Researchers at Universidad Carlos III de Madrid (UC3M) and the Severo Ochoa University Hospital in Leganes, part of the Madrid City Public Health Network, have developed a methodology that uses artificial intelligence (AI) to analyze the electrical activity of the brain during sleep, facilitating early diagnosis of Alzheimer’s disease. The study, recently published in the scientific journal GeroScience, demonstrated that by analyzing nighttime brain waves using machine learning techniques, early neurological changes can be non-invasively identified and patients can be classified into three distinct biological subgroups.
Alzheimer’s disease is a progressive neurodegenerative disease whose clinical symptoms usually appear 10–20 years after the onset of pathological processes in the brain. Currently available drugs are only effective if given early in the course of the disease.
Plasma testing for p-tau217 is starting to be integrated into clinical practice in some Spanish hospitals, but its availability is not yet uniform across the Salud National Hospital. Clinical evaluation of patients often requires supplementing this information with advanced techniques such as positron emission tomography (PET) and invasive procedures such as lumbar puncture to analyze the cerebrospinal fluid surrounding the brain and spinal cord. These tests are usually performed at a later stage, when symptoms are already evident.
Research focused on sleep
The study focused on sleep during periods of high activity when the brain performs cellular removal of metabolic waste products, including beta-amyloid protein. This process is bidirectional: Alzheimer’s disease alters sleep architecture, and as a result, sleep disturbances favor mechanisms that accelerate disease progression.
By recording electrical activity during sleep, we can learn about the biological processes taking place that can lead to forgetfulness and the characteristic symptoms of Alzheimer’s disease over many years. ”
Arrate Muñoz-Barrutia, Principal Investigator, Full Professor, Department of Neuroscience and Biomedical Sciences, UC3M
“In our research, we are looking for tools that allow early diagnosis of Alzheimer’s disease, tools that are non-invasive, affordable and applicable to the majority of people,” explains Dr. Anna Michela Gaeta, a pulmonologist at Severo Ochoa University Hospital.
Applying artificial intelligence to sleep recording
The researchers analyzed a unique database that combines recordings of nighttime electrical activity, known as polysomnography, with protein expression data from patients’ cerebrospinal fluid. This database was created in a previous study led by Dr. Gerardo Piñol Ripol, a neurologist at the University Hospital of Santa María de Lleida de Lleida, and Dr. Ferran Barbe, head of the respiratory department at the University Hospital Arnau de Vilanova de Lleida and director of the Respiratory Medicine Translational Research Group at IRBL Leida. The study included 42 patients with mild to moderate Alzheimer’s disease and 58 cognitively healthy controls. Thanks to funding from the Instituto de Salud Carlos III, Lleida’s team collected all the data and built a repository that was then analyzed by researchers in this new study.
“Signals were recorded using a series of electrodes placed on the scalp, capturing the electrical activity of neurons throughout the night. We used AI to analyze this electrical activity and identify specific changes that may later detect accumulations of proteins in the brain that lead to neurodegenerative diseases,” explains one of the study’s authors, Lorena Gallego Viñaras from UC3M’s Department of Neuroscience and Biomedical Sciences. She further clarified that “AI is not intended to replace medical testing, but rather to support early diagnosis. This allows information to be analyzed in more detail, facilitating early diagnosis and, as a result, starting treatment sooner.”
This AI-based technology was able to differentiate between healthy people and Alzheimer’s disease patients with high accuracy. Furthermore, by cross-referencing these data with key cerebrospinal fluid biomarkers such as beta-amyloid (Aβ42), phosphorylated tau (p-tau181), total tau (t-tau), and neurofilament light chain (NfL), the algorithm identified three distinct subgroups or subclusters of Alzheimer’s disease patients. These profiles showed gradual differences in biomarker levels and demonstrated that the disease exhibits distinct biological features from the earliest stages.
Results with potential for future clinical applications
In the future, these tools based on night-time sleep research may complement blood tests such as p-tau217 protein. This approach provides a cost-effective screening route that can be performed from a patient’s home and could help detect preclinical Alzheimer’s disease while simultaneously treating sleep disorders, thereby slowing the progression of cognitive decline.
“Another thing we learned from this study is that the future of science in this field depends on the integration of different specialties, such as neurology, pulmonology and engineering. This collaboration is essential, as currently drugs are being developed that act only in the early stages of the disease,” Dr. Anna Michela Gaeta concluded.
sauce:
Carlos III University of Madrid – OIDCI
Reference magazines:
Gaeta, A.M.; Others. (2026). Quantitative sleep EEG identifies CSF core biomarker-associated subgroups in Alzheimer’s disease. Gero Science. DOI: 10.1007/s11357-026-02266-z. https://link.springer.com/article/10.1007/s11357-026-02266-z
