Digital speech biomarkers in motor neuron disease

Motor neuron disease (MND) is an incurable, rapidly progressive, and fatal neurodegenerative disease. It affects motor neurons, causing prominent difficulties with movements, walking, speech and breathing. Most affected people unfortunately do not survival beyond 1-2-years, and only a single drug has been shown to marginally benefit survival by 2-3 months.

Progress to deliver new and effective treatments have been hampered by difficulties in timely diagnosis, identification of subtypes, and monitoring of disease severity. Current standard investigations and tests are often invasive, time-consuming, subjective, and crude. There is clear urgent and unmet need for more sensitive and high frequency data that reflects individual impact of disease to facilitate diagnosis and monitoring which are accessible and acceptable to people living with MND.

Speech is profoundly affected in MND and offers unique potential as a non-invasive biomarker. Advances in digital technologies and artificial intelligence (AI) to study speech are opening a new era of opportunities in disease diagnosis and monitoring. This project aims to develop digital speech biomarkers, derived using accessible digital devices to identify patterns in speech which are sensitive to changes seen in MND.

We developed a user-friendly and widely accessible App for smartphones, tablets and computers in a co-production process with participants, specialised in the collection of speech data from people living with neurodegenerative conditions. Speech recordings are analysed using state-of-the-art AI approaches to derive digital biomarkers, which will be evaluated to assess their performance in supporting diagnosis and monitoring of MND

The major benefit of this approach is the ease and non-invasiveness of speech recordings which can be collected remotely and at scale, offering an opportunity for earlier diagnosis and closer monitoring in MND. Applications include screening individuals suspected of MND, and providing sensitive measures of disease severity for clinical care and drug trials.

This would reduce patient burden and provide additional valuable information to healthcare professionals, such as alerting clinicians to significant events like breathing or swallowing difficulties to prompt earlier interventions. This will ultimately benefit carers, relatives, and healthcare professionals by enabling better understanding of MND progression and facilitate appropriate care planning.

Digital speech biomarkers have a unique potential to significantly improve the care and research for people living with MND. This translational AI research project will develop these tools for use in MND clinical and research practice.