Ears wide open to prevent disease

BETHLEHEM RODRIGOABC Updated: 03/20/2022 00:48h
Keep

The brothers Carla and Javier Zaldua, founders of the startup Accessible, have brought some hope to the harsh reality experienced by those who face a case of cognitive impairment, both patients and family members and medical staff. The model developed by this Spanish firm, based on 'machine learning', identifies vocal biomarkers that allow diseases to be detected early and better monitored.

Carla was working in a Swiss foundation with R&D projects and remembers reading a report from Canada on how speech analysis, in this case in English, could detect mild cognitive impairment. In 2018, together with her brother, she carried out a pilot model in Spain to collect the speech of people.

“We achieved an accuracy of over 80%. We bundled up our heads and set up the company”, explains Carla. They started with their own funds and achieved a 'neotech' fund of 250,000 euros that allowed them to make a more solid prototype. Since then, different projects have begun to emerge that make them optimistic about the future and hope to be able to develop the full potential of the project, which has already begun to receive awards. They are also part of a consortium for the health data network that will receive European funds and that will be used as an alert system to refer specialists. And it has just been chosen to enter LG's global incubator (LG Nov a).

“Two-thirds of people with cognitive impairment are undiagnosed. And of those that are done, only 18% are in the early stages," explains Carla Zaldua. She recalls that between the time memory loss begins to be detected and she goes to the doctor, "an average of 20 months pass."

Accessible is carrying out detection and screening work in the Basque public health system and also in the Quirón Salud group.

Its operation is very simple. The person performs an oral test on a computer or device with a microphone and internet connection for ten minutes. Later, the voice is analyzed with tests that study more than 100 linguistic variables and finally a doctor formulates a diagnosis on whether or not the user suffers from any pathology. Accessible's early dementia and mild cognitive impairment detection model is 93% accurate. It is being used to support the specialist, it is an alert system.

In the last edition of the Mobile World Congress, the startup presented its prototype to detect depression, which is already being used as a test by two psychology cabinets and they want to make the leap to primary care. Through questions, the system detects the state of the person regarding their mental health. In addition, together with a pharmaceutical laboratory, they are looking at the possibility of developing a prototype capable of detecting Parkinson's.

View commentsTopics



Source link