Genetics, previous medical history, medication and other exposures influence which diseases we develop later in life and how fast they progress. The researchers use healthcare data from registries and electronic patient records, socioeconomic data, and genetics to feed machine learning algorithms that can predict new events. Using the Danish National Genome Center’s supercomputer, the researchers group patients, identify patterns, and define various significant trajectories that can contribute to predicting and possibly prevent future diseases of individual patients.
A focus area is patients with diabetes, where the researchers investigate how certain mental disorders and cancer co-occur in connection with diabetes. The approach addresses in a population-wide manner, complex patients with many multimorbidities, which, due to the general aging of the population, now is a significant societal problem.
At the Danish National Genome Center we use GPU resources to construct predictors of new cardiac events, time to diagnosis, or prediction of complication trajectories. We make predictions using Danish National Genome Center's security infrastructure that can be implemented in electronic patient record systems, for example Sundhedsplatformen in Eastern Denmark. We have a specific focus on polypharmacy where patients are receiving many drugs simultaneously, an area where Danish National Genome Center’s genomic data , along with other health data, over time can be used in pharmacogenomic screenings that can be fed back to the clinic and gude medication selection and tweaking of dosages, says Principal Investigator, research director, professor Søren Brunak.