Crowdsourcing: Fostering Progress Through Partnership

The Prostate Cancer DREAM Challenge attracted a substantial number of participants to address specialized questions in prostate cancer research. The best performing teams developed very credible methodologies that could one day have implications for moving beyond overall survival as an endpoint for prostate cancer studies.
Oliver Sartor, MD, Medical Director, Tulane Cancer Center

Open competitions – “crowdsourced” challenges – represent an innovative and effective method to engage diverse communities to apply the power of data mining to advance health care. Utilizing our open-access database of clinical trial data, Project Data Sphere is uniquely positioned to aggregate clinical datasets and conduct crowdsourced challenges that can deliver practice-changing insights.

In 2015 Project Data Sphere hosted the Prostate Cancer DREAM Challenge, the first crowdsourced prostate cancer data-mining competition. More than 500 participants accessed de-identified, patient-level prostate cancer data hosted within the Project Data Sphere cancer research platform to investigate two important challenges posed by world-class urologic oncologists:

  1. Predict overall survival for patients with Metastatic Castrate-Resistant Prostate Cancer
  2. Predict docetaxel discontinuation due to adverse events

The Prostate challenge led to the development of the world’s largest curated prostate cancer data set and has yielded 25+ publications in peer-reviewed journals. The main results – including a new predictive model for patient survival and identification of aspartate aminotransferase as an important, under-reported prognostic biomarker – were published in an article in The Lancet .  The practice-changing insights that resulted from the challenge served to validate data mining as an effective way to realize new insights about disease from patterns in patients' clinical data.

Following the success of the Prostate Cancer DREAM Challenge, Project Data Sphere is initiating multiple new crowdsourced challenges:

  • In collaboration with the U.S. Food and Drug Administration and as part of PDS' Immune-Related Adverse Events research program, we are developing a crowdsourced challenge to derive insight from the FDA's Adverse Event Reporting System (FAERS) data.  Contact our program manager to become involved. 
  • In collaboration with key opinion leaders and clinician experts, PDS is developing a challenge around the treatment of metastatic breast cancer. Contact our program manager to become involved.