Machine-based Algorithm for Increased Physical Activity and Sustained User Engagement

NCT06388317 | Completed

• 1 other identifier: 123632
• Study Type: interventional
• Target: 1,249
• Locations: 1 country
• Sites: 1

Brief Summary

A cohort analytic study (i.e., observational study following a cohort over time to compare outcomes between those exposed and unexposed to an intervention) was conducted over a 17-week period (March 7, 2022 to June 30, 2022). The study consisted of a five-week baseline period (March 7, 2022 to April 10, 2022). During the baseline, Telus Wellbeing users earned FI for meeting personalized but static daily step goals. A 12-week intervention period followed (April 11, 2022 to June 30, 2022). During the intervention period, users were randomized by a senior Telus Wellbeing software engineer into either control (i.e., continuing with their personalized static goal) or intervention groups (i.e., testing the new ML-driven 'Smart Mode' feature offering personalized and adaptive daily step goals) using a random number generator in Python (version 3.9; i.e., non-pi package random binomial). The randomization ratio was about 1:2 in favour of the intervention group. Both groups were informed they would be testing a new 'Smart Mode' feature at intervention onset, however, only those allocated to the intervention group received ML-driven goals. Notably, the study sample consisted of more highly engaged users (e.g., registered on the app at least five weeks prior to the intervention, took initiative to synchronize their activity trackers with the app). The Strengthening the reporting of observational studies in epidemiology (STROBE) statement flow chart and checklist for cohort studies is provided.

Conditions

Health Behavior

Keywords

machine learning; physical activity; step goal; mHealth; financial incentives

Outcome Measures

Primary Outcomes (1)

• Step count

  Relative change in daily steps from the 5 week run-in period to the 12-week follow-up, in groups receiving adaptive daily step goals versus those receiving non-adaptive goals.

  17 weeks

Secondary Outcomes (1)

• Number of app opens, total time spent on app, number of pages opened

  12 weeks

Study Arms (2)

Adaptive Step Goal (Intervention)

EXPERIMENTAL

The 'Smart Mode' feature used a propriety ML algorithm to generate 16 user clusters based on daily step count patterns from approximately 100,000 global Telus Wellbeing users over the previous two years (March 1, 2020 to March 1, 2022; characteristics of these users unknown). The algorithm then compared time series data from participants' five-week baseline against the repository of 16 user cluster patterns to determine to which cluster they would be assigned. Once assigned, a difference in proportions was calculated (ratio of previous week's average daily step count over the step count goal for that week vs. ratio of average daily step count from two weeks prior over step count goal for that week \[e.g., 8000 steps/7500 steps=1.067 vs. 8000 steps/9500 steps=0.8421\]). Chi-square testing was used to assess statistical significance of the difference (p\<0.05). This was done to determine whether the next calculated goal would go up, down, or stay the same.

Behavioral: Adaptive Step Goal (Intervention)

Static Step Goal (Control)

ACTIVE COMPARATOR

Over the course of the 12-week intervention period controls continued with their static daily step goal, equivalent to their baseline period weekly average daily step count. Control participants received an out-of-app push notification at 8:00 a.m. each Monday during the intervention period reminding them of their static daily step goal. If the participant achieved their daily step goal before 8:00 p.m. on any day a push notification was sent to congratulate them on reaching their goal for the day. No notifications were sent past 8:00 p.m. Participants achieving daily step goals were rewarded with FI in the form of points (i.e., "SproutBucks"). They could redeem "SproutBucks" either "on-platform" (e.g., gift cards to StarbucksTM, BestBuyTM, iTunesTM, VisaTM) or "off-platform" (e.g., employer specific rewards like vouchers for fitness membership discounts). The value of "SproutBucks" was unique to companies, ranging from $0.00 to $1.00 USD per daily goal achieved.

Behavioral: Static Step Goal (Control)

Interventions

Adaptive Step Goal (Intervention) BEHAVIORAL

Receiving automated personalized daily step goals,

Adaptive Step Goal (Intervention)

Static Step Goal (Control) BEHAVIORAL

Receiving personalized static daily step goals.

Static Step Goal (Control)

Eligibility Criteria

• Age: 18 Years+
• Sex: all
• Healthy Volunteers: Yes
• Age Groups: Adult (18-64), Older Adult (65+)

You may qualify if:

• Adults aged 18 years or older
• User intent to become physically active in the next 12 weeks
• Own a smartphone device
• Willing to install and use the Telus Wellbeing app (which requires Internet connection) every day for 12 weeks
• Ability to speak and read English

You may not qualify if:

• Known medical conditions or physical problems that require special attention in a physical activity program
• Planning an international trip during the next 3 months, which could interfere with daily server uploads of mobile phone data
• Pregnant or gave birth during the past 6 months
• Current participation in lifestyle modification programs or research studies that may confound study results

Sponsors & Collaborators

• Western University, Canada: lead

Study Sites (1)

Western University

London, Ontario, Canada

Location

Related Publications (3)

• Michie S, van Stralen MM, West R. The behaviour change wheel: a new method for characterising and designing behaviour change interventions. Implement Sci. 2011 Apr 23;6:42. doi: 10.1186/1748-5908-6-42.

  PMID: 21513547 BACKGROUND

• Mitchell MS, Orstad SL, Biswas A, Oh PI, Jay M, Pakosh MT, Faulkner G. Financial incentives for physical activity in adults: systematic review and meta-analysis. Br J Sports Med. 2020 Nov;54(21):1259-1268. doi: 10.1136/bjsports-2019-100633. Epub 2019 May 15.

  PMID: 31092399 BACKGROUND

• Armijo-Olivo S, Stiles CR, Hagen NA, Biondo PD, Cummings GG. Assessment of study quality for systematic reviews: a comparison of the Cochrane Collaboration Risk of Bias Tool and the Effective Public Health Practice Project Quality Assessment Tool: methodological research. J Eval Clin Pract. 2012 Feb;18(1):12-8. doi: 10.1111/j.1365-2753.2010.01516.x. Epub 2010 Aug 4.

  PMID: 20698919 BACKGROUND

MeSH Terms

Conditions

Health Behavior; Motor Activity

Interventions

Methods

Condition Hierarchy (Ancestors)

Behavior

Intervention Hierarchy (Ancestors)

Investigative Techniques

Study Design

• Study Type: interventional
• Phase: not applicable
• Allocation: RANDOMIZED
• Masking: SINGLE
• Who Masked: PARTICIPANT
• Masking Details: Participants are unaware of the other intervention arm (i.e., adaptive step goals versus static step goals).
• Purpose: TREATMENT
• Intervention Model: PARALLEL
• Sponsor Type: OTHER
• Responsible Party: SPONSOR

Model Details: Retrospective, cohort-analytic study.

Study Record Dates

• First Submitted: July 29, 2023
• First Posted: April 29, 2024
• Study Start: January 1, 2023
• Primary Completion: March 1, 2025
• Study Completion: March 24, 2025
• Last Updated: April 25, 2025

Record last verified: 2024-04

Data Sharing

• IPD Sharing: Will not share

Locations

CA (1)