Using Ultra-Abridged Individual Difference Scales for Personalization in Digital Mental Health to Improve Uptake, Engagement, and Experiences: Three-Tiered Decision Framework for Scale Shortening

Clinical Snapshot

Key Findings

• Effect Size: Not applicable - theoretical framework

• Primary Outcome: Three-tiered decision framework for scale shortening method selection

• Nnt Or Sensitivity: Not applicable - no diagnostic or therapeutic outcomes measured

• Confidence Interval: Not provided

Clinical Application

Moderate - requires technical expertise and validation studies before implementation Relevant to Australian digital health initiatives, but would require local validation studies and consideration of TGA digital therapeutics guidelines Digital mental health platform developers and researchers seeking personalization strategies

Abstract

Given the diversity of human characteristics and experiences, personalization in nudges, messages, choice presentations, interventions, and overall product design has been increasingly adopted in digital health to promote engagement. Past studies on moderators and personalization in digital health and mental health services generally focused on demographic and symptom variables, with generally inconsistent findings or null findings. Cognitive, motivational, and decisional psychological attributes are largely overlooked. Psychology often uses long self-report scales to measure various psychological attributes. Although they are useful in tapping into individuals' psychological profiles, when applied in real-life, everyday settings to assess individual differences, people are most likely unwilling to complete them. With the pressing need to personalize digital health platforms to enhance uptake, retention, and engagement, ultrashort versions of these psychological scales may be considered to allow assessment of multiple attributes at the same time. Scale shortening can be achieved through regression analyses of each item, factor analyses, item response theory, ant colony optimization, and machine learning methods, with each method having advantages, disadvantages, and conditions required to make it suitable. To illustrate, we provided examples of regression analyses of each item and factor analyses, with potential implications for personalizing narrative versus research-based messages in digital mental health contexts. We present a 3-tiered decision framework for scale shortening method selection depending on goals and possible constraints, with guidelines on validation methods for ultrashort scales. Moving forward, more validation studies and field studies in digital health platforms are needed to evaluate the ecological validity, reliability, and generalizability of these methods, bearing in mind the limitations and conditions where such shortening methods may not work well. Researchers may compare the effectiveness and limitations of personalization using ultrashort scales with other commonly adopted personalization methods (eg, based on longer scales, behavioral data, and large language models). Ethical concerns need to be considered and mitigated carefully, respecting diverse preferences, informed choices, and the privacy of service users. Our viewpoint piece is primarily intended for digital mental health researchers and practitioners, but may also be informative for the fields of digital health and medicine as well as personalization (eg, personalized health care, personalized nudging, and message matching) more broadly, given the common goal of boosting uptake and engagement as well as improving service users' experiences.

References

1. 1.Yeung, S. K., Tong, A. C. Y., Zhao, H., & Mak, W. W. S. (2026). Using Ultra-Abridged Individual Difference Scales for Personalization in Digital Mental Health to Improve Uptake, Engagement, and Experiences: Three-Tiered Decision Framework for Scale Shortening. Journal of Medical Internet Research, 28(6). https://doi.org/10.2196/80662