Enhancing EHR based risk scores with genotyping and metabolomics.
The aim of this project is to develop scores for risk assessment and early detection of various diseases based on multiple data sources. These scores can be used by physicians and public-health coordinators to tailor personalized treatment and improve their patients health. We wish to combine the following data sources –
- information sources that are widely available through the patients' medical records - laboratory test results, vital signs (e.g. blood pressure), medications and diagnoses made by physicians
- Genomic information characterizing the patients - this is less common nowadays, but will probably become ubiquitous in the near future
- Potentially new markers that are not commonly measured nowadays - in particular the characteristics of various small molecules circulating in the blood (metabolites). If such molecules prove important for predicting patients' health, they may be measured as part of the standard of medical care. In the longer run, we believe that checking the whole profile of such molecules (through technologies such as Nuclear Magnetic Resonance (NMR) and Mass-Spectrometry) would also become part of the standard medical care.
Once we obtain the relevant data, with a collection of individuals who either develop the conditions (cases) or not (controls), we will be able, within several months, through methods of machine learning, to construct scores that use the available data to asses individual's risks and provide early detection and warnings for multiple diseases. The conditions we wish to investigate include cancer, metabolic disorders, such as type II diabetes, kidney disease, and any other morbidity that is common enough in the UK biobank.