Doctors and technology

Neurology: how AI is helping to provide personalised care and early diagnosis

Artificial intelligence will only have an impact on neurology if we are able to manage it, validate it and integrate it into a model in which technology and clinical responsibility go hand in hand

by Tommaso Bocci *

12 June 2026

3' min read

Translated by AI

Versione italiana

Key points

3' min read

Translated by AI

Versione italiana

Several centuries have passed since Homer, in the *Iliad*, offered a brilliant, ahead-of-its-time definition of artificial intelligence and ‘machine learning’, describing humanoid beings crafted by Hephaestus to serve the gods, capable of learning, making decisions and correcting their own mistakes.

Much has changed since then. The computational foundations of artificial intelligence have been in place since the second half of the twentieth century; what makes AI central today is the need to organise, systematise and understand a mass of data that cannot always be interpreted using traditional tools. With the development of neuroimaging, the brain has become readable through millions of pixels, bytes and signals.

The role of AI

Why is artificial intelligence so important in neurology? For some years now, particularly in relation to conditions such as Parkinson’s and Alzheimer’s, we have been witnessing a promising yet complex situation: we have more information, but we are not always able to manage it. AI can process vast amounts of data, recognise patterns, monitor patients via wearable sensors and build predictive models useful for clinical decision-making. But it also raises a question: how can we use these tools without undermining the doctor’s responsibility and the relationship with the patient?

The opportunities and challenges that formed the focus of the session “Artificial Intelligence in Neuroscience” at the 65th National Congress of the SNO (Hospital Neurological Sciences), held at the Mediterranean University of Reggio Calabria. Clinicians, researchers and ethics experts discussed the transition of AI “from the bedside to the cloud”, from everyday neurological practice to large collaborative data analysis networks.

The 'Parkinson's case'

In the case of Parkinson’s disease, for example, three innovations have revolutionised deep brain stimulation (DBS) in recent years: directional electrodes, brain sensing and adaptive brain stimulation. DBS involves electrically stimulating a small nucleus, the subthalamus, which plays a central role in the complex circuitry that generates and sustains the disease.

Electrodes are becoming increasingly sophisticated: the stimulation contacts, which are smaller and more numerous, allow for better targeting of the current, conserve battery life and reduce adverse events. However, greater precision also multiplies the possible combinations. Often, even the most accurate medical and anatomical knowledge of the neurologist and neurosurgeon is not enough to identify the best parameters for the individual patient.

Brain Sensing has added a further layer of complexity: after years of experimentation, we have succeeded in recording the ‘voice’ of the subthalamus and other key players in the disease, identifying brain frequencies correlated with symptoms. This has paved the way for devices capable of suggesting which contacts to activate or, in adaptive stimulation, of modifying certain parameters in real time according to the situation.

Has all this improved our work? Only to a certain extent. Because as the amount of information increases, so does the entropy of the system: we do not always know whether what we are recording is noise, a cause or effect of the disease, or an epiphenomenon of processes that still elude our understanding. Recent studies show that disease-specific frequencies are not present everywhere, do not appear in all patients, and the two cerebral hemispheres may respond differently.

Wearable devices

In addition to these data, there are those from wearable devices, which record the speed and amplitude of movements, and those obtained using less invasive techniques, such as surface electroencephalography. Looking ahead, sensors and brain signals could contribute to earlier diagnoses and increasingly personalised treatments, tailored to the individual patient like a bespoke suit.

However, a proper understanding of the technology remains fundamental to the effective use of artificial intelligence. Training is essential: that is why, in Milan, we were among the first in Italia to introduce a specific course on artificial intelligence for healthcare professionals. Innovations of this scale must be accessible and understood by all, not reserved for the few.

The verification of information generated by AI must always involve human oversight. Artificial intelligence may well have an impact on neurology, but only if we are able to manage it, validate it and integrate it into a care model in which technology and clinical responsibility go hand in hand.

* Neurologist (SNO), Senior Consultant at ASST Santi Paolo e Carlo, Associate Professor of Neurology at the University of Milan

Brand connect

I prossimi eventi

Tutti gli eventi

Notizie e approfondimenti sugli avvenimenti politici, economici e finanziari.