Motivation

Dementia, one of the most common neurodegenerative diseases, is on the rise due to demographic changes and an ageing population. The disease is characterized by cognitive decline and often manifests itself in the form of memory loss, aphasia, apraxia, agnosia or deficits in executive function.

While recent breakthroughs in antibody therapy offer hope for slowing the progression of Alzheimer’s disease when applied in the early stages, the early detection itself remains a major challenge. The insidious onset of dementia often delays detection, which is exacerbated by overlap with age-related memory loss and leads to underdiagnosis. However, early detection is crucial for timely intervention and reducing healthcare costs and the burden on patients and caregivers.

Secondary diagnoses such as depression add complexity to the diagnostic process. Depression increases the risk of developing dementia and further exacerbates cognitive and functional impairments. The accurate discrimination between dementia and pseudodementia (caused by depression) is challenging even for experts, as they have shared symptoms.

Dementia screening and monitoring enable early detection and tracking of disease progression, with a combination of medical examination, psychological history taking, cognitive testing and rating scales forming the gold standard. Current biomarkers, which are primarily invasive and costly, lack sensitivity in the early stages. Speech analysis is proving to be a promising, non-invasive and cost-effective tool that makes biomarkers accessible from clinical assessment or spontaneous speech.

Project Goals

This project focuses on two main goals:

It involves ongoing data collection in clinical settings such as memory clinics and care facilities, where subjects’ speech is recorded while participating in interviews and standardized neuropsychological tests. Participants’ diagnoses range from no or mild cognitive impairment to mild to severe forms of dementia and include various genesis (e.g. Alzheimer’s disease) and secondary diagnoses (e.g. depression).

Results

The results related to this project were published in the following peer-reviewed conferences:

  1. Braun, F., Bayerl, S.P., Pérez-Toro, P.A., Hönig, F., Lehfeld, H., Hillemacher, T., Nöth, E., Bocklet, T., Riedhammer, K., 2023. Classifying Dementia in the Presence of Depression: A Cross-Corpus Study, in: Proc. Annual Conference of the Int’l Speech Communication Association (INTERSPEECH).
  2. Braun, F., Förstel, M., Oppermann, B., Erzigkeit, A., Hillemacher, T., Lehfeld, H., Riedhammer, K., 2022. Automated Evaluation of Standardized Dementia Screening Tests, in: Proc. Annual Conference of the Int’l Speech Communication Association (INTERSPEECH).
  3. Braun, F., Erzigkeit, A., Lehfeld, H., Hillemacher, T., Riedhammer, K., Bayerl, S.P., 2022. Going Beyond the Cookie Theft Picture Test: Detecting Cognitive Impairments Using Acoustic Features, in: Sojka, P., Kopeček, I., Pala, K., Horák, A. (Eds.), Text, Speech, and Dialogue. Springer International Publishing.

Partners

The data is recorded at the Memory Clinic (“Gedächtnissprechstunde”) of the Department of Psychiatry and Psychotherapy, Nuremberg Hospital and the Bavarian Red Cross retirement home, Kronach. Both partners also provided meta-information such as medical diagnoses of the participants.