A methods study of mental-health research in China identifies recurring workflow gaps in ecological momentary assessment (EMA), the repeated prompting approach used during a study. It proposes an implementation-oriented way to assess localized platforms and applies it in a preliminary evaluation of Huixin EMAI.
To map the field, the team searched CNKI and Wanfang from database inception through October 31, 2025. Two researchers screened the records independently, resolving disagreements through discussion or senior adjudication.
The search produced 241 records. After 32 duplicates were removed, 209 unique records were screened and 66 empirical studies were included. Instant-messaging workflows were most common, appearing in 36 studies, or 54.5%; 14 studies, or 21.2%, used specialized EMA platforms, while 11 used paper diaries and five did not report the tool.
A framework for the whole workflow
Six experts with at least five years of relevant experience refined an initial set of 26 indicators through a moderated workshop held over two rounds. The resulting MEPEF framework contains 43 observable indicators across six domains.
The framework sets targets for research-grade protocol control, lower-burden implementation, multimodal data integration, analysis-ready data, privacy-preserving local deployment and participant accessibility. The team then benchmarked seven anonymized platforms through operation, trial use, installation and documentation review. The comparison was exploratory, not a definitive ranking.
Across 1,806 indicator-level ratings, 113, or 6.3%, were left unrated. Agreement between raters was relatively high, with average-measures intraclass correlation coefficients ranging from 0.78 to 0.97, although access and verification depth varied and some functions could not be verified.
Huixin handled most prompts, but not all
The Huixin case involved 48 adults: 26 healthy participants and 22 participants with anxiety disorders, using 27 Android and 21 iOS devices. Monitoring lasted about eight study days per participant, with six randomly scheduled prompts each day.
The deployment recorded 1,893 of 2,472 expected prompts, a completion rate of 76.6%. Participants averaged 4.52 completed EMAs per day across 8.58 study days. Among valid prompted responses, the median time to answer was four minutes, with the middle half arriving between zero and 13 minutes. The study reported an 82.1% match in timing between subjective reports and objective data. The latency analysis covered 1,122 prompted responses and excluded 771 resident questionnaires.
In the study-specific questionnaire, 44 respondents gave favorable ratings for understanding the system, using it as part of normal phone use, ease, display, completing the study over multiple days and completing four or more questionnaires per day. The reported item means ranged from 4.11 to 4.64.
Still, timely completion was often difficult. Being too busy and failing to notice a notification were each reported by 31 of 44 respondents, or 70.5%; 14 respondents, or 31.8%, said they had been asleep.
A preliminary implementation test
Researchers interviewed six people with at least three months of Huixin Research experience and analyzed the discussions using the framework’s six domains. They described support for routine EMA deployment, but also reported continuing friction in researcher-side control, delivery monitoring, multimodal stability, downstream data usability, governance support and interface smoothness.
Exploratory comparisons found higher prompt completion among healthy participants than among participants with anxiety disorders, 82.0% versus 69.8%, and more completed EMAs per day, 4.94 versus 4.03. Android users had a longer mean response latency than iOS users, 11.32 minutes versus 5.58 minutes. The comparisons had no multiplicity adjustment and should be read cautiously.
The paper frames the work as an implementation assessment, not a definitive test of platform superiority or clinical outcomes. The Huixin case was relatively short, involving 48 adults monitored for about eight days per participant; the questionnaire was study-specific and descriptively summarized. The authors also note that the framework requires independent external validation.
The work was supported by the Construction Fund of Key Medical Disciplines of Hangzhou under grant 2025HZZD14.
Paper data and sources
Original title: Localized Ecological Momentary Assessment for Mental Health Research in China: An Implementation-Oriented Framework and Preliminary Case Application
Authors: Xinying Zhao, Yue Li, Jiafeng Wang et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-20
DOI: Not available
Original paper · Full text