This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on http://formative.jmir.org, as well as this copyright and license information must be included.
Intelligent personal assistants such as Amazon Echo and Google Home have become increasingly integrated into the home setting and, therefore, may facilitate behavior change via novel interactions or as an adjunct to conventional interventions. However, little is currently known about their potential role in this context.
This feasibility study aims to develop the Intelligent Personal Assistant Project (IPAP) and assess the acceptability and feasibility of this technology for promoting and maintaining physical activity and other health-related behaviors in both parents and children.
This pilot feasibility study was conducted in 2 phases. For phase 1, families who were attending a community-based weight management project were invited to participate, whereas phase 2 recruited families not currently receiving any additional intervention. Families were randomly allocated to either the intervention group (received a smart speaker for use in the family home) or the control group. The IPAP intervention aimed to promote positive health behaviors in the family setting through utilization of the functions of a smart speaker and its linked intelligent personal assistant. Data were collected on recruitment, retention, outcome measures, intervention acceptability, device interactions, and usage.
In total, 26 families with at least one child aged 5 to 12 years were recruited, with 23 families retained at follow-up. Across phase 1 of the intervention, families interacted with the intelligent personal assistant a total of 65 times. Although device interactions across phase 2 of the intervention were much higher (312 times), only 10.9% (34/312) of interactions were coded as relevant (related to diet, physical activity or well-being). Focus groups highlighted that the families found the devices acceptable and easy to use and felt that the prompts or reminders were useful in prompting healthier behaviors. Some further intervention refinements in relation to the timing of prompts and integrating feedback alongside the devices were suggested by families.
Using intelligent personal assistants to deliver health-related messages and information within the home is feasible, with high levels of engagement reported by participating families. This novel feasibility study highlights important methodological considerations that should inform future trials testing the effectiveness of intelligent personal assistants in promoting positive health-related behaviors.
ISRCTN Registry ISRCTN16792534; http://www.isrctn.com/ISRCTN16792534
The high incidence of childhood obesity has been well documented, with 29% of children aged 2 to 15 years in England [
Interventions to promote healthy behaviors in children have largely focused on the school setting [
Alongside family involvement, incorporating technology within the family setting has been identified as a potential means of enhancing the effectiveness of interventions targeting childhood obesity [
Researchers and practitioners have used technology to change how we deliver interventions (eg, moving from print-based information to web-based resources) and how we incorporate behavior change techniques within interventions. To date, interventions using interactive electronic media [
There is a strong need for research studies to target the family setting [
This study (1) outlines the development of the GetAMoveOn+ Intelligent Personal Assistant Project (IPAP), (2) compares the acceptability of intelligent personal assistants alongside an existing intervention or as a stand-alone intervention, and (3) evaluates the potential of intelligent personal assistants for promoting and maintaining physical activity and other health-related behaviors in both parents and children.
IPAP was a 12-week RCT conducted in 2 phases. Phase 1 was an RCT that evaluated the effect of a home-based intelligent personal assistant intervention on obesity-related behaviors (diet and physical activity) in families attending a community-based weight management project.
Phase 2 was an RCT that evaluated the effect of the home-based intelligent personal assistant in families not attending a weight management project. Randomization for both phases of recruitment took place at the family level, with families (a parent and 1 or 2 children) randomly allocated to an intervention or control group. Randomization was performed by a university staff member who was independent of the research team. Sealed, opaque envelopes were used to randomly assign families to a study arm.
Families were eligible to participate when at least one child (aged 5-12 years) and one parent or adult responsible for their care consented to participate in the study. Given the nature of the intervention, access to internet connection with their home (Wi-Fi) and ownership of one smart device within the home (eg, a tablet or smartphone) or access to a computer or laptop to enable the family members to interact with the home-based intelligent personal assistant was required. The adult and child or children taking part in the study also had to live within the same household. No restrictions were placed on the family type. No inclusion criteria were placed on parents or children in relation to any medical condition. Participants were asked to notify the research team of any related issues that might affect participation in the intervention. No issues that limited or affected participation or resulted in adverse events were reported.
All families (n=16) attending a community-based obesity prevention project, Safe Wellbeing Eating & Exercise Together (SWEET) as a family, were invited to participate in the study. The SWEET project is a community-based obesity prevention and management program aimed at children and families across a number of sites (community organizations, healthy living centers, etc) in the Western Trust area of Northern Ireland. It aims to work with families in areas of high economic deprivation and targets lifestyle characteristics, such as dietary habits, physical activity, and mental well-being. Families are recruited to the SWEET project via social media sites, flyer distributions in schools, and local paper advertisements. Before approaching families, permission was obtained from the Healthy Lifestyle Coordinator of the Healthy Living Centre where the project was being delivered. Members of the research team attended the first session of the project and provided a verbal overview of the research study. Written informed consent was obtained from all parents or guardians, and written parental consent and child assent were obtained for each child. Phase 1 of the study was conducted from January to April 2019.
Phase 2 was subsequently undertaken to further assess the acceptability of intelligent personal assistants as a stand-alone intervention. Potentially eligible families (as mentioned earlier) were invited to take part in the study (not restricted to those attending the SWEET project) through a number of recruitment strategies. Local community group leaders were contacted and asked to provide permission for a member of the research team to approach families (parents) at relevant events, for example, parent or child groups, youth club, sports training sessions etc. Similar to phase 1, prospective families were provided with a verbal overview of the study and detailed written information on the study. Written informed consent was obtained from all parents or guardians, and written parental consent and child assent were obtained for each child. Efforts were made by the research team to ensure families in phase 1 and phase 2 were recruited from similar community groups to avoid any potential sampling bias. Phase 2 of the study was conducted from May to August 2019. Families were only able to participate in one phase, that is, families who took part in phase 1 were not eligible to take part in phase 2.
A smart speaker (Amazon Echo) and its linked intelligent personal assistant (Amazon Alexa) were chosen as the tools for intervention delivery in this study. A market survey (n=2274) highlighted that 33% of respondents based in the United States and the United Kingdom owned a smart speaker [
Intelligent personal assistants can perform a range of basic home assistant functions, including playing music, setting alarms, checking the weather, and searching for information. Users can also personalize the devices by adding apps or
Following the completion of baseline measurements, families recruited to both phase 1 and phase 2 of the study were randomly allocated to either the intervention group (receive an intelligent personal assistant) or the control group (continue as usual without the provision of additional technology within the home). The IPAP intervention aimed to promote positive health behaviors in the family setting through the utilization of the functions of a smart speaker and its linked intelligent personal assistant. Each family in the intervention arm of the study received a smart speaker (Echo Dot, third generation, Amazon 2018 release) for use in the family home for the duration of the intervention (12 weeks).
The research team set up an individual user account for each family, creating a new email and password, not linked to the family’s other email accounts (for security purposes). Each family was provided with their log-in detail, meaning that the research team and family members could both access the accounts during the intervention period. Each family was provided with a detailed information sheet on how to set up and use the device and were instructed to contact a member of the research team for support or troubleshooting throughout the intervention period.
The research team was able to remotely access the devices and set weekly tasks, prompts, and reminders for family members. The prompts and reminders provided by the research team were developed in line with recommendations for the management of childhood obesity [
Examples of intervention components delivered by the intelligent personal system.
Intervention component and type of interaction | Interaction content | |
|
||
Skill | Ask |
|
Task | Plan your shopping list for the week and add foods to your list using Alexa | |
Tip | Fruit and vegetables that are fresh, frozen, or tinned all count toward your 5-a-day | |
Reminder | How much water have you had today? | |
|
||
Skill | Use Alexa to find some fun games that can help you be active | |
Task | Kids, do 10 star jumps every morning | |
Tip | You should aim to be active daily—try going for a 30-min walk on most days this week | |
Reminder | Have you been for a walk as a family this week? |
In addition, families were informed that the devices were to be used as a health promotion tool within the home setting and were free to add their own reminders at times convenient to them and had complete autonomy over what
Families were informed during the recruitment and throughout the intervention that the research team would also be able to view and manage their user accounts. Families were also made aware that all interactions with the device would be noted by the research team, including interactions that may not be linked to the goals of the intervention, for example, asking the intelligent system nonrelated questions.
Within this pilot feasibility study, we aimed to evaluate the potential of intelligent personal assistants for promoting and maintaining physical activity and other health-related behaviors in both parents and children. Data collection was carried out at local community centers or at the university by trained researchers, and all participant outcome measures were assessed at baseline and follow-up (12 weeks).
Physical activity was measured using an ActiGraph GT3 accelerometer (ActiGraph LLC). Participants (parent and child or children) were instructed to wear the device on the waist for 7 consecutive days, removing it only for bathing, water-based activities such as swimming, and when asleep. During the measurement periods, participants were asked to keep a family log of when they wore the accelerometer and took it off. A sampling epoch of 15 seconds was used for data collection. Periods of ≥60 minutes of zero counts were classified as nonwear and were removed. Cutpoints were used to estimate time spent in sedentary behavior and light-, moderate-, and vigorous-intensity physical activity for adults [
Height (nearest 0.1 cm) and weight (nearest 0.1 kg) were measured according to standardized protocols. BMI was calculated and converted to BMI z-scores using the World Health Organization AnthroPlus software (version 1.0.4).
Behaviors related to eating and activity habits were assessed using the Revised Family Eating and Activity Habits Questionnaire (FEAHQ-R). The FEAHQ-R is a 32 item self-report instrument designed to assess changes in eating and activity habits of family members as well as obesogenic factors in the overall home environment (stimulus and behavior patterns) related to energy balance [
The research team was able to access each family’s account via their log-in details and view each interaction with the device across the intervention period. An interaction was defined as any engagement with the device made by a parent or child, in addition to the reminders and information provided by the device from the research team. A copy of all interactions was downloaded from the device website and anonymously stored. The research team recorded the number of interactions and the type of interaction. Interactions were primarily coded as
A record of any technical issues in relation to the smart speaker was held by the research team. All parents in the intervention arm of phase 1 and phase 2 were invited to participate in focus group discussions. These discussions focused on the acceptability of the intelligent personal assistants, intervention fidelity, any challenges that arose during the intervention, and suggestions for future improvements. Owing to practical issues (timing and location), it was not possible to facilitate focus groups with all parents, so these were replaced with semistructured interviews. One focus group (n=4 parents) and 3 semistructured interviews (n=3 parents) were conducted with participating parents in the intervention arm of the study. All discussions were audiorecorded. The mean duration of the recordings was 26 (SD 20) minutes.
Participants were provided with detailed instructions on the use of the device and the functionality of the device, that is, what the device is capable of doing and picking up. The mute or temporality disable functions of the device were also highlighted to families. These instructions were developed using the manufacturer’s instructions. As these devices were present within the home and accessible to both parents and children, a protocol was developed to consider the potential issue of disclosure and unintended collection of data. No such issues were observed during the intervention period. The search history of the device was kept confidential, and the device was not used for any other purpose during the intervention, for example, recording information or conversations within the home. This pilot feasibility study was approved by the Ulster University Research Ethics Committee and was registered retrospectively (ISRCTN16792534).
Frequencies, percentages, means, and SDs were used to describe data related to recruitment, retention, outcome measures, intervention acceptability, device interactions, and usage. Data analysis was conducted using SPSS for Windows (version 25; SPSS Inc).
Focus groups and semistructured interviews were transcribed verbatim and analyzed thematically, following a deductive approach [
A total of 16 families attending the SWEET project were invited to participate in the IPAP study (
Consolidated Standards of Reporting Trials 2010 flow diagram for phase 1 participants. SWEET: Safe Wellbeing Eating & Exercise Together.
Individual participant characteristics at baseline.
Characteristic | Phase 1 | Phase 2 | ||
Adults (n=11) | Children (n=16) | Adults (n=15) | Children (n=18) | |
Sex, female, n (%) | 10 (91) | 9 (56) | 11 (73) | 8 (44) |
Age (years), mean (SD) | 40.5 (5.4) | 9.1 (2.0) | 38.9 (5.2) | 7.9 (2.0) |
Height (cm), mean (SD) | 166.0 (6.2) | 141.1 (14.5) | 166.9 (8.5) | 130.0 (12.8) |
Weight (kg), mean (SD) | 97.0 (22.8) | 49.5 (15.4) | 81.4 (15.8) | 28.3 (7.7) |
BMI (kg/m2) | 35.0 (6.4) | N/Aa | 29.1 (4.9) | N/A |
BMI, z-score | N/A | 2.61 (1.23) | N/A | 0.02 (1.17) |
aN/A: not applicable.
A total of 20 families from local community groups were approached to take part, of which 16 were assessed for eligibility (
Consolidated Standards of Reporting Trials 2010 flow diagram for phase 2 participants.
In phase 1, 91% (10/11) of adults and 69% (11/16) of children met the minimum inclusion criteria for accelerometer wear time. At baseline, the mean valid wear time was 720 (SD 90.3) and 657.2 (SD 47.8) minutes per day for adults and children, respectively. At follow-up, the proportion of participants meeting the minimum inclusion wear time dropped to 55% (6/11) of adults and 19% (3/16) of children. In phase 2, 86% (13/15) of adults and 89% (16/18) of children met the minimum inclusion criteria for accelerometer wear time. At baseline, mean valid wear time was 782.1 (SD 63.2) and 695.4 (SD 36.3) minutes per day for adults and children, respectively. At follow-up, the proportion of participants meeting the minimum inclusion wear time remained at 87% (13/15) of adults and dropped to 72% (13/18) of children, indicating greater compliance to the accelerometer outcome measure in phase 2 of the IPAP study.
Of those who fulfilled the minimum wear time criteria, 70% (7/10) of adults and 36% (4/11) of children achieved the recommended physical activity guidelines at baseline for phase 1, compared with 77% (10/13) of adults and 38% (6/16) of children in phase 2 of the study. Owing to the small sample size, statistical testing was not undertaken to assess changes in physical activity before and after intervention (
Change in accelerometer measured physical activity and sedentary behavior across the Intelligent Personal Assistant Project study (adults).
Physical activity and sedentary behavior | Intervention, mean (SD) | Control, mean (SD) | ||
|
||||
|
||||
Daily physical activity (minute per day) | 268.5 (35.3) | 234.2 (67.4) | ||
Sedentary behavior (minute per day) | 440.5 (115.5) | 492.8 (52.5) | ||
|
||||
Daily physical activity (minute per day) | 293.7 (57.8) | 201.1 (9.5) | ||
Sedentary behavior (minute per day) | 587.6 (132.8) | 531.4 (26.9) | ||
|
||||
|
||||
Daily physical activity (minute per day) | 260.7 (35.6) | 241.8 (47.7) | ||
Sedentary behavior (minute per day) | 562.3 (10.1) | 492.7 (56.6) | ||
|
||||
Daily physical activity (minute per day) | 218.9 (40.7) | 244.8 (33.1) | ||
Sedentary behavior (minute per day) | 513.9 (65.1) | 498.3 (21.4) |
Questionnaire data were provided by 85% (22/26) of adult participants at all time points. In phase 1, positive improvements in scores for eating style were observed for adults (−1.75, SD 2.06) and children (−0.50, SD 2.81) in the intervention group, with increases observed in the control group. In phase 2, there was a slight improvement in both the activity level score and stimulus exposure and control for children in the intervention group, with all other summary scores increasing across the intervention period (
Change in scores for Family Eating and Activity Habits Questionnaire for adults and children in phase 2.
Characteristics | Adults, mean (SD) | Children, mean (SD) | ||
Intervention (n=6) | Control (n=5) | Intervention (n=7) | Control (n=4) | |
Activity level | 0.75 (2.72) | 1.70 (2.11) | −1.07 (8.23) | −0.25 (6.65) |
Eating style | 1.80 (8.56) | 5.33 (1.15) | 3.33 (6.65) | −1.00 (3.00) |
Eating related to internal cues | 0.83 (1.33) | 0.00 (2.00) | 1.14 (1.46) | −0.13 (1.55) |
Stimulus exposure and control | 1.80 (4.09) | 1.25 (4.99) | −0.25 (6.65) | 0.00 (4.63) |
Across phase 1 of the intervention, families who received a smart speaker on average interacted with the intelligent personal assistant (Alexa) 65 times.
In phase 2, families did not attend the SWEET project, but the intervention content largely reflected the prompts or reminders provided to families in phase 1. Device interactions across phase 2 of the intervention were much higher, with families interacting with the device 312 times across the intervention period (equivalent to 31.3 interactions per week). Only 11% of interactions were coded as relevant (related to diet, physical activity, or well-being). Of the interactions that were coded as relevant, the most frequent interactions were when families asked questions about nutrition (healthy eating) or used
In total, 7 parents took part in focus groups and semistructured interviews to discuss their experiences of the IPAP project. At the offset of these discussions, parents acknowledged the prominent role of technology in their family’s everyday lives and the need to use it in a positive way:
Technology is there, and it can be used for good and evil. And it’s not going to go away. The way they are growing up, they can’t avoid it really so might as well try and use it for good.
...
Parents commented that the intelligent personal system motivated the child to engage with the intervention:
It actually motivated her quite a bit, because she was saying “mummy, we need to go for a family walk now...or I need to eat my fruit or...”
Families found the intervention content acceptable and discussed how the prompts or reminders encouraged them to change their behaviors in a fun way (
Families highlighted several ways to increase engagement with the intervention, including further suggestions on how to use the device within the home and more personalization of the prompts or reminders. The timing of prompts or reminders was a key component of the intervention delivery, and families noted practical issues with this, in addition to the importance of ensuring that families were at home when the device was interacting with them (
If it was connected to your phone, like a phone reminder as well, because Alexa’s in the house.
In addition, families felt that the device needed to be linked to some type of feedback to increase accountability and provide families with opportunities to log their healthy eating or physical activity (
Parents felt that the intelligent personal assistants played an additive role in encouraging children to be healthier and could work alongside other types of intervention:
I still think you need the traditional ways of activity rather than reliance on a device.
...if there was an intervention or like, if there were a, a class or some sort of, erm, programme that was with, sent home with families and Alexa reminded you to do it...
In terms of concerns about having a smart speaker within the home, most parents commented that they were cautious of both increasing engagement with technology and the potential issues with social media and young people (
...he’s downloading games and I don’t know what they are—I would be quite worried; not so much that it’s listening, I wouldn’t worry about Alexa listening, it’s not gonna hear anything in my house.
Supporting quotes from family focus groups and semistructured interviews.
Subheading | Supporting quotes | |
|
||
Device setup | “It was easy to set up and easy to use. Quite interesting but, and the prompts were very good |
|
Prompts or reminders from the research team |
“We got a prompt, quick do 10 sit ups, and I’m like come on children, everyone on the floor, let’s do it! It was some craic [fun] like, and everybody just downed the phones and going to do that challenge. They loved it.” [Family 3, female] |
|
Using other device features |
“...the easy access to the workouts so that you could just do it at a time whenever it suited you “There were a couple of occasions where we asked Alexa for a healthy recipe to make something so we made a chilli one day and we asked Alexa for a recipe ‘cos we were prompted by the device about, you know, healthy, eating healthily and stuff “[Child name] was new-fangled with it, she was more into the music in it, bopping about but it got her active too, she was asking me how to do this, and will you do this “Even her homeworks, she was going out and asking, she was asking me how to spell this, I said ask Alexa, just to get her doing things for herself.” [Family 3, male] |
|
Overall device usage |
“We probably could have utilised it much more but it’s just the pure fact if we had more time. Erm, and the fact that we were away from it all day long and then we came in, in the evening, it’s usually kind of a race, get the dinner made and...” [Family 6, female] “...but after, like, a week or so they kind of almost forgot it was there and maybe that was our fault, we didn’t encourage them to use it as much, erm, but the prompts I think were a good idea.” [Family 5, male] |
|
|
||
Timing of prompts or reminders |
“I think there was a couple of technical glitches where the timing wasn’t right because we didn’t seem to get the prompts and we used Alexa a lot like, we do ask a lot of questions and stuff but, erm, it didn’t seem to prompt us; maybe we were out at the time.” [Family 4, male] “You know, if we weren’t at home..., I don’t know how many prompts there were.” [Family 7, female] |
|
Lack of feedback provided |
“...but what it would say to me, ‘Have you had your five a day?’ Do I shout back, ‘Yes,’ or, ‘Alexa, yeah,’ I don’t know what way to answer...” [Family 7, female] “If you had to log what you did, you know, because it’s fair enough, erm, you could say, ‘Right, go for a family walk,’ but you know, then they come back and say, ‘Well how many kilometres did you do?’ or whatever...to close the loop. |
|
Concerns |
“I just worry about that whole side of technology, erm, never mind Alexa but all social media, erm, in terms of how, how that can be utilised against them and I suppose that’s a worrying thing for me as a parent...” [Family 6, female] “I think if you find the right balance where, you know, I don’t like the idea of my kids being constantly engaged to technology but I can see the benefit of, of that via a prompt or something like that but, you know, I wouldn’t want them to be constantly going to Alexa...” [Family 3, male] |
|
Increasing device usage |
“You know, I think they would maybe be set challenges to do because I think if they’ve, just can get an app and do so much, I’m not sure that they’ll benefit from it “I think if it was maybe a wee bit more personalised...I don’t actually know what I was supposed to be doing with Alexa, you know...and maybe it was in the documentation somewhere, maybe there was a letter written somewhere that I didn’t see, that I didn’t read.” [Family 7, female] |
Most families were able to set up their user accounts and link these to the smart speaker device. Overall, 2 families did not set up their devices in phase 1 of the study. Of these families, one parent noted that they could not set up the device because they shared the house with another family who did not want the device used, and the other family failed to respond to follow-up instructions from the research team, meaning they did not receive the intervention content. All families in phase 2 successfully set up and used the device.
The smart speakers had to be
To our knowledge, this is the first study to outline the development and usage of intelligent personal assistants to promote positive health-related behaviors within the home setting. Given the constraints that exist within current family-based interventions, including time and travel restraints [
To date, there is a paucity of research on both the development of interventions using this technology and the potential effectiveness of such interventions. An ongoing study is examining the role of a voice coach intervention (Amazon Alexa/Echo) on increasing levels of physical activity among overweight and obese cancer survivors [
Recent research has highlighted the plethora of
The mean frequency of device interactions across phase 2 was much greater (312 vs 65), but a higher proportion of interactions were coded as
Given the small sample size in this study, it was not possible to statistically compare the effectiveness of these 2 intervention approaches. As the families in phase 1 were already attending the SWEET project, the results from phase 1 and phase 2 could not be combined. A recent systematic review highlighted that most family-based eHealth interventions combined technology with other types of delivery, for example, face-to-face counseling, nutrition lessons, and so on, and from this literature, it is difficult to ascertain the exact effect of the eHealth component versus other approaches [
The development and feasibility testing of the intervention identified several important methodological considerations. First, the research team was not able to control the content, or indeed validity, of the responses families received when they asked for information on healthy eating or physical activity. At present, there is limited insight on whether these apps are developed based on evidence-based guidelines or available materials [
The implementation of the intervention was dependent on a few factors. An important practical consideration was the capacity of the research team to access the family’s device remotely. If the device was switched off or the family had Wi-Fi connection issues, the delivery of the intervention was affected, as the research team was unable to set new reminders and prompts during these periods. During the focus group and interview discussions, parents highlighted how the timing of the prompts or reminders may have affected their adherence to the intervention. Although attempts were made to tailor the intervention to suit the schedules of individual families, future studies using similar intervention components should seek to provide families with further guidance and ownership in relation to managing the devices themselves.
The IPAP study adopted a cross-sectoral, interdisciplinary approach to explore the role of intelligent personal assistants within the home environment to promote and maintain physical activity and other health-related behaviors in families. The intervention development and evaluation used novel methods to capture intervention engagement, addressing key recommendations for research in this field to adopt appropriate methodologies that enable interventions to be effectively evaluated [
This study demonstrates the feasibility and acceptability of a family-based intervention using intelligent personal assistants. This novel intervention has highlighted important methodological considerations and provides important suggestions to further optimize the potential of intelligent personal assistants to promote positive health-related behaviors in the home setting. This work will inform future pilot and fully powered studies to build upon this feasibility work and test whether such interventions are effective at changing health-related behaviors, including physical activity and healthy eating.
CONSORT EHEALTH checklist V1.6.1.
Revised Family Eating and Activity Habits Questionnaire
Intelligent Personal Assistant Project
randomized controlled trial
Safe Wellbeing Eating & Exercise Together
The authors thank the families that participated in the study. The authors are grateful to the SWEET project team (George McGowen, Julie White, and Aoibhin Kelly) for their involvement in the study and for their assistance with the project management. The authors thank Jordan Fleming, Naomi Bell, and Sofia Marini for their assistance with data entry. This project was funded by the GetAMoveOnNetwork+ (Engineering and Physical Sciences Research Council grant EP/NO27299/1). The funder had no role in the study design, data collection and analysis, decision to publish, or preparation of the paper.
None declared.