PhD Scientific Days 2025

Mental Health Sciences I.

How Algorithms Shape Online Health Information-Seeking: A Case Study on Cow's Milk Protein Allergy in Hungarian Google Searches

Name of the presenter

Kovácsné Hegedűs Dóra

Institute/workplace of the presenter

Semmelweis University

Authors

Kovácsné Hegedűs Dóra¹, Dr. Sztárayné Dr. Kézdy Éva², Lengyel Lívia¹, Dr. Hidvégi Edit Phd³, Prof. Dr. Albert Fruzsina¹, Susovits Kitti⁴

1: Semmelweis University
2: Károli Gáspár University of the Reformed Church in Hungary
3: Semmelweis University Clinic of Pulmonology
4: .

Text of the abstract

Introduction:
Search engines not only organize but also shape how users think about health. While digital health research often relies on English-language data, the influence of search algorithms in smaller linguistic communities is understudied. This study focuses on Hungarian-language searches related to cow's milk protein allergy (CMPA)—a misunderstood condition often confused with lactose and milk protein intolerance.
Aims:
We aim to examine how algorithmic tools such as Google Ads and Google Autocomplete influence health information-seeking behavior regarding CMPA in Hungary. The study explores thematic patterns, search volume trends, and the emergence of misinformation through digital recommendation systems.
Methods:
We conducted a mixed-methods infodemiological analysis using quantitative data from Google Ads Keyword Planner (2020–2023) and qualitative data from Google Autocomplete. Keyword suggestions for "milk allergy", "milk protein allergy," and "cow's milk protein allergy" were categorized and analyzed. Quantitative analysis was performed using Stata 18, while qualitative coding was conducted in Atlas.ti using thematic content analysis.
Results:
CMPA-related search volumes increased annually, with seasonal peaks in January. Among all keyword suggestions, 37% addressed symptoms (especially dermatological), 25% concerned nutrition, and 24% reflected terminological confusion. Although diagnostic terms were less frequent, interest in such queries grew steadily. Autocomplete suggested non-evidence-based tests (e.g., IgG tests) over professional guidelines. Algorithmic prioritization often favored commercial over medical content. Infant-related searches were the most dominant, suggesting parents were engaging in proxy-seeking behavior.
Conclusion:
Search engines do more than guide users—they shape what users find. Our findings highlight how algorithmic design influences online health information, leading to conceptual confusion and potential overdiagnosis, especially among concerned parents. This study demonstrates how the Nobeco Effect operates within a lesser-studied language space, revealing risks and opportunities in digital health literacy.
Funding:
This research was conducted without external funding.