Mika Gustafsson

Key Publications

Martínez-Enguita D, Dwivedi S K, Jörnsten R, Gustafsson M, NCAE: data-driven representations using a deep network-coherent DNA methylation autoencoder identify robust disease and risk factor signatures, 24 (5) bbad293 Briefings in Bioinformatics 2023.

Åkesson J, Hojjati S, Hellberg S, …, Olsson T, Ernerudh J, & Gustafsson M. Proteomics profiling reveals biomarkers for predicting diagnosis, disease activity and long-term disability outcome in multiple sclerosis. 14 (1) 6903; Nature communications. 2023.

Dwivedi, S.K., Tjärnberg, A., Tegnér, J. & Gustafsson M, Deriving disease modules from the compressed transcriptional space embedded in a deep autoencoder. Nature Communications 11 (1), 1-10, 2020.

I create machine learning models embedding large-scale knowledge utilizing functional maps with omics. These models has often been described as disease modules of co-localizing disease genes using network analysis. We have seen that such dense sub-graphs also can be identified using data driven machine learning models applied to omics. Our current interest lies in integrating knowledge with data-driven models and apply these to complex and common diseases. Of particular interest for us is to apply these methods on epigenetics and thereby create explainable AI for health status.

Group Members

Thomas Hillerton
David Martinez
Julia Åkesson
Axel Josefsson

Last updated: 2025-12-03

Content Responsible: Hampus Pehrsson Ternström(hampus.persson@scilifelab.uu.se)