More than 4,000 people have tried FAV ZČU’s app for early dementia detection

The DigiDiaDem app, developed by FAV experts at the University of West Bohemia, can detect early symptoms of Alzheimer’s and frontotemporal dementia. So far, 15% of users have been advised to see a doctor.

All you need is a computer or tablet with a microphone and speakers. The DigiDiaDem web app, unveiled in mid-May by computer scientists from the Faculty of Applied Sciences at the University of West Bohemia (FAV) and capable of recognizing the first symptoms of cognitive disorders, has already been tested by 4,000 people in just four months of operation.

According to the researchers, results to date show that the system assessed cognitive performance as above average in 18% of those tested. The largest group, 67 % of users, achieved an average result consistent with that of the general healthy population. The remaining 15% of those tested received a result indicating reduced or deficient performance. “The app is not intended as a definitive clinical diagnosis, but rather as a reliable early-warning system. We recommend that individuals for whom our tool indicated reduced scores visit their primary care physician,” explained Luboš Šmídl from the FAV Department of Cybernetics.

The app is available online and open to everyone. It guides users through 10 short tasks focused on auditory and visual perception, memory, spoken language, and the ability to describe what is happening in a picture coherently. “For example, a person repeats a sentence, names pictures, or recalls previously presented objects. They speak their answers aloud naturally. They don’t have to type long texts or navigate a complex interface. The entire assessment usually takes about 10–12 minutes,” said Luboš Šmídl, explaining the self-assessment process.

Conventional speech recognition systems are typically developed for the continuous speech of healthy speakers. However, the speech of older adults and people with cognitive difficulties may include longer pauses, slower response times, corrections, repetitions, or unfinished statements. The project team therefore integrated several technologies, ranging from voice dialogue to specialized recognition of spoken Czech and response analysis, all the way to the automatic generation of a final report.

“The report can be printed and used as a reference during a follow-up consultation with a primary care physician, neurologist, or geriatrician. “The success of the DigiDiaDem app confirms the enormous potential of integrating artificial intelligence and natural language processing methods into routine preventive healthcare and care for the aging population,” emphasized Luboš Šmídl. He also pointed out that the doctor always has the final say.

Three institutions are collaborating to develop the DigiDiaDem app. UWB FAV experts are providing the technological solution. The Department of Neurology at the Královské Vinohrady University Hospital is responsible for examining participants and providing expert interpretation of the results. Last but not least, the Institute of Physics of the Czech Academy of Sciences is primarily involved in natural language processing, data analysis, and classification methods.