shared way. These tools can facilitate some essential aspects in the practice of palliative care, for example the activation of interviews that have as their objective the advance care planning and the definition of treatments consistent with the needs and desires of patients, especially in final stages of life. The development, also in our country, of projects for the application of artificial intelligence in palliative care requires particular attention to the possible organizational repercussions and to some ethical and relational aspects. It will be necessary to reflect on the most appropriate organizational models and on the specialized resources necessary in relation to the foreseeable increase in the number and variability of patients with early identified palliative care needs. These tools must not interfere in fundamental elements of the relationship between patient and doctor, that is the ability to communicate a poor prognosis in an individualized and ethically appropriate way.