Imperial College London

Christophe Stevens

Faculty of MedicineSchool of Public Health

Software Engineer/Data Manager
 
 
 
//

Contact

 

+44 (0)20 7594 8596christophe.stevens Website

 
 
//

Location

 

309Reynolds BuildingCharing Cross Campus

//

Summary

 

Projects

Imperial Centre for Cardiovascular Disease Prevention:  EAS FHSC

I am the main software developer for the EAS Familial Hypercholesterolaemia Studies Collaboration (FHSC), an international initiative aiming to generate large scale robust data on how Familial Hypercholesterolaemia (FH) is detected and managed and the clinical consequences of the current practice on outcomes.

My role consists in programming and testing the data collection webapp as well as the backend data harmonisation/integrations functions. I am interacting with national investigators in order to integrate their requirements and suggestion within the application.

Sheffield University: School of Health and Related Research (SCHARR)

As part of my Dissertation for the MSc. in Medical Statistics I was involved in the Sheffield Accelerated Value of Information(SAVI) project and more specifically in developing the Expected Value of Sample Information(EVSI) tool. 

This tool will allow health economists and scientists involved in cost-effectiveness analysis to compute the EVSI relatively easily. EVSI models the cost of extra information (i.e. a new study) per patient and for a subset of parameters. The EVSI is useful to do research prioritization and to estimate the net benefit of a trial in monetary value. 

Continuous Update Project

Database manager of the WCRF Continuous Update project (CUP) at Imperial College London for about 4 years.The CUP merges all the studies concerning Food, Nutrition, Physical Activity, and the Prevention of Cancer in a unique database. 

My main duty consisted in coding SQL queries and designing databases in order to answer epidemiologist`s demands. In complement, I programmed the Java 8 application that allows epidemiologists to analyse and manipulate the data.
I integrated R (Rserve) in the java application so that parts of the data analysis process can be done directly in the application ( high vs low forest plot).

A module developed by me let user generate large datasets(csv, excel) and large reports (rtf, word) directly from the database and via the application. 
Our project being a systematic literature review, the amount of pdf documents in our drives is huge (>13,000), I implemented a pdf library using mongoDB and solr (search engine) to reference these pdf via a java web application (jsf2.2).

It is now possible to do advanced searches on these pdf's text content. In an attempt to speed up the process of literature selection by the epidemiologist, I designed a classification algorithm (using regex-text mining and decision tree CART) that automatically exclude studies from a MedLine file. The output of this algorithm is a RIS file that can be imported in reference manager.