This breakthrough, allowing health systems to track long-term maternal-child health outcomes using routine electronic health record data.
The ability to connect a mother’s medical history with her child’s health records has long been a significant hurdle for public health research. Electronic health record (EHR) systems typically operate in silos, often failing to standardize data across generations. A new probabilistic linkage algorithm developed by researchers at the Regenstrief Institute and Indiana University School of Medicine aims to bridge this gap, enabling large-scale observational studies that were previously considered impossible.
Machine Learning Performance and Methodology
Dr. Colin Rogerson, a research scientist at Regenstrief and the Indiana University School of Medicine, emphasized the novelty of this achievement.
“No one before this has been able to do what we’ve done here. Other researchers have tried this with administrative or state-level data, but it has been hard to generalize their results. Our approach uses standard information that every hospital collects, which means other states and health systems should be able to use the same algorithm and achieve similar results.”
Dr. Colin Rogerson, Research Scientist, Regenstrief and IU School of Medicine
Predicting Pediatric Asthma Readmissions
The application of machine learning in pediatric care extends beyond record linkage to predictive modeling for specific chronic conditions. A study published in Pediatric Pulmonology on April 15, 2026, examined the use of an XGBoost model to predict 180-day readmissions for children hospitalized due to asthma. Gabbay of the Albert Einstein College of Medicine, analyzed 173,730 encounters across 47 U.S. children’s hospitals between January 2016 and December 2024.
As Medscape reported, the model demonstrated a statistically significant improvement over conventional regression methods. The ML approach achieved an area under the receiver operating characteristic curve (ROC AUC) of 0.718, compared to 0.702 for standard regression models. While the performance was described as moderate, the authors noted the potential for clinical application.
“Despite the moderate performance of an ML model, this study represents an important benchmark in the use of administrative data for pediatric readmission prediction, demonstrating that even limited data may support early identification of high-risk children that is generalizable across children’s tertiary hospitals in the United States.”
Study authors, via Medscape
Clinical Decision Support and Future Research
In a separate pilot randomized clinical trial registered on April 12, 2023, researchers investigated the usability of an EHR-integrated Passive Digital Marker (PDM) for assessing asthma risk in preschool-aged children. This trial, which enrolled 34 pediatricians in Indiana, utilized a Solomon 4-group design to isolate the effectiveness of AI-driven decision support tools.
These developments signify a broader trend in medical informatics.
“By leveraging high-quality real-world data and modern machine learning, this work demonstrates how we can responsibly apply AI to answer questions that matter for public health. The ability to generate reliable maternal-child linkages across systems opens the door to discoveries that weren’t possible before.”
Dr. Shaun Grannis, Vice President for Data and Analytics, Regenstrief
As researchers move forward, the focus remains on the scalability of these models. The primary limitation identified in recent asthma readmission research involves the inability to track patients across hospital networks outside of the specific study database, as well as the exclusion of outpatient visits and prescription medication data. Future iterations of these algorithms aim to incorporate broader data sets to refine accuracy and generalizability across diverse health systems.