EPIX.AI: Harnessing AI and Epigenetics to Decode and Extend Human Healthspan,

Milena Georgieva, Nikolay Vasev, Andrey Bachvarov, Martin Klosi

Martin Klosi: My career began as a software engineer in e-commerce, but a growing passion for longevity and human health drew me into bioinformatics and AI. At 23andMe, Grail and Soley, I led teams that built large-scale data and machine learning systems for genetics and epigenetics, spanning projects like Genmo-Wide Association Studies, early cancer detection, and high-throughput target-agnostic drug discovery. These experiences led me to my current work at Epix.ai, where our mission is to harness AI and epigenetics to help people understand and extend their healthspan. At Epix.ai, we are developing machine learning pipelines that turn raw biological data into meaningful insights. A flagship project is our epigenetic biological clock, a deep neural network trained on thousands of DNA methylation samples, which estimates biological age - an algorithm based on Steve Horvath’s pioneering work and further improved with deep learning approaches like DeepMAge. I will discuss how we optimized this model, addressed challenges such as inconsistent public datasets and batch effects across labs, and deployed a scalable inference system on the cloud. Finally, I will discuss our pipeline for Polygenic Risk Scoring and our exploration of Oxford Nanopore Sequencing, which offers unprecedented resolution for next-generation aging clocks. Together, these efforts move us closer to true personalized, preventative healthcare.