Detecting Alzheimer's disease from semantic verbal fluency

Alzheimer's disease (AD) is the most common form of dementia, with around 50 million people suffering from dementia
worldwide. Due to ageing population this number is expected to rise to 152 million by 2050, causing major
socioeconomic challenges of today's world. (WHO, 2017) Dementia is often difficult to diagnose as notable
manifestations appear in a later stage and screening for dementia requires seeing a clinician which is costly and timeconsuming and assumes having access to healthcare. In recent years, numerous studies have shown that language
dysfunction is the first sign of declined cognitive abilities. Natural language processing (NLP), signal processing and
machine learning techniques have provided promising results in automatically detecting dementia from language, making
it low-cost and accessible. One of the most promising approaches in early AD detection is using semantic verbal fluency
(SVF) tasks, (e.g. naming as many animals as possible in one minute). Several studies have achieved promising results
in differentiating between healthy, mild cognitive impairment (MCI) and AD population based semantic cluster sizes,
switches between clusters, word production strategies and word position in time. My research proposal is to improve the
NLP techniques used in the analysis of SVF tasks to contribute to early detection of AD. I would include languages that
have not yet been analysed (English, Swedish, Estonian, Russian), and pair SVF task with imaging.