With biointelligence we mean a multidisciplinary field of science: biointelligence is where machine learning meets biology and medicine.
In the last decades, biology transformed into a data rich science. The giant amount of biomedical data originating from various sources such as DNA sequencing, computer tomography (CT), magnetic resonance imaging (MRI), microarrays, magnetoencephalography (MEG), electroencephalography (EEG), electrocardiography (ECG), textual descriptions, etc., may often be used directly to address biomedical challenges. In most cases, however, advanced analytic techniques are required, including approaches based on machine learning. This is challenging because the straight forward application of machine learning is likely to lead to suboptimal results. Instead, adaptation of the methods and, in many cases, development of new approaches is required. This is a complex process involving the integration of biomedical knowledge into the analytic algorithms.
For example, regarding the identification of the origin of DNA replication, the straight forward application of the motif detection algorithm fails because it finds too many motifs, the vast majority out of them being false positives in the sort of sense that they do not correspond to the actual origin of replication. However, for many organism, the origin of replication can be successfully identified if we take into account the biochemical mechanisms underlying the replication process which lead to increased frequency of mutations at some particular, strand-specific locations. Finally, this allows to filter out the false positives identified by the original motif discovery algorithm.
The BioIntelligence Lab is an informal group of researchers, PhD students and volunteers with various backgrounds ranging from medicine to computer science. The group was established in 2014 and is coordinated by our institute.
Our mission is to develop machine learning techniques for biomedical tasks. In particular, we focus on
- classification and regression techniques for biomedical data, such medical time series (EEG, ECG) and gene expression data,
- link prediction in (biological) networks, and
- analysis of next generation sequencing (NGS) data and its clinical applications.