Identifying unmet needs in food protein–induced enterocolitis syndrome using search query analysis and a large language model
7 affiliations
- 1Division of Immunology, Boston Children’s Hospital, Harvard Medical School, Boston, Mass
- 2Tulane University School of Medicine, New Orleans, La
- 3Department of Community Medicine, Information and Health Decision Sciences, Faculty of Medicine, MEDCIDS, University of Porto, Porto, Portugal; Centre for Health Technology and Services Research, Health Research Network, Faculty of Medicine, CINTESIS@RISE, University of Porto, Porto, Portugal
- 4Department of Statistics, The Pennsylvania State University, University Park, Pa
- 5Boston Children’s Hospital, Boston, Mass
- 6Division of Gastroenterology & Nutrition, Boston Children’s Hospital, Harvard Medical School, Boston, Mass
- 7Division of Immunology, Boston Children’s Hospital, Harvard Medical School, Boston, Mass; Corresponding author: Lisa Bartnikas, MD, 300 Longwood Ave, Boston, MA 02115.
Abstract
Background: Food protein–induced enterocolitis syndrome (FPIES) is a non–IgE-mediated food allergy with many unanswered clinical and research questions. Existing educational resources, patient-facing education materials, and guidelines may not fully reflect families’ information needs. Digital data and artificial intelligence (AI) offer new ways to identify public concerns in rare diseases. Objective: We developed a framework utilizing online search data and AI-assisted analysis to generate data-informed questions for future patient advice guidance and educational resource development in FPIES, and evaluate these questions with parents/guardians. Methods: We collected Google Trends data from January 2023 to December 2024 to identify search queries related to FPIES in the United States. Using ChatGPT, query terms and frequencies were analyzed to generate 10 representative questions reflecting public interest. Parents/guardians of children with FPIES enrolled onto the Boston Children’s Hospital FPIES Registry were surveyed to evaluate these questions. Results: Forty-four parents/guardians completed the survey (93.2% mothers; median [interquartile range] child age was 3.0 [2.0, 5.0] years). Nearly half (43.2%) reported more than one FPIES trigger food. Nine of 10 AI-generated questions were rated “important” or “very important” by >90% of participants, confirming their relevance. Questions most frequently reported as difficult to find answers for were managing reactions, outgrowing FPIES, and distinguishing FPIES from other allergies. Similar topics were selected as research priorities. Conclusion: A framework built from Google Trends data and AI-supported analysis can generate questions that align closely with parent/guardian priorities, offering a practical approach to informing patient advice guidance and educational resource development, with the ultimate goal of supporting patient-centered guideline development in FPIES.
Related
- Vedolizumab use for severe chronic enterocolitis in a child with Hoyeraal–Hreidarsson syndrome: A case report
- MAXWELL: Calibrating the probabilistic outputs of protein language models to the mutation-induced stability change landscape
- Identifying diagnostic testing gaps for antiphospholipid syndrome within a single healthcare system
- PDI-Spot: A Robust Representation Learning Framework for Identifying Energetic Hotspots in Protein–DNA Complexes
- Limitations of general protein language models and public TCRpMHC data in training specificity prediction models
BibTeX
@article{thinhhnguyen2026identifying,
title = {{Identifying unmet needs in food protein–induced enterocolitis syndrome using search query analysis and a large language model}},
author = {Thinh H. Nguyen, MD, PhD and Mia Vaida, BS and Bernardo Sousa-Pinto, MD and Thinh Doan, MS and Danielle Crethers, BA and Brichelle Peña, BS and Emeline Kao, BS and Wanda Phipatanakul, MD, MS and Erin Syverson, MD and Elena Crestani, MD, MS and Lisa M. Bartnikas, MD},
journal = {Journal of Allergy and Clinical Immunology: Global},
year = {2026},
month = nov,
doi = {10.1016/j.jacig.2026.100776},
url = {https://doi.org/10.1016/j.jacig.2026.100776},
note = {CC BY},
}
RIS
TY - JOUR
TI - Identifying unmet needs in food protein–induced enterocolitis syndrome using search query analysis and a large language model
AU - Thinh H. Nguyen, MD, PhD
AU - Mia Vaida, BS
AU - Bernardo Sousa-Pinto, MD
AU - Thinh Doan, MS
AU - Danielle Crethers, BA
AU - Brichelle Peña, BS
AU - Emeline Kao, BS
AU - Wanda Phipatanakul, MD, MS
AU - Erin Syverson, MD
AU - Elena Crestani, MD, MS
AU - Lisa M. Bartnikas, MD
PY - 2026
DA - 2026/11/08
JO - Journal of Allergy and Clinical Immunology: Global
DO - 10.1016/j.jacig.2026.100776
UR - https://doi.org/10.1016/j.jacig.2026.100776
AB - Background: Food protein–induced enterocolitis syndrome (FPIES) is a non–IgE-mediated food allergy with many unanswered clinical and research questions. Existing educational resources, patient-facing education materials, and guidelines may not fully reflect families’ information needs. Digital data and artificial intelligence (AI) offer new ways to identify public concerns in rare diseases. Objective: We developed a framework utilizing online search data and AI-assisted analysis to generate data-informed questions for future patient advice guidance and educational resource development in FPIES, and evaluate these questions with parents/guardians. Methods: We collected Google Trends data from January 2023 to December 2024 to identify search queries related to FPIES in the United States. Using ChatGPT, query terms and frequencies were analyzed to generate 10 representative questions reflecting public interest. Parents/guardians of children with FPIES enrolled onto the Boston Children’s Hospital FPIES Registry were surveyed to evaluate these questions. Results: Forty-four parents/guardians completed the survey (93.2% mothers; median [interquartile range] child age was 3.0 [2.0, 5.0] years). Nearly half (43.2%) reported more than one FPIES trigger food. Nine of 10 AI-generated questions were rated “important” or “very important” by >90% of participants, confirming their relevance. Questions most frequently reported as difficult to find answers for were managing reactions, outgrowing FPIES, and distinguishing FPIES from other allergies. Similar topics were selected as research priorities. Conclusion: A framework built from Google Trends data and AI-supported analysis can generate questions that align closely with parent/guardian priorities, offering a practical approach to informing patient advice guidance and educational resource development, with the ultimate goal of supporting patient-centered guideline development in FPIES.
SN - 2772-8293
ER -