AI biology company Prima Mente has introduced a groundbreaking approach to diagnosing early-stage neurodegenerative diseases, including Alzheimer’s and Parkinson’s. The company has developed an artificial intelligence model named Pleiades, which leverages epigenetic data to enhance diagnostic accuracy and offers a less invasive alternative to traditional methods.
Neurodegenerative conditions are complex and often begin manifesting subtle changes in gene regulation long before clinical symptoms appear. Hannah Madan, PhD and co-founder of Prima Mente, emphasized in a recent interview that while imaging and cognitive assessments provide some information, they do not inform treatment pathways effectively. “This is what epigenetics can start to unravel,” she noted.
Instead of relying solely on fixed genome sequences, Prima Mente’s model analyzes changes in DNA methylation within cell-free DNA (cfDNA) extracted from patient blood samples. This innovative technique has achieved an impressive 89% success rate in detecting early-stage Alzheimer’s and Parkinson’s in real-world clinical cohorts. When combined with leading neurodegenerative biomarkers, such as pTau-217, the model’s accuracy increased to 97%.
Innovative Collaboration and Clinical Applications
Prima Mente is a member of the NVIDIA Inception program and has partnered with NVIDIA Cloud Partner, Nebius, to utilize the DGX Cloud Lepton AI platform for model training and development. Founded in 2022, the company operates from offices in London, San Francisco, and Dubai, employing over two dozen staff members.
Madan outlined the rationale behind the collaboration, stating, “The hypothesis we had from the beginning was that bigger models would help understand complex biology over time, which requires both compute and the ‘know-how’ for building these models.” Currently, Pleiades is in use within a live clinical study named SANDBOX, spanning 15 sites across the U.K. This initiative aims to classify and stratify patients suspected of having neurodegenerative diseases, ultimately improving clinical outcomes and addressing waiting lists within the U.K. National Health Service.
The Pleiades model is structured as a 7 billion parameter epigenetic foundation model, trained on a vast dataset comprising 1.9 trillion tokens of epigenetic information. This data includes human methylation and genomic sequences sourced from various cell types, sequenced using methods that preserve the cytosine methylation state. By focusing on tissue-specific information, Pleiades accurately identifies the cfDNA cell type of origin, pinpointing components relevant to Alzheimer’s, including neurons and immune cells.
Future Aspirations and Broader Applications
Looking forward, Prima Mente is expanding its research scope to include the development of therapeutics. The team is also creating a virtual cell model called Parthenon to identify novel therapeutic targets, particularly for microglia-related cells. Pleiades has revealed that cfDNA fragmentation patterns can serve as a new class of biomarkers for Alzheimer’s detection. These fragments provide insights into biological processes, including epigenetic states and mechanisms of cell death. Impressively, Pleiades reconstructs cfDNA fragments with 83% per-nucleotide accuracy and maintains a 91% match to original methylation patterns.
Moreover, the model can enhance signals from specific cell types in blood samples by up to 20-fold, facilitating the detection of rare disease signatures even from limited material. As the research progresses, Madan indicated that Prima Mente aims to extend its model beyond epigenomics to encompass proteomics, transcriptomics, and additional clinical data. Longitudinal studies involving both blood samples and brain tissue are already underway.
Madan concluded by highlighting the platform’s versatility, stating that it is designed to be generalizable for application in various disease areas by 2026. With its innovative approach and commitment to advancing neurodegenerative diagnostics, Prima Mente is poised to make significant strides in the field of precision medicine.
