Section: New Results

Early detection of cognitive disorders such as dementia on the basis of speech analysis ELEMENT

Participants : Alexandra Koenig, Antitza Dantcheva, François Brémond.

This year we have contributed to the ELEMENT activity (see 9.3.2). The goal of this activity has been directed towards automated screening for cognitive decline in non-clinical settings, resulting in faster, earlier diagnosis and intervention. Inria in collaboration with the Association Innovation Alzheimer (AIA )created the French speech corpus allowing automatic detection of dementia on the basis of speech analysis. The employed speech analysis has been augmented by the means of facial expression recognition.

Due to the rapidly ageing population, the number of people with dementia in the EU will triple by 2050. The proposed speech-based screening app supports early detection and intervention, which in turn significantly reduces cost of care and preserves quality of life. The people best placed to spot early cognitive decline are carers, social workers, and family. But there is a clear lack of affordable, usable screening apps that people without medical training can use to validate these concerns and to provide actionable data for medical professionals. The approach can also be used to track mental health and other neurological conditions. The proposed solution supplements neuropsychological assessment with sophisticated and unobtrusive natural biomarkers extracted from speech data that is collected outside of medical consultations. It provides rich information about cognitive and emotional characteristics and can be used to inform clinical judgment during consultations, saving time and money. The project will bring to the European market a new product for fully-automated, reliable, unobtrusive, self-managed screening for cognitive decline, in particular dementia, and other cognitive disorders. It will allow earlier detection and, through that, more effective interventions resulting in the reduction of overall costs associated with treatment and rehabilitation. For users is will offer the comfort of flexible usage without visiting professional physicians.

The target customer group can be characterized as individuals 60+ living either at home or in residential care facilities, as well as their families, caregivers, charities, social services, other stakeholders involved in supporting older persons. The Activity will enable the first-of-that-kind product allowing implementing sophisticated and unobtrusive neuropsychological assessment within minutes right at home or at easy reachable locations without the support of professional clinicians. The initial target markets are France and UK, with the goal to start focused marketing there at the end of 2017 - beginning 2018. The focus of initial marketing will be threefold: (1) residential care facilities as access points for groups of users, (2) social services providing homecare, and (3) pharmacies with the modern trend of turning them from sales points to service providers. Therefore, the main method of revenue generation will be corporate subscriptions purchased for specific number of users. In the middle term (2-3 years), building on the footprint at the market for corporate clients, the company will start sales for individual clients. In this case the focus of marketing strategy will be on general practitioners as major recommenders and market agents.

Inria in collaboration with AIA created the French speech corpus allowing automatic detection of dementia on the basis of speech analysis. Until now, the corpus contains samples of 149 recorded participants from which 40 Healthy controls, 40 Major cognitive disorder and 57 minor cognitive disorder, for the rest the diagnosis in missing). Data collection is ongoing and will be coupled next year systematically with the video recording.

The following shows the list of transcribed audio files :

Semantic verbal fluency : 74/149

Phonetic fluency: 36/149

Pictures: 167/ 289

Counting backwards: 7/149

Sentence repeating : 168/ 596

Postive story: 77/140

Negative story: 76/149

Motivation: 75/149

In this year, target use case scenario and user requirements were defined. ki elements UG (haftungsbeschraenkt) was established on March 15th, 2017, with the purpose of commercializing the technologies matured and integrated within the ELEMENT project. Significant efforts have been invested in the preparation of the first public demonstration of the product's prototype Delta. In collaboration with DFKI, several research papers were published on the audio data collected in Nice (see [33], [32], [25]).