Definitions of digital biomarkers: a systematic mapping of the biomedical literature

Abstract

Background Technological devices such as smartphones, wearables, sensors, or virtual assistants allow to collect data on health and disease processes and are thus increasingly considered a useful digital alternative to conventional biomarkers. We aimed to provide a systematic overview of the emerging literature on "digital biomarkers" with their definitions, features, and citations in biomedical research. Methods We analyzed all articles in PubMed that used the term "digital biomarker(s)" in title or abstract, considering any study involving humans and any review, editorial, perspective, or other opinion-based article up to 8 March 2023. We systematically extracted characteristics of publications and research studies, and any definitions and features of "digital biomarkers" mentioned. We described the most influential literature on digital biomarkers and their definitions using thematic categorizations of definitions considering the FDA BEST framework (i.e., data type, data collection method, purpose of biomarker), analysing the structural similarity of definitions by performing text analyses (hierarchical clustering on the distance-matrix) and citation analyses (based on citation metrics obtained from OpenAlex via Local Citation Network; last search 26 June 2023). Results We identified 415 articles prominently using the term "digital biomarker". They were published between 2014 and 2023 (median 2021), mostly describing primary research (283 articles; 68%). Most articles did not provide a definition of a digital biomarker (n=287; 69%). The 128 articles providing a definition of a digital biomarker reported 127 different definitions. Of these 127 definitions, 78 considered data collection, 56 data type, 50 the purpose, and 23 were based on all three key components. The 128 articles with a definition were cited a median of 6 times (interquartile range 2-20) with up to 517 citations. Of the ten most frequently cited articles using a definition, all used a different one. Conclusions The most frequently used definitions for digital biomarkers are highly different and there is no consensus about what this emerging term means. Our overview highlights key defining characteristics of digital biomarkers which can inform the development of a harmonized and more widely accepted definition.

Machine Learning and Digital Biomarkers Can Detect Early Stages of Neurodegenerative Diseases

“…Similarly, Babrak (2019) discusses the basic differences and similarities between traditional and digital biomarkers and points out the necessity for synchronization and unique feature definition claiming, therefore, that the unclear definition of digital biomarkers, population groups, and their intersection with traditional biomarkers hinders their discovery and validation [ 22 ]. Finally, and recently, Alonso et al (2023) even concluded that there is no consensus about what this emerging term means [ 75 ].…”