This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR mhealth and uhealth, is properly cited. The complete bibliographic information, a link to the original publication on http://mhealth.jmir.org/, as well as this copyright and license information must be included.
Strong evidence shows that an increase in cardiorespiratory fitness (CRF) and physical activity (PA) reduces cardiovascular disease risk.
To test whether a scientifically endorsed program to increase CRF and PA, implemented on an easy-to-use, always-accessible mobile app would be effective in improving CRF.
Of 63 healthy volunteers participating, 18 tested the user interface of the Cardio-Fitness App (CF-App); and 45 underwent a 2-week intervention period, of whom 33 eventually concluded it. These were assigned into three groups. The Step-based App (Step-App) group (n=8), followed 10,000 steps/day prescription, the CF-App group (n=13), and the Supervised Cardio-Fitness (Super-CF) group (n=12), both followed a heart rate (HR)-based program according to American College of Sports Medicine (ACSM) guidelines, but either implemented on the app, or at the gym, respectively. Participants were tested for CRF, PA, resting systolic and diastolic blood pressures (SBP, DBP), resting, exercise, and recovery HR.
CRF increased in all groups (+4.9%;
A 10,000 steps/day target-based app improved CRF similar to an ACSM guideline-based program whether it was implemented on a mobile app or in supervised gym sessions.
Low cardiorespiratory fitness (CRF) and physical activity (PA) have been shown to be two key independent risk factors for cardiovascular disease (CVD) and all-cause mortality [
Ten thousand steps a day is a well-accepted PA goal, which has shown benefits in improving people’s cardiovascular health [
Next, to the issue of what is the most effective training program to reduce CVDs risk, there is the problem of getting people to adhere to such a program. Low adherence to PA and exercise programs has been a known issue for decades [
Recently, smart, wearable, and mobile technology has enabled a large number of health applications [
Knight et al [
Therefore, the primary aim of this study was to test whether a scientifically endorsed program to increase CRF implemented in an easy to use, and always-accessible mobile app, would be effective in improving CRF. A secondary objective of this study was to investigate how a HR-based training would compare with a steps-based training in terms of CRF changes.
In order to test our research hypotheses a three group, pre- and posttest, 2-week intervention study was designed. The study protocol was approved by the Internal Committee of Biomedical Experiments of Philips Research as well as the Departmental Ethics Committee of the Milan University according to the Declaration of Helsinki.
Sixty-three participants were included in this study. Eighteen took part in a user experience test of the app. Forty-five participants took part in the actual experimental intervention protocol. Those participants were recruited via posters and at the local health center. After first interest in the study, volunteers were informed about the study protocol via information letters, where inclusion and exclusion criteria were already mentioned. These criteria specified, among others, the absence of chronic health conditions, no cognitive impairments, a low cardiorespiratory fitness (<45 mL/kg/min), an age ranged between 20 and 55 years, and a body mass index (BMI) limit not exceeding 35 kg/m2. After signing the informed consent participants were screened for cardiovascular risk using the American Heart Association/ACSM Health/Fitness Facility Preparticipation Screening Questionnaire [
The latter 45 participants were assigned into three groups: the Step Count App group (Step-App) (n=16), the Cardio Fitness App group (CF-App) (n=17), and the Supervised Cardio Fitness group (Super-CF) (n=12). For logistic reasons only two groups, the Step-App group and the CF-App group, could be randomized. This is because the participants included in the Super-CF group, which was included as a training intervention quality check, had to be living in the vicinity of the fitness center used for the supervised exercise intervention. This has resulted in an inhomogeneity at baseline between this group and the other two. Participant characteristics at baseline are reported in
Participant characteristics.
Step-Appa group (n=8) | Super-CFb group (n=12) | CF-Appc group(n=13) | |
Mean ± standard deviation | Mean ± standard deviation | Mean ± standard deviation | |
Male/female | 3/5 | 5/7 | 5/8 |
Age, years | 40 ± 10 | 45 ± 3 | 42 ± 6 |
Height, m | 1.69 ± 0.03 | 1.69 ± 0.12 | 1.71 ± 0.09 |
Weight, kg | 68.20 ± 11.80 | 78.10 ± 19.10 | 81.60 ± 10.10 |
BMI, kg/m2 | 23.70 ± 3.53 | 27 ± 4.70 | 26.60 ± 1.43 |
SBPd, mm Hg | 124 ± 11.60 | 132 ± 11.14 | 129 ± 16.80 |
DBPe, mm Hg | 76.90 ± 8.06f | 87.40 ± 5.68f | 83.70 ± 9.60 |
N-Exg questionnaire | 3.50 ± 2.27 | 0.58 ± 0.79 | 3.08 ± 2.27 |
Self-efficacy questionnaireh | 57.70 ± 3.31 | 44.20 ± 9.60 | 61.60 ± 13.98 |
Estimated VO2 max, mL/kg/mini | 36.8 ± 4.90j | 26.9 ± 5.60j | 31.70 ± 6.40 |
aStep count app group.
bSupervised cardio fitness group.
cCardio fitness app group.
dSystolic blood pressure.
eDiastolic blood pressure.
fSignificant differences at the baseline between Step-App and Super-CF groups are
gNonexercise aerobic capacity questionnaire (where 0 is inactive and 7 is very active).
hThe self-efficacy questionnaire is on a 0 to 100 scale.
iSelf-paced treadmill walk test was used to estimate VO2 max.
jSignificant differences at the baseline between Step-App and Super-CF groups are
Study enrollment flow-chart. Step-App, 10,000 steps/day training plan provided by a mobile app; CF-App, ACSM guidelines-based cardio-fitness training plan provided by a mobile app; Super-CF, ACSM guidelines-based cardio-fitness training plan provided by a personal trainer.
Participants were asked to visit our laboratories on three different occasions; for the baseline tests, the pretests, which occurred 1 week after the baseline tests, and the posttests, which took place immediately after the 2-week intervention. The design of the study is depicted in
Two submaximal exercise tests were then conducted. The Ruffier-Dickson squat test [
RDI=(P1-70)+2(P2-P0)/10 (1)
where P0 is 15-seconds mean resting HR, P1 is the maximum HR recorded during the first 15 seconds of recovery, and P2 is the 15-seconds mean after the first minute of recovery (the period from 1 minute and 00 seconds to 1 minute and 15 seconds) [
The Ebbeling test consisted of a 4-minute walking session at an adequate speed so that the participant’s HR would be between 50% and 75% of the estimated max HR (220 minus Age), followed by a 4 to 5 minutes session at a 5% incline at the same speed [
VO2 max=15.1+(21.8 · Speed)-(0.327 · HR)-(0.263 · Speed · Age)+(0.00504 · HR · Age)+(5.98 · Gender) (2)
where treadmill speed was in miles per hour, HR in beats per minute, age in years, and gender was 0 for females and 1 for males. Except for the screening questionnaires and the height measurement, all other physical tests performed at baseline, were being repeated at pretest and after the 2-week intervention.
Study protocol outline.
Throughout the entire study period participants were asked to wear two HR monitors, a chest strap–based one, the same as mentioned above, and a wrist mounted optical sensor, validated by Valenti and Westerterp [
During the pretest visit, participants received instructions about the training that they would follow during the two intervention weeks. The Step-App group was asked to complete 10,000 steps per day. Participants could access feedback on their progress via the device and via a standard mobile app, only during week 1 and 2, and not during the control week. No specific instructions were given to this group on how to achieve their goal. The pedometer has a display were steps could be read upon request, and it can be synchronized via Bluetooth connection directly with its dedicated mobile app. The Fitbit app did not provide any strategy on how to achieve the 10,000 steps/day target neither gave reminders.
The CF-App and Super-CF groups were asked to follow an intensity training based on the guidelines of the ACSM [
The training programs for the CF-App and the Super-CF groups were designed according to the recommended frequency, intensity, time, and type framework outlined in Table 7.4 of the ACSM’s guidelines for exercise testing prescription consistent with the United States Department of Health and Human Services Physical Activity Guidelines for Americans [
For the CF-App group, the daily training target was visualized based on the concept of the Training Impulse method by Banister and Calvert [
This training monitoring metric was called “mBeats” and consisted of the number of heart beats in a personalized heart rate zone. This mBeats score was calculated over the week as target HR (bpm) × session duration (minutes) × the training frequency. Target HR was defined according to Box 7.2 of the ACSM guidelines [
Target HR=[(HRmax–HRrest)] · %intensity desired]+HRrest (3)
where the desired intensity is determined according to the baseline fitness level, corrected for age and gender. For instance participants with a target HR=120 bpm (eg, 30% of HRR) and classified as sedentary would receive a training frequency of three times a week, and a training session duration of 30 minutes, making a weekly mBeats target equal to target HR × session duration × frequency per week of 10,800 mBeats. The week was divided into training days and resting days. In this example, a training day would have a daily target of 3600 mBeats, while a resting day would have a target of 0 mBeats. The daily targets are not fixed, but depend on the remaining mBeats for the week. In other words, if the weekly target mBeats was 10,800 and on the first training day 7000 mBeats were already achieved, the daily target on the second training day would have been 1900 mBeats (the remaining 3800 mBeats divided by 2 remaining training days of the week).
The participants in the Super-CF group were instructed to visit the gym three to four times per week according to their training program consistent with the ACSM guidelines described above. During the training session at the fitness center participants were supervised and motivated by an experienced personal trainer. They were also asked to wear the HR monitors and step counters for the whole day during the entire intervention period without a specific goal.
Multiple screenshots of the Cardio Fitness mobile app used in this study.
A mobile app was specifically designed for the purpose of this study. Main features included in this first prototype of the CF-App were a HR feedback element, daily and weekly mBeats targets, an activity planner based on the theory of implementation intentions [
Our App included a planning option, in which the participants could decide how to distribute their training days according to their own private schedule (
User interaction experience was tested during a 3-week pilot test by 18 healthy adults (age: 26-50 years, BMI: 18-25 kg/m2). The test started with a baseline week followed by two intervention weeks. During the baseline week, participants were instructed to wear the HR monitor and keep the app running, in order to collect physical activity data. All functionalities of the app were disabled and participants were asked to be as physically active as usual. After the baseline week, the 2-week intervention period started in which participants were coached to achieve their daily and weekly mBeats targets. At the end of the study, a one-on-one, semistructured interview was conducted to discuss usability and user experiences in depth. In addition, usability was measured with the Computer System Usability Questionnaire (CSUQ) [
The Statistical analysis was performed using the Statistical Package for Social Sciences (version 21) and the level of statistical significance was set at 0.05. Data were presented as means ± standard deviations unless otherwise indicated. Dependent variables were analyzed with a two-way, repeated, measures-mixed analysis of the variance (ANOVA), where the two factors were: time (pre- and post-intervention) and group (Step-App, CF-App, Super-CF). Because only the Step-App and the CF-App groups were randomized, a two-way, repeated measures ANOVA was performed also only on those two groups. If there had been violations of the sphericity, the corrections of Green-house Geisser if ɛ<0.75 and Huynh-Feldt if ɛ>0.75 were applied. The significant interactions were followed by post-hoc Tukey test. Pearson correlation coefficients were calculated between RDI and estimated VO2 max.
Overall, participants were very positive about the potential of the CF-App and the concept of collecting heart beats to increase PA. Nonetheless, the pilot test revealed some points for improvement for the app. Most participants stated that their target zone was too narrow, and for some participants the zone was too high. Based on this feedback, the target HR zone was widened in the intervention study. Moreover, more features and functionality could be added to make the CF-App useful for people with various health goals. The current app is specifically designed to improve CRF, but over half of the participants stated they would want more information about the effects of all activities, including activities that are in a different HR zone. Furthermore, they wanted to be able to set their own exercise goals, to, for example, maintaining health or losing weight.
Other than many other popular fitness apps, the CF-App stimulates users to achieve a weekly exercise target, rather than daily targets. The majority of participants (15/18, 83%) were in favor of a weekly target, as opposed to a daily target, because the amount of exercise they do is not the same for every day, but is fairly similar every week. Five of them stated that although they preferred a weekly target, they would be okay with having a daily target, as long as the daily target is integrated with what they have planned in their activity planner. Only three participants were in favor of the daily target because it would motivate them to get enough exercise on a daily basis. Overall usability from the CSUQ was toward the positive end of the scale. On a scale from 1 to 7, the participants rated usability 4.46 ± 1.46, quality of the interface 4.96 ± 1.26, system usefulness 4.72 ± 1.34, and information quality 4.21 ± 1.28.
The ratings on the CSUQ questionnaire were in line with the qualitative findings obtained during the interviews. Ratings on the three variables were on the positive side of the scale, confirming the potential of the app. However, with an average rating of 4.5 on a 7-point scale, usability is not rated extremely high. This can mostly be accounted to some technical issues and the limited functionality of our first prototype. Moreover, participants identified missing elements that would make the app better, more interesting, and easier to interpret. Half of the participants explicitly stated they would add more parameters, such as speed and distance. This would make it easier for them to relate the mBeats to something they are already familiar with. Ten participants would like to add global positioning system tracking, or at least link the timestamps in the history view to a location, because it could give them insight in the effects of certain routes on their exercise performance.
A majority of the participants mentioned that they would like advisory or motivational messages in the app. These could be tips on what kind of activities they could do to reach their target (8 participants), or information on how they are doing during exercising (6 participants). In addition, six participants mentioned they want to get messages about their achievements so far, like how many beats they still need to achieve, and another six participants wanted motivational messages such as ‘good job’ or ‘you have been idle for a while, isn’t it getting time to go for a jog?’
Some participants did not fully understand the mBeats concept. They would need more education on advantages of the concept and measuring heart rate. Furthermore, they would need support in interpreting their data, to gain better understanding of their performance.
Because people were uncertain about their current level of fitness and how to improve or maintain it, they would want the app to objectively assess their current fitness level and use this information to create a personalized training program, with feasible goals and reliable, and accurate HR zones.
There were no significant time × group interactions (F(1;30)=0.466,
Baseline (W0) and 2 weeks (WL2) systolic and diastolic blood pressure changes induced by the interventions. Step-APP, 10,000 steps/day training plan provided by a smartphone application; CF-App, ACSM guidelines-based cardio-fitness training plan provided by a smartphone application; Super-CF, ACSM guidelines-based cardio-fitness training plan provided by a personal trainer. a, Significant main effect of time. b, Significant main effect of group. c, Significant difference between Step-App group and Super-CF group.
Maximal oxygen uptake estimated by the Ebbeling treadmill walk test did not show any significant time × group interaction (F(1;30)=0.543,
A significant time × group interaction was found in week mean step counts (F(2;60)=4.903,
HR rest did not show a time × group interaction (F(1;30)=2.169,
HR recovery after 1 and 3 minutes did not show significant time × group interactions (F(1;30)=1.368,
In order to confirm the validity of the RDI as CRF index, we correlated it with VO2 max estimated using baseline values. A negative significant correlation was present between RDI and the estimated VO2 max (r=-0.46,
When only Step-App and CF-App were compared, only two significant main effects of time were found. One for HR after 1 (F(1;19)=4.619,
A) VO2 max week 2 – baseline deltas. B) Mean weekly steps showed for the baseline week 0 (W0), and the two intervention weeks, week 1 (W1), and week 2 (W2). C) Mean weekly mBeats expressed as percentage of target mBeats (for the definition of mBeats see methods section). D) Week 2 - baseline Heart rate (HR) deltas at rest, for the maximal recoded during a squat exercise test (peak), 1 and 3 minutes during the recovery from the squat exercise test, and for the Ruffier-Dickson Index as defined in the method section. a, Significant time x group interaction. b, Significant main effect of time. c, Significant main effect of group. d, Significant difference between Step-App group and Super-CF group. e, Significant difference between CF-App and Super-CF group.
Our research showed that a scientifically endorsed program to increase CRF, in line with ACSM's guidelines, implemented on smartphone in an easy to use and always-accessible app can improve fitness, and other health-related parameters. According to the treadmill walk test, CRF increased in all three groups not showing any interaction between the groups. Although, the weekly mean number of steps walked by the CF-App group did not increase drastically from baseline levels (6808 steps/day at baseline; 7534 steps/day week 1; and 7775 steps/day week 2), and it may seem that participants in the CF-App group were more efficient than those in the other two groups in improving their CRF levels; yet no interaction and no main effect of group were found between the Step-App and the CF-App groups indicating that steps/day did not differ that much among the two App groups. The mBeats for the CF-App group did increase significantly from 43.5% of the mBeats target at baseline, 113.7% at week 1 and 105.5% at week 2, whereas mBeats in the Step-App group did not increase (ie, 60% of the hypothetical target mBeats). Although the Super-CF group achieved a higher mBeats level (ie, approximately 170% of the target mBeats), the participants of this group have done that by walking the highest number of steps per day. Most probably because these participants trained at the gym under the supervision of a personal trainer, three to four times a week. These results seem to suggest that intensity as well as volume training delivered by means of an easy to use mobile app accessible at any time, may be an efficient alternative to attending fitness classes. Although programs targeting steps can be a useful tool for sedentary people with a very low CRF level [
Resting blood pressure results showed a small but significant decrease in SBP in the two intensity groups (CF-App and Super-CF), but not in the volume group (Step-App). DBP decreased in all three groups. Previous studies have found that both volume and intensity training are able to reduce blood pressure in hypertensive people [
Resting as well as recovery HR did decrease, as expected [
Our study had a number of limitations. Cardiorespiratory fitness was only indirectly estimated by using a treadmill submaximal test and a squat test [
Our CF-App was built taking into consideration the user experience feedback. As shown in the Methods section an additional qualitative study was conducted on 18 people to improve the look and feel and the experience flow of our software app. Qualitative reports confirmed the high relevance and acceptance of our app. However, it could be improved by providing educational, interpretational, and motivational messages. Ultimately, it is important for people who have a low PA level and a low CRF to start doing something to improve their lifestyle behavior. Adhering one to one to guidelines and recommendations can be overwhelming for most people [
Another important aspect of our research was to encourage people to take any occasion throughout the day to engage in moderate to vigorous PA. For the less fit people this meant taking the stairs more often, cycling to work at a faster pace, or brisk walking during the lunch break instead of having a stroll. By no doubts motivation is key, and people with low readiness to change and self-efficacy will still struggle to adopt an active lifestyle. In the current study, given 100 as the maximal motivation to exercise, we had moderate levels in self-efficacy in all three groups, averaging approximately 54.3 ± 9.1. Yet all groups showed good short-term adherence throughout the two intervention weeks, still longer-term adherence, which is the hardest to achieve, remains to be investigated.
For logistics reasons, participants were not randomly assigned to the Super-CF group. This has resulted in a baseline difference in CRF. However, this group was mainly used as a quality check of the intervention, to control for the main effect of time. This study, as mentioned above, was also kept rather short, only 2 weeks of intervention. We have shown in the past that sedentary people can improve their CRF when undergoing a vigorous but short training program [
In conclusion, a 10,000 steps/day target-based app improved CRF similar to an ACSM guidelines-based program whether it was implemented on a mobile app or in supervised gym sessions. Moreover, HR-based training improved CRF in equal measure as a steps-based training, but with a higher number of heart beats in a training zone for a similar number of steps/day than a Step-based training.
American College of Sports Medicine
analysis of the variance
body mass index
cardio-fitness app
cardiorespiratory fitness
computer system usability questionnaire
cardiovascular disease
diastolic blood pressures
heart rate
heart rate reserve
physical activity
Ruffier-Dickson index
step-based app
Supervised Cardio-Fitness
systolic blood pressures
The authors would like to thank Dr. Jelle Alten for developing and contributing to the design of the mobile app and Ir. Tess Speelpenning for her user interface guidance for the mobile app.
Francesco Sartor, Alberto G. Bonomi, and Saskia van Dantzig work for Royal Philips Electronics.