Predictive modeling for disease risks

Genetic disorders often have complicated genetic causes. We provide disease risk prediction models based on comprehensive analysis of genetic, epigenetic and clinical data.

Key features of our prediction models:

• Risks of many diseases are greatly affected by environmental factors. However, many environmental factors function via producing somatic mutations. Thus, genetic features are maximized while environmental factors are minimized in our models.


• The "common disease-common variant" belief cannot be applied to all genetic disorders. Different mutations may cause the same diseases. Our modeling takes disease heterogeneity and rare variants into account.


• The causal relationships may not be straightforward. Our deep experience tells that some complex diseases may result from accumulation of small-effect variants rather than the existence of particular causal variants.


• Causal mutations may have different origins. Some are inherited from parents, some are germline mutations, while some are somatic mutations resulting from life styles/environmental factors.


• We use sophisticated machine learning or AI models to select the most relevant features based on which we build the optimal prediction model.

Below is a quick example:

Deep Learning (TensorFlow) for precision medicine modeling