Finding new ideas from population data
How do the patterns in population health data illuminate new ideas? Can we use such information. sometimes amassed over decades, to sprout new ideas and test hypotheses?
Our population health team mines a wealth of existing psychiatric, medical, behavioural, socio-demographic, biomarker, and microbiome data to uncover sometimes unexpected links between illness and biomarkers, behaviours, or environmental factors.
We use combine machine learning algorithms and traditional statistical techniques adapted to epidemiology, and recently used this hybrid approach to successfully identify biomarkers that most strongly predict depression in a large, US-based population study. Agents unearthed using these methods have become the basis of clinical trials.
The datasets we use include:
- The Geelong Osteoporosis Study: one of the largest and most comprehensive psychiatric epidemiological cohort studies in the world, running for over 20 years
- The ASPREE Longitudinal Study of Older Persons (ALSOP)
- Scandinavian databases like the Nord-Trøndelag Health (HUNT) Study (CI Williams)
- The Norwegian Mother & Child Cohort Study (MoBa) – some of the largest health surveys ever performed
- Danish and Norwegian population registry studies
