# HappyMums mobile application study protocol: use of a smartphone application to gather data predictive of antenatal depression

## TL;DR

The HappyMums app uses smartphone data to predict antenatal depression in pregnant people across six European countries.

## Contribution

This study introduces a novel mobile application combining passive and active data to predict antenatal depression using machine learning.

## Key findings

- The app will assess user engagement and usability in predicting mental health outcomes.
- Machine learning models will be tested for their ability to identify depression risk during pregnancy.
- Findings will be shared with both academic and non-specialist audiences via open access platforms.

## Abstract

Mobile health (mHealth) technologies have become increasingly popular for monitoring mental health symptoms and lifestyle behaviours, and are largely reported to be feasible and acceptable to users. However, to date, the efficacy of such technologies to improve perinatal mental health outcomes has been mixed. Within the perinatal context, much of this work has been done in the context of postpartum depression, stemming from electronic health records as well as cohort studies. There is, however, a dearth of studies focusing on depression in pregnancy, and machine learning-based clinical decision support systems remain underexplored. The HappyMums application has been developed to meet this need, and its use across Europe will be tested in this study.

A total of 1000 pregnant people currently suffering from, or at risk of, antenatal depression will be recruited across six countries. All participants will be between 13 and 28 weeks’ gestation and will be given access to the new purposefully developed HappyMums mobile application, to use from enrolment until 2 months postpartum. The application leverages passively collected data from smartphone sensors relating to physical activity and behaviour, as well as requiring active engagement from the user to complete mental health questionnaires and ‘game-like’ activities. Digital data types will be combined with traditional mental health measurement methods, such as standardised questionnaires and interviews, to develop novel predictive models capable of identifying mental health trajectories in women at risk of developing antenatal depression and to test the app’s utility for use as personalised risk prediction and depression identification tool. The primary outcome of this study is to determine what proportion of users will continue to use the mobile application and engage with its tasks and activities at least weekly, while secondary exploratory outcomes include assessing usability of the app and testing the predictive ability of a novel machine learning-based model. These outcomes will, for the first time, be assessed by integrating active as well as passive data.

Ethical approval has been granted by local research ethics committees in each recruiting centre. At King’s College London (leading the clinical study), the study was reviewed by the East of England—Essex Research Ethics Committee and granted favourable opinion (REC reference 24/EE/0129). All other sites collecting participant data have the study approved for local delivery. Findings relating to the primary and secondary outcomes will be submitted for publication in open access, peer-reviewed journals, as well as presentations at conferences as symposia or posters. Findings will be made available to a non-specialist audience through open access digital mental health magazines and promotion on social media.

NCT06578845.