Author: Abel García Abejas, Faculdade de Ciências da Saúde da Universidade da Beira Interior

Artificial intelligence is increasingly entering palliative medicine through prediction, algorithms can now estimate mortality, identify clinical deterioration and recognise end-of-life trajectories long before they become apparent to clinicians, in many respects, this represents genuine progress.

Earlier identification may improve advance care planning, reduce unnecessary interventions and help patients receive palliative support sooner. In healthcare systems where referrals often happen too late, predictive technologies may even help address important gaps in care, yet prognosis is never simply a technical matter.

Medicine does not deal only with disease; it also shapes how people experience uncertainty, vulnerability and the future. This is especially important in palliative medicine, where prognostic information often carries significant emotional meaning for patients and families.

Knowing that someone may be approaching the end of life is not the same as living through that reality, artificial intelligence therefore has the potential to change the experience of illness itself, information about the future may arrive before a person is emotionally ready to receive it, a risk that remains statistical may begin to feel immediate and personal.

Patients may start viewing their lives through the lens of predicted outcomes, while families and clinicians may unconsciously adjust their expectations according to what the algorithm suggests.

This can create what might be called a burden of anticipation, anticipation itself is not the problem, palliative medicine has always involved recognising change, preparing families and discussing uncertainty with honesty and compassion.

The challenge arises when predictions begin to influence how people experience their lives today, human beings are not simply clinical trajectories, even in the face of serious illness, people continue to find meaning, sustain hope, maintain relationships and redefine what matters most to them, life is not lived as a prediction.

Yet predictive systems may subtly encourage healthcare professionals to focus more on projected futures than on the person who is living in the present moment, this is particularly important in palliative medicine.

A prediction may help organise care, but it may also influence how patients see themselves and how others see them, there is a risk that people become increasingly defined by what is expected to happen rather than by who they are and how they are living now.

The deeper concern is therefore not simply that technology might be wrong, it is that predictions may begin to shape how people understand their future and how they live the time they still have.

Artificial intelligence may help clinicians recognise deterioration more accurately, but no algorithm can determine how a person should live when facing mortality, no predictive system can fully understand reconciliation, fear, hope, love, unfinished relationships or the importance of human presence at the end of life.

Palliative medicine has long served as a reminder that healthcare is about more than diagnosis and treatment. Its foundations remain deeply human: listening, accompanying, understanding and being present.

For this reason, the integration of artificial intelligence into palliative medicine requires more than technical validation, it also requires careful reflection about how predictive information affects patients, families and clinicians.

There is an important difference between helping people prepare for what may come and making them feel that their future has already been decided.