Artificial Intelligence and Glioblastoma: What Is Really Changing for Patients
In this series of articles based on the final report of the Sixth Biennial World Summit of Brain Tumour Patient Advocates organised by the IBTA in Rome in November 2025, we turn today to a topic that is increasingly shaping every aspect of our lives: artificial intelligence. An entire panel discussion session at the Summit was devoted to it, moderated by Petra Hoogendoorn (Netherlands) with the participation of Martin Glas (Germany), James Buckley (United Kingdom), Chris Tse (New Zealand), and, representing Italy, Professor Riccardo Soffietti of CancerSucks APS in Turin. Roberto Pugliese of Glioblastoma.IT ODV was also present at the Summit — and the application of artificial intelligence to patient support is a topic our organisation has been working on concretely for some time.
Artificial intelligence is everywhere. We hear about it every day, often in a confused way — oscillating between exaggerated enthusiasm and equally exaggerated fear. But what does it mean in practice for someone facing glioblastoma or a brain tumour? What is already changing in diagnosis, in treatment, in the way patients search for information? These were the questions at the heart of the Summit session, which sought to separate what exists today from what remains a promise for the future.
Artificial intelligence, in its simplest definition, is the ability of machines to simulate typically human behaviours — abstract reasoning, problem-solving, learning from experience. It is not magic, nor is it an autonomous entity: it is a tool, powerful but dependent on the quality of the data it is trained on and the questions it is asked.
In neuro-oncology, the most concrete and already operational applications concern above all diagnostic imaging. The analysis of MRI scans is one of the areas where artificial intelligence is showing the most significant results: algorithms trained on thousands of images are able to identify tumour characteristics — size, margins, enhancement patterns — with a precision and speed that supports and complements the radiologist’s work. The aim is not to replace the doctor, but to provide a tool that reduces the risk of assessment errors and can be particularly useful in centres with less specific expertise.
Artificial intelligence is also being applied to histopathological and genomic analysis: computer vision systems can analyse sections of tumour tissue and identify molecular characteristics — the presence of certain mutations, the tumour grade — that sometimes escape the human eye or require lengthy analysis. In the longer term, these tools could significantly accelerate the integrated diagnosis we discussed in the previous article. AI is also entering the fields of biomarker identification and drug development as a data analysis tool, capable of identifying patterns in large datasets that no research team could examine manually.
But there is an aspect that concerns patients even more directly: many of us are already using artificial intelligence, often without fully realising it. Anyone who has received a glioblastoma diagnosis has almost certainly searched for information on Google, perhaps using the AI synthesis features now integrated into search engines. Many have asked questions of ChatGPT or other chatbots. This is already reality, and at the Summit it was discussed without judgment: it is understandable and natural that someone facing such a serious diagnosis should seek answers wherever they can, including from AI tools.
The issue is not the use itself, but the quality and reliability of the responses. The panel identified three main challenges. The first is the accuracy of information: language models generate plausible text, but not always accurate text, and in the medical field an imprecise piece of information can create false hopes or unjustified fears. The second is consistency: different tools, or the same tool at different times, can give similar but not identical answers to the same question, creating confusion. The third is interpretability: when an algorithm produces a prediction — about response to a treatment, about disease progression — it often cannot explain why it reached that conclusion, making it difficult to assess its reliability.
That said, there are AI applications oriented towards patients that are showing concrete value. Apps for cognitive rehabilitation, for example, can support patients with cognitive deficits — memory, attention, executive function — which are among the most frequent and disabling consequences of the disease and its treatments. Symptom monitoring apps allow patients to record daily how they feel, what symptoms they are experiencing, and how they are progressing towards their goals: this information, shared with the clinical team, can improve the quality of therapeutic decisions and make medical appointments far more effective. In both cases, AI does not replace the doctor-patient relationship, but enriches it with data that would otherwise be lost.
A concrete example close to home is Gliobot, the chatbot available on the Glioblastoma.IT ODV website. Gliobot is integrated with active clinical trial databases — both the American ClinicalTrials.gov registry and its European counterpart — and allows patients and families to search for clinical studies relevant to their situation, as well as to obtain information about specialised centres. Integrated within Glioblastoma Navigator, and given sufficient data on the patient’s profile, it is also able to estimate compatibility with specific ongoing clinical trials — a practical application of artificial intelligence in the service of those who need to navigate a complex therapeutic landscape. It does not replace the doctor, but it can be a very useful starting point for arriving at an appointment with the right questions.
One aspect worth highlighting is the potential of AI to address disparities in access to care. Not all glioblastoma patients today have the same opportunity to be followed by a multidisciplinary expert team at a specialised centre. AI tools that support diagnostics or guide therapeutic decisions could, in time, help reduce the gap between high- and low-specialisation centres — bringing advanced expertise to places where it is currently scarce.
The panel’s message was measured: artificial intelligence will change neuro-oncology, that much is certain. But the pace and manner will depend on the quality of available data, on the ability to clinically validate tools before adopting them at scale, and on the willingness to involve patients and advocacy organisations in the development of these technologies — not as passive recipients, but as active interlocutors. As the Summit discussion emphasised, linking people with technology can transform information into influence and data into progress. But this requires that technology be developed with its users in mind.
For anyone who wants to learn more or is curious about the responsible use of artificial intelligence tools in managing their illness, the advice is always the same: use them as a starting point for asking the right questions, not as a source of definitive answers. And then bring those questions to your doctor.
Source: Report of the Sixth Biennial World Summit of Brain Tumour Patient Advocates, IBTA, Rome, 2–5 November 2025. Session Ten — Panel Discussion: Artificial Intelligence and Neuro-Oncology.
