1928 Talks - Erik Kristiansson


Erik Kristiansson, professor in Biostatistics and Bioinformatics at Chalmers University of Technology, together with Kristina Lagerstedt, CEO at 1928, and Fanny Boije af Gennäs, CSO at 1928, in the second 1928 talks.

Erik Kristiansson, Professor in Biostatistics and Bioinformatics at Chalmers University of Technology, as well as a co-founder of 1928, joined us at 1928 Talks. His research aims to understand data in the space of molecular biology and medicine and his curiosity continuously drives him into new fields of research and interdisciplinary collaborations.

In this talk we discussed bioinformatics and what type of data is required to excel further in implementing AI as a support for infection control.

Bioinformatics is the computational science of biological data. The birth of high throughput methods, like next generation sequencing, has both enabled and put higher demands on the analysis of the generated data and bioinformatics have today therefore a central role within the life sciences. Bioinformatics deals with a wide range of data-types, especially data from omics technologies and, within microbiologi, it has been pivotal for providing insights into the promotion and spread of antibiotic resistant bacteria.

Erik describes an upcoming project between Chalmers, Gothenburg University and Sahlgrenska University Hospital, where they will use large volumes of data from diagnostic tests in combination with artificial intelligence (AI) to predict antibiotic resistance. The ultimate aim is to improve time to treatment and reduce the financial burden caused by the increased prevalence of antibiotic resistance. The project will be ongoing until 2023.

The conversation ventures into trying to understand what is required to apply AI in the field of infection control and antibiotic resistance. Mathematical models have been used for years to describe our reality, but with the increasing complexity of data, the models need to become more complicated. This is where AI, with its black-box approach, excels. However, to be successful, just like a sportsman, the AI model has to train and learn, which requires data. In particular, to train an AI model you need to have unbiased datasets where the answers are known, and these datasets are, in many areas within the life sciences, hard to assemble. Prof. Kristiansson says that this is the cost we pay to use AI, and that further resources need to be put into both collecting data and understanding it better.

As for the future of what we can accomplish, Erik is positive that bioinformatics research is still in its early days and moving fast forward.

Interested in participating in the context of Erik’s research on AI and antibiotic resistance? Don’t hesitate to reach out at info@1928diagnostics.com.

Watch the video here: https://www.youtube.com/watch?v=hyifimI-pTk