Cornerstone AI and Bristol Myers Squibb Present Findings at 2025 IMPACCT RWE
Overview
Recently, our team and partners at Bristol Myers Squibb (BMS) gave an oral presentation at the IMPACCT RWE Summit in Boston MA, focused on how Cornerstone’s AI-powered platform helped extract and harmonize laboratory data from the Alliance for Genomic Discovery (AGD) dataset, as well as dynamics for how Sponsors can make “buy” vs. “build” decisions in the age of AI. We’re excited to share this presentation as it provides a strong example of how researchers can leverage Cornerstone to accelerate the time needed to reach analysis-ready data and maximize data quality.
Presenters:
Clara Oromendia, MS (CPO, Cornerstone AI)
Cara Carty, PhD (Senior Principal Scientist, BMS)
You can access a copy of the presentation here, and we’ve included a few highlights below:
BMS’ Challenge: Curating Messy, Complex Lab Data For Preclinical Research
The team collaborates with AGD to leverage a linked, longitudinal dataset comprised of EHR and genomic data. Currently, analysts need to complete various manual steps when working with the data - such as identifying patients with a certain condition based on lab values - which can lead to challenges, including:
Test names can be misspelled, inconsistent, and/or missing
Irrelevant tests can be captured by simple text string searches
LOINC codes may not be assigned, old codes may not be taken into account for capturing tests
Heterogeneous unit data - including misspellings, inconsistencies, need to convert units, biologically implausible values, etc.
These issues are time consuming to address and can compound quickly as teams need to analyze lab data across multiple tests, patient cohorts, etc - often requiring weeks of manual data cleaning. A solution was needed to avoid delaying preclinical teams’ analyses and to ensure high quality output.
Examples of ways in which Cornerstone’s data curation increased cohort size and precision across several tests of interest.
Cornerstone Solution: AI-powered Data Standardization and Harmonization
Cornerstone’s algorithms helped standardize and harmonize the full dataset rapidly, including:
LOINC prediction and assignment
Component splitting
Unit standardization
Error detection (e.g., misassigned LOINC codes)
The result: BMS received a cleaned data file including ~250 million curated lab measurements to support multiple internal analyses (e.g., disease progression, safety), with the team noting meaningful increases in speed to data availability, test coverage, and reproducibility.
Looking Forward
We’re excited to share this project with the RWD research community as it highlights how AI-powered tools are well positioned to solve these types of challenges. Effective data cleaning and data quality infrastructure is essential to ensuring that high quality research can be conducted efficiently - and we look forward to sharing more case studies with our biopharma and data ecosystem partners soon.
Thank you to Cara, BMS, and our IMPACCT RWE Summit hosts for the opportunity! See below for a few more pictures from the event.
About Cornerstone AI
Cornerstone AI is a healthcare data platform that automates assessment, standardization, schema conversion, and error detection for leading life sciences and healthcare technology companies, resulting in more utilizable records, higher quality data, and faster insights. As a neutral third party working at the intersection of consumers and producers of healthcare data, Cornerstone is utilized by leading companies for rapid, consistent, and comprehensive AI-powered data quality assessments to inform decision-making. Learn more at www.cornerstone.ai.