Data Analytics Program Takes Leading Role in NIH Challenge Around Data Sharing

Sean Hennessey

Date

August 6, 2026

Topics

Research

MSDA faculty and teaching-team members named on team that finished among winners of national competition

The foundation of discovery isn’t the research papers that populate the academic landscape, it’s the underlying data. And the only way for the data to be effective is if it’s validated, scored, and shared. But how is that effectively handled?

Master of Science in Healthcare Data Analytics (MSDA) faculty and teaching-team members Dr. Allison Goff, Dr. Pedram Safari, and Shivani Pimparkar, MS, were named on one of the winning teams in the National Institute of Health’s S-Index Challenge, a national competition designed to find the best, most rigorous, well-engineered approach to measure scientists and their data.

The team was led by the MSDA Program Director Shuhan He, MD. Seven teams reached the final round: Yale University, Mount Sinai Hospital, Oregon Health & Science University with the University of Michigan, the California Medical Innovations Institute with the University of Washington, and a group from Harvard Medical School that included the director of the MIT Press, co-founder of bioRxiv and medRxiv, and scientific director of arXiv.

“This is a win for the Institute,” said He, who has led the Master of Science in Healthcare Data Analytics (MSDA) program for nearly four years. “Creating and validating new metrics is what this program does, and it is why our people get named to national efforts.”

Traditionally, a researcher’s impact has been measured by the H-index, which reflects the productivity and influence of their published scholarly work. The S-index recognizes the scientific impact of creating and sharing high-quality datasets that serve as foundational resources for the research community. Shared datasets can enable many future discoveries by supporting hypothesis generation and results validation.

With that in mind, He led a team of researchers and analyzed 76-million pieces of data using 25 categories that document, shepherd, and upkeep high-quality data that can be reused later. They built and validated a new framework for measuring the quality and impact of biomedical research-data sharing, helping create new ways to recognize and reward the data that make discovery possible. The goal is the kind of widely usable recognition metric for data sharing that the H-index became for publications.

The result is the website sharescore.org.

This was a national, trans-NIH effort led by the National Eye Institute with contributions from multiple institutes, centers, and offices. NIH Director Dr. Jay Bhattacharya was present at the final event.

“Every AI model is downstream of the data it learned from. Until we can tell good data from bad at scale, we cannot tell a good model from a bad one, and that is the question the whole field is standing in front of. Students who come here learn from the people answering it. If you want data analytics that sets the national standard, this is where you come.”

More than Sending an Email

He points out that data sharing is much more complex than you might think. While some might think of file sharing, that’s just the beginning.

A five-column chart highlighting the 25 signs of care — While creating the website sharescore.org, MSDA Program Director Shuhan He and his team created 25 categories to score data, a central differentiator in the team’s success during the National Institute of Health’s data challenge.

“The methodology you created, the license to use it again, the protocol, the provenance to show it actually happened,” said He. “It’s not the file. It’s everything that makes the file mean something.

“We asked the questions any researcher asks when they open someone else's data. Is there a guide? Is there analysis code? Is there a file list? Is it organized? Can I reuse it? Can I get it? Are the methods included? Is attribution clear? These are the signs of care.”

Producing sharescore.org was a massive engineering undertaking. He’s engineering team at Conduct Science spent six months and roughly 85,000 lines of code pulling records out of nine separate public repositories — Zenodo, Figshare, Dryad, OSF, Dataverse and others — each with its own way of describing what a researcher had shared. The system ranks 140,000 researchers from around the world; one can find every dataset, map, data diversity, Application Programming Interface documentation, all of which produces rankings of researchers and their S-index, or data-sharing score.

Why It Matters

“This work positions the MSDA program of the MGH Institute at the forefront of building the measurement infrastructure for a more open, rigorous, and equitable science,” noted He. “It demonstrates a program that does more than prepare learners to analyze healthcare data, machine learning, and AI. Our MSDA program is helping define how data stewardship, sharing, and real-world reuse can be recognized, rewarded, and trusted across the national research enterprise.

“Somebody who does this work carefully has had no way to show it. A paper gets counted. The dataset behind it does not. That’s the gap we were trying to close, and it is the part that will outlast the competition.”