Understanding the prospective significance of published scientific findings today is of critical interest. Estimating the future impact of a given publication is currently an imperfect process, often overly relying on the prestige of the journal in which it is published.

Utilizing machine learning to develop algorithms to predict scientific paper citation count has steeply advanced in the past few years.  The goal of this challenge is to present a software tool to predict future impact / citation counts of peer-reviewed scientific publications.

Publications on selected topics of interest will be used to judge the effectiveness of a tool based on the following criteria:

• Accuracy: Since current citation counts for publications are known, the algorithm derived score will be evaluated for accuracy by this crucial metric.    
• Speed: Quickly digesting and applying the algorithm to the thousands of publications associated with each topic is key.
• Originality and Inventiveness: Utilizing your own newly developed original algorithm will score higher than using developed algorithms.

Anyone around the world may participate in the challenge. This challenge is not open to HCPs. Amongst up to 50 submissions, the best performing algorithm(s) will be awarded with a prize of EUR 10,000.

This challenge is now closed.

Anyone from around the world may participate, as an individual or in a team. This challenge is not open to HCPs.

• Submission is by completing the Application Form and submitting by e-mail to open.innovation@emdgroup.com before 23:59 (CET) August 31st, 2020.  
• Only the first 50 working algorithm and tool entrants will be accepted into this challenge.
• Detailed instructions on how to use your developed algorithm and tool with a group of publications or a simple to use front end with instructions must be supplied.  We will work closely with entrants to run their tool, but we are not liable if it does not run during the evaluation.
• The tool must be able to process XML papers used in the PubMed Central (PMC) open paper shared files.  ftp://ftp.ncbi.nlm.nih.gov/pub/pmc/oa_bulk/
• You may not create a tool that looks for existing citation counts. The tool delivered to us must be a predictive tool and must deliver a theoretical impact / citation count and rank papers for this future impact/citation count.
• Accuracy is a crucial criterion to win this challenge.
• The tool must be able to evaluate thousands of papers at a time, not one at a time.
• Your code and code repository must be shared with Merck KGaA, Darmstadt, Germany for evaluation.
• An already developed algorithm may be used in this challenge to develop the tool, but originality and inventiveness will be considered when judging the entrants.
• You retain any rights that you hold with the developed tool and publishing rights you may have.
• Merck KGaA, Darmstadt, Germany retains the exclusive right to make any press release or any kind of public communication about the competition and the winning proposal. Confidential details on project content will not be revealed.
• Amongst up to 50 submissions Merck KGaA, Darmstadt, Germany will award the best performing algorithm with a prize of EUR 10,000.