Manuela Rehr, PhD MScPH

I am a microbiologist and epidemiologist working at the intersection of global health and AI.

I spent >16 years in international public health as epidemiologist and laboratory specialist with Médecins Sans Frontières, as a diagnostics senior technical officer with KNCV Tuberculosis Foundation, and as an independent consultant for USAID, FIND, and others.

My global health work focuses on geospatial optimization of access to health care and diagnostics, with country-level projects across Afghanistan, Bangladesh, Kenya, Mozambique, Namibia, Nigeria, the Philippines, Uganda, Vietnam, and Zimbabwe. 

A my heart is always with open-source work, I developed and open-sourced the Diagnostic Network Explorer geospatial analysis app and published a multi-country accessibility analysis for novel TB diagnostics across 11 countries in East and Southern Africa.

Recently, I trained in machine learning, neural networks, and the mathematical foundations of transformers. In my current AI research, I focus on how language models make decisions under uncertainty at the mechanistic level, i.e. how internal representations determine decision boundaries, how stored knowledge modulates evidence processing, and where these mechanisms introduce bias. I use quantitative experimental frameworks from microbiology & epidemiology with white-box mechanistic interpretability tools to make these questions measurable. This work aims to inform the use of AI in high-stakes decision-making, such as health care.

I hold an MSc in Chemistry from the University of Marburg, a PhD in Natural Sciences from the ETH Zurich and a postgrad MSc in Public Health from the London School of Hygiene and Tropical Medicine.