Machine learning in healthcare citizen science: A scoping review

Area of Science:

• Healthcare research
• Citizen science
• Machine learning

Background:

• Citizen science is increasingly integrated into healthcare research.
• Understanding the scope and impact of citizen involvement in health projects is crucial.
• This review clarifies the role of citizen scientists in health-related machine learning.

Approach:

• A comprehensive search of major databases (PubMed, Scopus, Web of Science, EBSCOhost) was conducted in January and September 2024.
• Peer-reviewed publications combining citizen science and machine learning in healthcare were included.
• Articles were excluded if citizens were passive data providers or if only professional scientists were involved.

Key Points:

• 56 articles from 2013-2024 met inclusion criteria, primarily from the U.S. and Europe.
• Citizen scientists most commonly collected (51.8%) and annotated (26.8%) data for machine learning.
• Mosquitoes and air pollution were key data types, with classification tasks and Convolutional Neural Networks being prevalent ML methods.

Conclusions:

• Citizen science in healthcare is predominantly an American and European phenomenon with emerging Asian contributions.
• Citizen engagement is mainly focused on data provision and annotation, with less involvement in data analysis or study leadership.
• Distinguishing projects based on the degree of citizen involvement is essential for accurate categorization.