Conditional deep generative modeling of blood-based infrared spectra enables controlled in-silico phenotyping studies
Abstract Infrared molecular fingerprinting of blood offers a scalable, minimally invasive window into human physiology, but limited follow-up, imbalanced cohorts, and restricted access to diverse phenotypes constrain systematic studies. Here, we introduce a conditional deep generative framework for synthesizing blood-b...
Niklas Leopold-Kerschbaumer, Timo Halenke, Selina Süzeroğlu et al.
· npj Digital Medicine · 0 citations