Speech reveals the first signs of neurodegenerative diseases
Neurodegenerative diseases such as Alzheimer’s or Parkinson’s disease rank among the greatest current health threats. Early detection of their initial stages is crucial, as it enables physicians to introduce treatments that can slow disease progression and improve patients’ quality of life. Responding to this challenge is a team led by Jiří Mekyska from the Brain Diseases Analysis Laboratory (BDALab) at the Faculty of Electrical Engineering and Communication of Brno University of Technology (FEEC BUT), which has developed a method that uses spontaneous speech analysis to detect these diseases at an early stage. This approach is simple, non-invasive, and could form the basis of future screening - for example, during a routine visit to a general practitioner.
Professor Jiří Mekyska, FEKT VUT. | Photo: Jakub Rozboud
Research has shown that even a short monologue, such as talking about one’s everyday routine, can contain valuable information about a person’s cognitive state. Using tools from natural language processing (NLP), signal processing, and machine learning, the team analyzes linguistic and acoustic features such as sentence complexity, the frequency of repeated phrases, or the length of pauses. These parameters change already in the early stages of disease. For instance, people with mild cognitive impairment associated with Alzheimer’s tend to use fewer function words (pronouns, conjunctions, auxiliary verbs, etc.), whereas patients with Parkinson’s disease often produce shorter sentences and longer pauses.
“Speech is a window into the brain. Just ninety seconds of spontaneous narration is enough for us to detect subtle impairments that would otherwise remain hidden. Unfortunately, Alzheimer’s and Parkinson’s cannot yet be cured, but treatments are gradually emerging that can slow their progression. To be effective, however, the disease must be detected early. Our goal is to develop an affordable and easy-to-use tool that will significantly improve early detection. Current methods used by general practitioners unfortunately fail to identify many individuals in whom neurodegeneration is already underway,” says Jiří Mekyska, who works at the Department of Telecommunications at FEEC BUT.
The results of the study show that speech analysis can not only distinguish between different types of cognitive impairment, but also correlates with structural brain changes observed on Magnetic Resonance Imaging (MRI). This means that linguistic and acoustic biomarkers can serve as early indicators of neurodegenerative processe - even before significant clinical symptoms appear. Details of the method and findings are available in a research article published in Computers in Biology and Medicine.
This research bridges electrical engineering, computer science, and medicine, demonstrating how modern technologies can help improve the quality of life of an aging population.
“This research is the result of a multidisciplinary team. I would like to thank Daniel Kováč, who led this significant study, as well as other colleagues from our BDALab, from the Central European Institute of Technology at Masaryk University, and from St. Anne’s University Hospital in Brno. The collaboration between VUT and its startup Scicake works exceptionally well, and we plan to transfer this method into clinical practice so that it can have real impact on the lives of patients and their families,” adds Jiří Mekyska.
Author: Zdeňka Koubová
| Responsible person | Ing. Zdeňka Koubová |
|---|---|
| Date of publication |