Machine Learning and Biological Age

Biological age is the deviation from chronological age and is hypothesized to be one causal factor for age-related disease. Biological age, however, is difficult to measure. Here, we analyzed 676,787 samples from 502,211 UK Biobank participants aged 37-82 years with deep learning artificial intelligence approaches to build a total of 331 biological age predictors on different data modalities (e.g., Magnetic Resonance Imaging)

The Confluence Projects

Analytics and data resources to study the geospatial exposome in exacerbating health outcomes of the elderly population.

Exposure-wide studies

Claims Analysis of Twins Correlation and Heritability

We estimate the relative contribution of genetics and shared exposome in 560 phenotypes in a large health insurance cohort analyzing data from ~60,000 twins and ~500,000 siblings.

Exposome Globe Browser

What is the “linkage disequilibrium” of the exposome? View how biomarkers of exposures are correlated with one another and disease-related phenotypes.

HIV+ Exposure-wide study across Sub-Saharan Africa

Predisposition to become HIV positive (HIV + ) is influenced by a wide range of correlated economic, environmental, demographic, social, and behavioral factors. A data-driven approach to identify risk factors for HIV+ in across Sub-Saharan Africa in over 600,000 individuals.

Non-linear relationships between physiological indicators and all-cause mortality in the US

We document linear and non-linear relationships of 27 physiological indicators with all-cause mortality to evaluate whether the current clinical thresholds are suitable in distinguishing patients at high risk for mortality from those at low risk.

Data-driven characterization of exposome risk variables for type 2 diabetes in the Netherlands

We query for environmental and modifiable drivers of tpe 2 diabetes risk in the Lifelines Biobank Cohort.



Developing new computational approach for predicting new drug repositioning candidates? Test your predictions with our standard database for drug repositioning.


MeSHDD uses MeSH-term enrichment to discover literature-based similarities between FDA approved drugs.


Vibration of Effects

How do inferences change based on the parameters of statistical models? Estimate the distribution of association sizes, and p-values based on model selection, called the Vibration of Effects (VoE).

  1. Journal of Clinical Epidemiology 2016: Mortality
  2. PLOS Biology 2022 Microbiome
  3. PLOS Biology 2021: Data-driven for quantitative phenotypes (quantvoe)

Polyexposure Risk Scores (PXS) via PXSTools

PXStools provides an analytical package to standardize exposome-wide studies as well as derive and validate polyexposure risk scores in the UK Biobank. PXSTools

Course Materials

Course materials and starter code for exposome-phenome data analysis.