The ADDF’s Diagnostics Accelerator Unveils SpeechDx Dataset and New Partnership to Catalyze Breakthroughs in Speech-Based Biomarkers

The Alzheimer’s Drug Discovery Foundation’s (ADDF) Diagnostics Accelerator (DxA) today announced two milestones for its SpeechDx program: the release of its first tranche of data and a new partnership with Callyope, a Paris-based AI healthtech company, to license the resource. Together, these advances mark a major step forward in the DxA’s mission to develop scalable, accessible tools that can detect Alzheimer’s at its earliest and most treatable stages – paving the way for precision medicine.

“By harnessing AI, we will be able to predict who will develop Alzheimer’s and when, unlocking the potential for personalized preventive care on a scale once unimaginable,” said Howard Fillit, MD, Co-Founder and Chief Science Officer of the ADDF. “Just as cholesterol tests transformed heart disease prevention, speech and digital biomarkers will revolutionize how we detect, track, and ultimately stop Alzheimer’s in its tracks.”

Launched in 2023, SpeechDx is a first-of-its-kind initiative to build the world’s largest harmonized longitudinal dataset of speech and biomarker data in Alzheimer’s and pre-Alzheimer’s patients. Subtle changes in speech are increasingly recognized as some of the earliest indicators of cognitive decline, even predicting it before it begins, and the newly released data provides a powerful foundation for innovators to develop and validate next-generation diagnostic tools.

“The release of SpeechDx’s first dataset and the addition of Callyope as a partner demonstrate the momentum building around this platform,” said Niranjan Bose, Managing Director, Health and Life Sciences at Gates Ventures. “By combining rich, longitudinal speech data with advanced AI, we’re opening up new possibilities for earlier detection, progression monitoring, and ultimately more effective intervention in Alzheimer’s disease.”

The DxA created SpeechDx to fill a critical gap in this emerging field, introducing a unified, harmonized speech resource that will include three years of speech, clinical, and biomarker data in up to 2,000 participants. Because speech can be captured quickly and inexpensively using microphones built into everyday devices like smartphones and computers, SpeechDx lays the groundwork for far more accessible and scalable screening and risk prediction for Alzheimer’s and related conditions.

Callyope will leverage the SpeechDx data to further strengthen their AI speech-foundation model – capable of analyzing just 30 seconds of speech to detect mood, cognitive, psychotic, and motor symptoms across mental health and neurodegenerative disorders including Alzheimer’s and Parkinson’s. It becomes the second partner to license the dataset, joining Siemens Healthineers and underscoring the growing ecosystem dedicated to advancing speech-based diagnostics.

“The SpeechDx dataset is a gold-standard resource that will allow us to train and validate our AI models on high-quality, ethically sourced longitudinal data,” said Martin Denais, Co-Founder and CEO of Callyope. “At Callyope, our vision is to make brain health monitoring objective, accessible, and transdiagnostic – and to deliver our technologies directly to clinicians through our platform, Callyope Copilot. This collaboration with the ADDF complements our focus on mental health disorders, as Alzheimer’s disease shares many overlapping symptoms with those conditions, such as depression, cognitive decline, and insomnia. Together, we will accelerate the development of tools that can make a real difference in patients’ lives.”