Table 2.

Features and delivery of just-in-time adaptive interventions

Author (year)	Real-time support: when provided and how triggered	Type of data used for real-time support, software and hardware	Content of real-time support	Theory base	Intervention duration
Bond et al. (2014) [30] Thomas & Bond (2015) [47]	Depended on the condition (three in total): 1) 3-mins break after 30 continuous sedentary minutes; 2) 6-mins break after 60 continuous sedentary minutes; 3) 12-mins break after 120 continuous sedentary minutes. Objective real-time data.	Sedentary behaviour time. Participants received an Android phone (Samsung Exhibit 4G SGH-I759).	Prompt to take a break from sedentary behaviour.	Not reported.	3 weeks.
Ding et al. (2016) [31]	When opportunistic walking moments were sensed. The message included a goal (‘take another 213 steps to reach 3000 steps!’) and a complimentary message was sent when this short-term goal was achieved. Objective real-time data.	Physical activity levels, sedentary behaviour time, smartphone use. Smartphone, Pebble smartwatch (smartwatch accelerometer used to determine whether user was eating), Android.	Prompt to walk and prompt to walk more (when already walking).	Fogg Behaviour Model, Goal Setting Theory (Locke and Latham), Habit Formation Theory.	3 weeks.
Finkelstein et al. (2015) [32] Ouyang (2015) [48]	When the person walked less than 15 steps in the past hour. No messages were sent during blackout conditions: 1) self-reported preferences collected from participant at enrolment; 2) participant texted S(X) (no messages for the next X hours), 3) participant texted ‘okay’ which meant no messages were sent during the next hour. Objective real-time data.	Physical activity levels, sedentary behaviour time. Fitbit and Android smartphone were given to the participants	Prompt to take a break from sedentary position: a tailored text message that sedentary period exceeded healthy limits, suggestions from message library on ways to have short activity breaks at work or at home, depending on the time of day.	Not reported.	4 weeks.
Gouveia et al. (2015) [35]	When participants were sedentary for 45 min. Possibly other times and locations too but authors do not report details. Objective real-time data.	Physical activity levels, sedentary behaviour time, and location-based sensor (details not reported). Smartphone, Android OS.	Prompt to take a break from sedentary position. Possibly other content too given 91 different textual messages.	Possibly Self-determination Theory (mentioned in the introduction but not intervention description); and Transtheoretical Model (used stages to categorise adoption).	10 months, but data were only analysed over a 12-week period from downloading the app.
He & Agu (2014) [36]	When participant was inactive (90% of the last 30 mins), sitting or showed sedentary patterns for an extended time. Objective real-time data.	Physical activity levels, sedentary behaviour time. GPS and time. Android OS 4.0+. Google Nexus 4 smartphone for development and testing.	Suggestions for physical activities, stand up and take a walk (when sedentary for 27 min).	Not reported.	2 weeks.
Hermens et al. (2014) [49] Tabak (2014) [20]	Suitable situations for delivery of motivational coaching (predicted by analysing previous cues and learning when a patient was likely to respond well to the message by relating relevant context factors to patient compliance and content). Objective real-time data.	Physical activity levels, previous motivational cues and ‘relevant context factors’. Hermens: smartphone and activity sensor (ProMove 3D wireless activity tracker). Tabak: HTC Desire S (smartphone) and Inertia Technology B.V. (accelerometer)	Motivational message (Hermens), prompt to walk (Tabak). Messages were encouraging, discouraging or neutral, based on physical activity levels measured in real-time.	Possibly Stages of change (mentioned in introduction and discussion but not intervention description (Tabak)).	3 months. Participants were asked to use the app at least four days per week.
Lin et al. (2011) [27] Lin (2013), Chapter 5	The system queried the geo-database, electronic diary, user profile, time and weather service, and sent support when all conditions for physical activity were met. Objective real-time data.	GPS/GSM localisation, electronic diary, weather, time and participant profile. Smartphone. Android version 2.0 and above. Software was built on Ruby on Rails platform. Participants used HTC Hero and Samsung GalaxyS.	Suggestions for physical activity.	Not reported.	4 weeks.
Lin (2013), Chapter 6	See Lin (2011)	See Lin (2011)	See Lin (2011)	Not reported.	5 weeks.
Pellegrini et al. (2015) [34]	When sedentary for more than 20 min as assessed by the accelerometer, a reminder prompt was triggered encouraging the participant to stand up for at least two minutes. Objective real-time data.	Sedentary behaviour time from a wireless accelerometer. Smartphone (Android).	Prompt to take a break from sedentary position.	Not reported.	4 weeks.
Rabbi et al. (JIMR, 2015) [28]	Based on automated sensing (accelerometer and GPS) when participants were in specific locations (on the way to work) or sedentary for prolonged period. Objective real-time data.	Physical activity levels, sedentary behaviour time. GPS. Smartphone, Android.	Suggestions for physical activities.	Learning theory, Social Cognitive Theory, Fogg’s Behaviour Model.	3 weeks.
Rabbi et al. (UBICOMP, 2015) [25]	See Rabbi et al. (JIMR 2015).	See Rabbi et al. (JIMR 2015).	See Rabbi et al. (JIMR 2015).	Decision-making theory models: multi-armed bandit and pareto-frontier. Fogg’s Behaviour Model, Social Cognitive Theory.	9 weeks (delivery ranged between 7 and 9 weeks).
Rajanna et al. (2014) [37]	After a period of sedentary time. Objective real-time data.	Sedentary behaviour time, GPS, time of day, weather, and personal calendar. Smartphone, Android.	Suggestions for physical activities	Fogg Behaviour Model, Theory of Meaning Behaviour.	1 h for the summative evaluation.
Van Dantzig et al. (2013) [33]	Study 1: when participants were sedentary for 60 min they received a prompt to take a break of 5 min, with a general daily activity goal of 50 min. Study 2: whenever 30 mins of nearly uninterrupted computer activity was recorded, a short SMS containing a hyperlink was sent to the participant’s smartphone, when clicked they were shown a message persuading them to be more active. Objective real-time data.	Sedentary behaviour time. iPhone 3G (study 1); own smartphone (study 2). Activity monitor (study 2 only) and software installed on computer to measure computer activity by registering keyboard and mouse activity.	Prompt to take a break from sedentary position.	Social influence strategies defined by Cialdini.	1 day (study 1). 6 weeks (study 2).
Van Dantzig et al. (2018) The Netherlands [29]	Support was sent during actionable moments in personally relevant geofence zones (e.g., home, work, nature area) identified in real-time based on sensor data interpretation. Objective real-time data.	Physical activity levels, time, location, weather, and behavioural events (participants achieved a step target or set a new step record). Smartphone, wrist-worn activity tracker, Philips health watch, operating system not reported.	Suggestions for physical activity, feedback about number of steps in specific contexts.	Not reported.	1 week.