27 August 2020

The research division of Philip Morris International (PMI) announces the winners of a computational challenge, the Metagenomics Diagnosis for Inflammatory Bowel Disease Challenge (MEDIC), launched through sbv IMPROVER, PMI’s international crowdsourcing platform.  

sbv IMPROVER is a collaborative initiative led and funded by Philip Morris International that was launched in 2012. Based on the principles of crowdsourcing and collaborative competition, and designed to facilitate the transparency of the research process, the project is designed as a series of open scientific challenges where computational methods and conclusions related to scientific problems are rigorously scrutinized. As the project continues, its focus will be expanded to new aspects of biomedical research. 

The competition, which came with a prize pool of $12,000, sought the best machine learning algorithm that can reliably classify Inflammatory Bowel Disease (IBD) using data obtained from non-invasive clinical samples.  

More specifically, MEDIC aimed to verify that shotgun metagenomics sequencing data is sufficiently informative to allow for accurate classification of human subjects as:

(i) IBD vs. non-IBD,  

(ii) Ulcerative Colitis (UC) vs. non-IBD,  

(iii) Crohn’s Disease (CD) vs. non-IBD, or  

(iv) UC vs. CD. 

 

sbv IMPROVER

MEDIC challenged participants to use machine learning to classify patients with IBD and non-IBD subjects based on metagenomics data

The Challenge was split into two sub-challenges: in the first sub-challenge (“MEDIC RAW”), participants were provided with shotgun metagenomics sequencing reads, so that they could process metagenomics data with the analysis pipeline of their choice to address the Challenge. In the second sub-challenge (“MEDIC PROCESSED”), participants were provided with pre-calculated taxonomic and pathway abundances matrices derived from the raw data. This allowed data scientists with no access to metagenomics analysis pipelines to solve the Challenge, as well as to compare the performance of classification methods beyond the role of pre-processing steps. The participants could participate in either one or both sub-challenges. 

As well as winning a cash prize, successful participants have the opportunity to collaborate with leading biologists and data scientists, and to publish their findings in peer-reviewed journals. 

The challenged received 81 submissions by 15 teams – with 50% of all registered teams having sent at least one submission for the MEDIC challenge, a percentage well above the average 20-30% seen in most crowdsourcing competitions. 

We congratulate the winners of the first subchallenge: 

  • Team CTLAB@ITMO, ITMO University, Russian Federation: Artem Ivanov, Vladimir Ulyantsev. 
  • Team GG-GUMC, Georgetown University Medical Center, USA: Garrett Graham. 
  • Team CDS-Lab, HPC and Networking Institute, CNRS, Italy: Mario Rosario Guarracino, Maurizio Giordano, Ichcha Manipur, Ilaria Granata, Lucia Maddalena, Marina Piccirillo. 

We congratulate the winners of the second subchallenge: 

  • Team CTLAB@ITMO, ITMO University, Russian Federation: Artem Ivanov (TL), Vladimir Ulyantsev. 
  • Team Mignon, University of Luxembourg: Enrico Glaab. 
  • Team GiGi, University of Padova, Italy: Barbara Di Camillo, Marco Cappellato, Mehdi Poursheikhali Asghari, Sebastian Daberdaku, Giacomo Baruzzo, Filippo Pietrobon, Ilaria Patuzzi. 

 

To learn more about the challenge, please visit https://www.intervals.science/resources/sbv-improver/medic

To learn more about sbv IMPROVER, please visit https://www.intervals.science/resources/sbv-improver