Deep speech analysis for cognitive assessment in clinical trials

DeepSpa will deliver a telecommunication-based neurocognitive assessment tool for early screening and monitoring of patients in clinical trials addressing Alzheimer’s and other neurodegenerative diseases. Combining AI and existing speech technologies, this novel service will allow for remote cognitive and psychological testing, thereby saving time and money.

Screenings to find participants for clinical trials that investigate Alzheimer’s disease have a 90% failure rate. A pharmaceutical company has to screen 10 people to successfully enrol one participant. Monitoring can span multiple visits (around 15 hours), and long travel times, costing around €8 000 for each potential participant. DeepSpa will provide this procedure with a more-efficient, less-expensive and patient-friendly telecommunication-based service that uses AI (artificial intelligence).


The project is led by INRIA a French Core Partner of EIT Health specialised in research, digital sciences, scientific excellence and technology transfer. INRIA is joined by other EIT Heath partners: INRIA is joined by other EIT Heath partners: Uni Maastricht, a faculty of health, medicine and life sciences from the Netherlands, and Janssen-Cilag, a Belgium pharmaceutical laboratory.

The project

DeepSpA’s mission is to deliver a technical solution for speech-based remote assessment of the cognitive status of potential candidates for clinical trials. The system uses a telecommunication service that will be accessed by clinical trial candidates via phone and/or teleconferencing, enabling performance of a battery of neuro-cognitive tests.

Speech data acquired during test implementation will be processed by back-end AI-based software that can extract the clinically-relevant metrics. This will make it possible to discriminate with up to 90% accuracy between people with Alzheimer’s disease, those with mild-cognitive impairment and those belonging to a healthy control group. The resulting data will be used to pre-select the cohort of candidates for further tests.

Along with being able to assess patients with Alzheimer’s disease, the solution is also effective for monitoring neuropsychiatric symptoms like depression. Moreover, the project explores the use of a video-conference system to perform more complete cognitive and psychological assessments (including visual-based tests) supplemented by automatic speech, as well as image analysis of facial expressions to allow remote monitoring of patients in rural areas.

The goal of this EIT Health project is to deliver a demonstrator of the tool that is ready to be marketed to external partners, such as big pharma. EIT Health support will also allow adaptation of the product, to cater to multiple European languages.


In the short-term, DeepSpA will have the greatest impact on segments of the pharmaceutical industry that are running trials of new pharmaceuticals targeting neurocognitive disorders – especially neurodegenerative diseases such as AD and related psychiatric syndromes. In the long term, the tool could benefit people with neurogenerative diseases by making clinical trials more easily accessible and cheaper, thereby facilitating the discovery of effective solutions.

Why this is an EIT Health project

This project is in keeping with the core EIT Health goal of improving healthcare. It is also in keeping with the EIT Health Focus Area of “Care Pathways” because it offers a method to improve clinical trials, with the goal of supporting development of treatments.

External partners

Dr. Alexandra König
| Psychologist, clinical researcher | Inria