Usefulness of web search queries for early detection of diseases in infants

Area of Science:

• Pediatric Medicine
• Computational Health Informatics
• Public Health Surveillance

Background:

• Early disease detection in infants is crucial for timely intervention and improved health outcomes.
• Guardians often use web searches to understand infant symptoms, presenting a potential data source for health monitoring.
• Biliary atresia and hypertrophic pyloric stenosis are significant infant conditions requiring prompt diagnosis.

Purpose of the Study:

• To evaluate the predictive accuracy of guardian web search queries for infant diseases.
• To explore the potential of machine learning models in identifying disease patterns from search data.
• To encourage earlier medical consultation by leveraging online search behavior.

Main Methods:

• Collected six months of Yahoo! JAPAN Search queries (October 2016 - March 2017).
• Developed and applied a machine learning model to analyze search query data.
• Investigated the predictive performance for biliary atresia and hypertrophic pyloric stenosis.

Main Results:

• The machine learning model achieved approximately 80% accuracy in predicting both biliary atresia and hypertrophic pyloric stenosis.
• Symptoms associated with these diseases were identified as significant predictive features in the model.
• Search queries demonstrated a notable ability to predict the diagnosis of these infant conditions.

Conclusions:

• Web search queries show promise as a tool for the early detection of specific infant diseases.
• Machine learning analysis of search data can identify patterns indicative of serious pediatric conditions.
• Future research can develop methods to utilize search data for timely guardian support and intervention.