記住我
Osteoarthritis is one of the most common chronic joint conditions and leads to pain, stiffness, and reduced physical function, which in turn diminishes quality of life (QoL) [,]. The most commonly affected weight-bearing joints are the knees, followed by the hips []. In 2019, about 344 million people worldwide were living with an osteoarthritis severity level that could benefit from effective therapy, and this number will grow over the next few years []. Appropriate mechanical stimuli, including physical activity (PA), are joint health protecting factors []. Furthermore, an adequate level of PA and reduced sedentary behavior (SB) results in reduced pain and improves physical function and stiffness in patients with knee osteoarthritis (KOA) [-]. Thereby, the QoL of people that are affected can also be improved []. Treating pain and overcoming functional limitations are essential for preventing a downward spiral of reduced PA and declining QoL []. Inadequate PA can lead to excess bodyweight, which increases mechanical joint loading, for example, during locomotion, and accelerates the degenerative process of KOA []. Furthermore, individuals with KOA are more prone to cardiovascular disease, as factors, such as reduced PA, are also correlated with elevated cardiovascular disease risk []. Consequently, PA and exercise are recommended for preventing such disease []. Additionally, PA and exercise have been shown to be beneficial in terms of alleviating KOA symptoms such as pain and stiffness []. Previous research has highlighted the importance of new strategies to encourage patients with KOA to participate in greater levels of PA [,]. However, KOA is a highly heterogeneous and multifactorial disease that affects individuals differently with variations in disease progression, severity, and response to treatments []. Consequently, an individualized therapy approach is suggested to effectively manage the symptoms, slow down the progression of the disease, and promote a healthy lifestyle specifically tailored to each person []. While this is also possible for interventions that are delivered in person, mobile health (mHealth) interventions are upcoming methods with the opportunity to deliver such highly individualized interventions at a larger scale. mHealth interventions are defined as medical and public health practices supported by mobile devices, such as mobile phones, patient monitoring devices, PDAs, and other wireless devices []. These interventions profit from the fast development of accompanying sensors such as accelerometers and heart rate monitors [], which allow for ecologically valid real-time interventions. Previous studies show promising results for PA promotion and the reduction of SB if a theoretical foundation and behavior change techniques (BCTs) were used for the intervention []. In the context of KOA, reviews point to good usability and the potential for effective interventions, but the research is in an early stage [,]. A review including mHealth interventions for KOA points out that most available apps lack scientific studies backing their effectiveness, but those evaluated by RCTs show promising results concerning pain, physical function, and PA []. The main features of these apps included exercise prescription and the tracking of symptoms. A special case of mHealth interventions of which patients with KOA could greatly benefit are the highly individualized just-in-time adaptive interventions (JITAIs) or ecological momentary interventions (from here on only the term JITAI is used). These aim to generate real-time or near–real-time feedback by leveraging prior knowledge of individuals and previously collected data. With this compiled, dynamic knowledge, tailored interventions and nudges can be delivered when needed through mobile technologies, aligned with the individual’s context and requirements []. These nudges can then change proximal outcomes such as step count to achieve the ultimate distal goals of the JITAI, such as better joint health and QoL []. An intervention is defined as a JITAI if it corresponds to real-time needs, adapts to input data, and is triggered by the system with the aim to deliver the intervention at a state of vulnerability, opportunity and receptivity of participants [-]. In the context of designing JITAIs, decision points represent the specific points in time when the JITAI can be triggered, while decision rules are the criteria that are applied at a decision point to determine if the JITAI should be triggered and which intervention option should be used. Intervention options refer to the possible actions that the JITAI can take at a decision point, while tailoring variables denote the sensor or user input that is used for adaptation []. A previous review on JITAIs for PA promotion by Hardeman et al [] and a meta-analysis for device-based measured health outcomes by Xu and Smit [] point to the feasibility and preliminary effectivity of JITAIs to promote PA and reduce SB. However, further variables (ie, physical function, stiffness, pain, and QoL) are important to KOA [-] that were not considered in previous works. Additionally, previous reviews did not aim to clearly depict decision points, tailoring variables, intervention options, and decision rules from the literature. This, however, is important to build a foundation for future JITAI designs. Finally, JITAIs are an upcoming field, with new studies getting published frequently warranting frequent updates of the literature.
ObjectiveThis review aimed to systematically summarize the literature regarding the features of JITAIs of important physiological health outcomes and proxies (ie, PA, SB, physical function, QoL, pain, and stiffness). The second aim was to extract the most promising features that should be considered by future JITAIs in the treatment of KOA.
This systematic review was registered on the Open Science Framework [] on March 7, 2022. It was carried out based on the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) expanded checklist [].
Inclusion CriteriaInclusion and exclusion criteria were formulated according to PICOS (Population, Intervention, Comparison, Outcomes, Study) and are presented in .
Textbox 1. Inclusion and exclusion criteria.Population: Participants in the studies must be aged >18 years, regardless of sex and gender. Thus, the only exclusion criterion relates to studies that only included participants aged <18 years of age. Studies were included if <50% of participants were aged >18 years.Intervention: All studies that use a digital physiological treatment that can be defined as a JITAI are included in this review. If studies use other types of interventions in addition to JITAIs, they are still included, and only the JITAI parts were considered. Furthermore, there were no restrictions regarding the duration of the intervention, the number of intervention sessions, or the technical device for the intervention.Comparator: There was no restriction regarding control. We included active control (eg, other physiological treatments or interventions), passive control (no intervention or treatment), and studies with no control group (CG).Outcome: Studies were included whenever they evaluated the effectiveness of a JITAI regarding one of the parameters PA, SB, pain, QoL, physical function, or stiffness as a main outcome. This also included all parameters that are synonyms or components of the previously mentioned parameters.Study design: We included all study types, except reviews or meta-analysis. The only important thing was that the study reported on the effectiveness of the intervention.Search StrategyThe electronic databases PubMed, Web of Science, Scopus, and EBSCO were used to identify relevant studies. Because search strategies exclusively relying on databases have been shown to be nonexhaustive [], the reference list of all selected articles was checked for further relevant studies. Initial searches were conducted on October 26, 2022, and the search was rerun on April 9, 2024. We combined different search terms related to the intervention type, the area of mHealth, and the outcome parameters. Further details are available in .
Paper SelectionAll identified studies were exported to a reference management tool (EndNote [Clarivate] or Citavi [Swiss Academic Software GmbH]). The reference management tools were used to remove all duplicates. Then, all titles were checked for eligibility by MRB and JF independently. Second, the abstracts were screened, and third, the full texts of all remaining papers were screened, and the reasons for exclusion were noted by the 2 authors. Any disagreements at one of the screening steps were resolved by discussion until consensus was found.
Study Quality AssessmentThe Quality Assessment Tool Effective Public Health Practice Project was used for the evaluation of the study quality. It is a specifically designed tool to test and support findings in public health research and refers to articles related to a broad number of health-related topics, including chronic diseases []. By evaluating 6 different criteria, a global rating was assessed at the end for each study. Every criterion could be rated weak (red), moderate (yellow), or strong (green), which finally influences the global rating. One weak rating leads to a moderate global rating, and 2 or more weak ratings result in a weak global rating. Only studies without a weak rating can achieve a strong global rating [].
Data ExtractionAll important information of the selected studies has been reproduced in tabular form in the Results section. The preparation of the tables and their contents were developed by MRB and JF in consultation with BS. For the most accurate rendering of the study content, reference was made to the study characteristics, including study design, sample size, population, and setting, as well as characteristics of the participants. To provide starting points for future JITAIs, the features and delivery of the JITAIs, including decision points, tailoring variables, intervention options, and decision rules based on the JITAI framework [], and the duration of the interventions were also extracted from the existing studies. To display effectiveness and theoretical foundation, all data regarding retention, measurement, and the significant within- or between-group differences of the intervention and the use of BCTs stratified by control and intervention group (IG) according to Michie et al [], and the theoretical foundation of the interventions are displayed in the Results section.
A total of 4340 papers were exported to Endnote (MRB) and Citavi (JF) and screened (). Initially, 571 (13.16%) duplicates were excluded by the reference management tools, and another 279 (6.43%) duplicates were removed manually. The remaining 3490 (80.41%) papers were screened for their titles, and 3133 (72.19%) were excluded by the authors. Thus, 357 (8.23%) abstracts were read, and 213 (4.91%) more papers were excluded at this step. The updated search yielded 1215 (28%) additional papers, including 145 (3.34%) duplicates. The 1070 (24.65%) unique papers were screened for title and abstract, and the remaining 88 (2.03%) papers for full texts. The main reasons for exclusion were that it was no intervention study, the intervention did not qualify as a JITAI, the outcome did not fit the inclusion criteria, it was a systematic review or meta-analysis, or all participants were <18 years old. Full texts of the remaining 144 (3.32%) studies were screened, and finally 18 (0.41%) papers were included in the review, supplemented by additional 14 (0.32%) papers from the updated search. During the screening process, 13 more studies were added. Nine studies were adopted from the systematic review by Hardeman et al [], and 4 additional papers were found during data extraction by checking references of the included studies. Finally, 45 studies are presented and summarized in and [-] and 3 [-].
Figure 1. PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram. JITAI: just-in-time adaptive intervention. Table 1. Study characteristics.Author and country of studyStudy designPopulation and activity profileAnalyzed sample sizeParticipant characteristics (age, gender, and ethnicity)Allicock et al [], United StatesFeasibility study or pilot evaluationaPA: physical activity.
bIG: intervention group.
cCG: control group.
dCOPD: chronic obstructive pulmonary disease.
eSCI: individuals with spinal cord injury.
fCVD: cardiovascular disease.
gMVPA: moderate to vigorous physical activity.
hWHO: World Health Organization.
iPAUL: The Playful Active Urban Living app.
Description of the Included StudiesA total of 25 (56%) studies were conducted in the United States, 9 (20%) in the Netherlands, 2 (4%) in Australia, 2 (4%) in Germany, and the others in New Zealand, China, Qatar, Finland, Israel, Spain, and Belgium. Most of them (26/45, 58%) were designed as feasibility study or pilot evaluations, and they were mainly randomized controlled trials (23/45, 51%). In total, 9 of 45 (20%) were conducted in a single-group pre-post design, 3 (7%) as a secondary analysis using a within-person design for the IG, 2 as a controlled within-person trial, 2 (4%) as a mixed-method evaluation, and all other studies were either a retrospective cohort study, a randomized within-person study, or combined different study designs. The minimum number of participants was 8 (3 studies), and the maximum was 981 participants with an average of 99 (SD 175) participants. Two studies focused exclusively on women, and overall, 62% (2434/3960) of all participants were women and 38% (1516/3950) were men. The mean age in 19 studies was >50 years, and they focused mainly on a specific population such as sedentary people, patients with chronic diseases (eg, chronic obstructive pulmonary disease and coronary heart disease), or survivors of cancer. Retention in 3 studies was <50%, 4 studies were below 80%, 21 were ≥80% and <100%, and the remaining 12 studies had full retention. PA was the most common outcome (39 studies), followed by 19 studies that used parameters of SB as an outcome, of which 18 studies considered both PA and SB. In total, 9 studies examined the impact on QoL, 4 used synonyms of physical function, and 3 studies used pain as an outcome. None of the included articles referred to stiffness. Detailed results are displayed in . An average of 8 BCTs (SD 7) were included for the IG and 4 (SD 5) for the CG. The number of BCTs in the IG ranged from 2 to 47, compared with 1 to 27 in the CG. The intervention duration ranged from 1 day to a maximum of 12 months, with a typical duration of 3 to 6 weeks. One study had no time restriction but a step count of 166.000 as a cutoff for the intervention duration []. Details on BCTs and theoretical foundations can be found in [-].
Key Features of JITAIsDetailed key features of the included JITAIs are displayed in .
Most decision points at which a JITAI could potentially be triggered were set to short time intervals, which were either specified to one decision point each minute (n=11) or not clearly specified but indicating short time intervals (n=21). Furthermore, 5 studies set decision points to time frames of 10 minutes up to 2 hours, and 11 were set to multiple times per day up to daily.
Tailoring variables used for JITAI adaptation were mainly device-based measured by accelerometer, smartphone, or smartwatch and concerned PA and SB outcomes supplemented with variables such as time, weather, or location. A total of 10 studies used self-reported tailoring variables through a baseline questionnaire [], through ecological momentary assessment (EMA) [,,,], or through nonspecified methods to assess pain and fatigue [,], to enable nonspecified user input [], or to supplement PA data if the device has not been worn [-].
Intervention options at decision points included mainly the delivery of an audible or vibration prompt to participants accompanied by tailored feedback, information about health behavior, and suggestions for behavioral alternatives. One study explicitly explained in the prompt why the prompt occurred []. The alternatives based on the decision rules were mainly not to send the prompt for the decision points with short time frames, while those with longer timeframes had different intervention options depending on the tailoring variables.
Decision rules defining opportune moments were mainly based on if-then conditions. This means that an opportune moment was defined if a certain threshold —for example, 20 minutes of SB [], <100 steps in 1 hour [], a bout (≥3 min) of moderate to vigorous physical activity (or higher) PA was performed [], lumbar posture [,], time, location, or weather-based thresholds [] were met. Most decision rules were based on one if-then condition, while some used if-then decision trees. More complex models applied multiarmed bandit algorithms (sequential decision-making algorithm []) [-], pareto-frontier algorithm (taking user preferences into account []) [], prelearned reinforcement learning models (models that make use of historical data and of data during the trial []) [], k-nearest neighbor classifier (one of the most fundamental and simple classification methods []), and real-time prediction and self-learning by a support vector machine (a classical machine learning technique []) [,]. If the conditions for an opportune moment were met, intervention options were delivered, which included both fixed responses and microrandomized suggestions (eg, Klasnja et al []) for behavior change.
EffectivenessDetailed results for intervention effectiveness are displayed in . For RCTs, 5 studies indicated effectiveness compared with a CG for PA [,,,,], 3 studies for SB [,,], and 1 study for pain []. For pre-post–designed and within-person studies, 8 studies indicated effectiveness in most outcome variables of the study for PA [,,,,,,,], 5 for SB [,,,,], 1 study for pain [], and 1 study for QoL []. Regarding the opportune-moment identification, decision points, and rules, most interventions that indicated effectiveness had near real-time decision points on a minute-to-minute basis, and simple decision rules depicting opportune moments when a threshold of 20 to 120 minutes of inactivity, SB, or computer activity was reached. Of these, the studies of Rabbi et al [,] were the only ones that used more complex algorithms. Effectiveness was also found for longer decision point periods: 2 times per hour [], 2 to 3 times per day [,], and daily []. Effective interventions had an intervention duration between 3 weeks and 12 months, where the majority of studies had an intervention length between 3 and 8 weeks. The other included studies that indicated no or limited effectiveness and reported heterogeneous features regarding decision points, decision rules, opportune-moment identifications, and the complexity of algorithms as well.
Study QualityThe quality assessment showed that 4 studies achieved a “strong” global rating ( [-]). Overall, 15 studies were rated “moderate,” and all the other studies were judged with 2 or more “weak” ratings. The most common weak ratings in 38 and 17 out of 45 were related to the context of blinding and data collection methods, respectively.
Figure 2. Study quality of included studies based on the Quality Assessment Tool Effective Public Health Practice Project. Green indicates a strong rating, yellow a moderate rating, and red a weak rating of the respective category. The objective of this study was to provide a systematic overview of the concept of JITAIs and to explore important parameters in KOA management. In total, 45 studies with 38 different JITAIs focusing on the outcomes PA, SB, pain, QoL, physical function, and stiffness were analyzed in this review. The majority of studies investigated PA and SB, while physical function, pain, and QoL were addressed by few studies, and the parameter “stiffness” was not included in any study, indicating a research gap. Reporting and use of the key facets of JITAIs [], along with the methodology and aspects of study design of the included studies, were heterogeneous. The best evidence for effective JITAIs exists for device-based measurements of PA and SB using frequent decision points and simple algorithms to deliver prompts after a period of physical inactivity or SB. This can be translated to the use case of KOA and accompanied by a specific knee load management to improve KOA treatments with JITAIs in the future.
JITAIs in the Context of PA, SB, Pain, Stiffness, Physical Function, and QoLOverall intervention effectiveness is hard to pinpoint and not comparable between studies due to heterogeneous study design, outcome measures, number of statistical comparisons, time windows, sample size, and focus on within- or between-person effects. Stratifying the results by study design, 5 out of 14 RCTs found the intervention to effectively increase PA and 3 studies had a positive effect on SB, and 1 study improved the pain level of participants. Pre-post study designs as well as the within-person studies found an increase in PA in 8 studies, a reduction of SB in 5 studies, a reduction of pain in 1 study, and 1 study positively influenced QoL. No evidence of effectiveness in studies for physical function outcomes was found. To obtain more results about causation under consideration of time-varying effects, it is important to increase the number of microrandomized trials in the future [] and focus on the state of receptivity of participants. This will also provide more information about what works in which situation and for whom, and it will help to provide valuable information to advance the field of JITAI research.
Although the studies showed a high degree of heterogeneity, some patterns can be observed. For example, decision points with a shorter period (eg, minute-to-minute) and an intervention duration of 3 to 8 weeks have shown good evidence for intervention effectiveness. Here, it is important to add that intervention duration and timing of decision points have to be viewed together to avoid overburdening participants and to keep the content engaging. Otherwise, JITAIs run at a high risk of disengagement [] that can hinder behavior change. The optimal dose for these parameters has yet to be completely explored. Conroy et al [] point out that self-monitoring prompts, but not behavioral feedback, show signs of a dose-response association with PA volume in JITAIs. This is an important finding that should be followed up by further studies. The included studies provide limited information regarding the long-term effectivity of JITAIs for behavior change. Of the 15 studies that included a follow-up measurement, 3 studies showed significant benefits of the intervention for QoL [,,] and 2 studies for PA [,]. This is not surprising, as JITAIs are an upcoming topic and still focus mainly on the opportune-moment identification, feasibility, usability, and technological challenges of the intervention []. In addition, decision rules that represent opportune moments when a threshold of 20 to 120 minutes of inactivity, SB, or computer activity has been reached resulted in behavior change. Here, shorter timeframes might be more useful to break SB, as Thomas and Bond [] have shown that opportune moments after 30 minutes of being sedentary are more effective than longer decision points of 60 and 120 minutes.
PA or SB was monitored in all significant studies by the smartphone app or by an accelerometer. In addition, prompts were used by each study as an intervention option to interrupt SB, send encouraging messages, give suggestions to increase PA, or send useful and interesting links and information. For example, Freene et al [] sent prompts in the form of do’s to help change behavior and learn a new behavior. The unique feature of this JITAI was that the do’s were sent based on a variety of tailoring variables, namely if participants showed a low score in PA, social opportunity, and variety on 3 consecutive days. Although it was not possible to positively influence PA or SB, the parameter QoL could be increased in 4 different domains, both after 6 and after 16 weeks. This result could mean that variety and social opportunity can be more easily influenced by do’s than PA. Another reason could be that shorter decision points are more effective for PA while for QoL longer decision points such as days are sufficient.
Importantly, Ding et al [] highlighted that 43% of the supposed “just-in-time messages” were not perceived as just-in-time by the participants. This is problematic, as sending prompts at moments when participants do not have the opportunity to change their behavior will not lead to the desired effect and might even lead to disengagement with the intervention []. The importance of adequate opportune-moment identification has also been highlighted by previous research [,] and should be assessed by future JITAI studies. This can be done by a repeated short EMA, which then can also inform and adapt the timing of the intervention and improve adherence and thus behavior change. One example is the study of Rabbi et al [], who included participants preferences into the opportune-moment identification, which allowed their algorithm to, for example, adapt learned opportune moments that were no longer valid.
Shortcomings of the included studies include the overall weak study quality and the mainly simple intervention design that is not optimized for opportune-moment identification as it seldomly includes the state of receptivity of participants but focuses on the vulnerability. This should be the focus of future studies that consider participants to, for example, not initiate a JITAI trigger when the tailoring variables (eg, calendar or user preferences) indicate that the participant is not receptive []. In this context, the role of contextual factors assessed by GPS or EMA [] are important aspects that are included in a few studies [,,-,,,-,,,] and point to the feasibility of such interventions. If interventions including context variables are more effective than those that do not, this cannot be answered by this review as no clear pattern emerges. The additional information provided by these studies is nonetheless very helpful for future interventions and should be explored further, including EMA studies on this topic []. The blinding of participants was the main issue for the quality assessment. While it is difficult to blind participants in mHealth research, microrandomization makes this process easier and has a lot of additional benefits, as discussed before. Furthermore, the included studies scored weak to moderate for selection bias. Here, more diverse samples and nonconvenient samples are required in future research to strengthen the evidence base and the generalization of JITAI study results. Additionally, future studies should test different sets of decision points, tailoring variables, decision rules, and intervention options within one study to improve the current literature on what works for whom and how. This would improve the evidence base for future JITAI interventions in different populations and settings.
Specific Considerations for the Transfer of the Finding to the Use Case of KOAAs a recommendation, the World Health Organization [] states that a healthy lifestyle should be cultivated and that maintaining a normal body weight plays an important role in preventing osteoarthritis and control the disease progression. Here, PA and SB are behavioral variables that influence pain, physical function, stiffness, and QoL of patients with KOA [-]. Consequently, PA and SB can be used to a certain extent as proxy variables to target the distal outcomes pain, physical function, stiffness, and QoL. However, even though appropriate mechanical stimuli are needed for promoting optimal joint function [], more PA is not necessarily better for patients with KOA [] and can even support the development of the disease []. It is recommended to reduce overuse of joints by refraining from PA that contain start-stop movements, rapid changes in direction, intense jumps, and landings [].
To design a JITAI for KOA, specific tailoring variables have to be considered. On the basis of our findings, starting points for intervention studies could include device-based measured PA and SB variables in addition to time, location, and pain scores assessed as user input through a mix of event-based and random sampling using EMAs [] to tailor the intervention to participants needs. From a theoretical perspective, additional JITAIs for stiffness would be very helpful to provide an evidence base for the creation of a JITAI for KOA. A further important tailoring variable that has not been found by this review is the knee load. Here, body-worn measurement technologies such as orthopedic shoe insoles or knee braces with built-in sensors to measure the forces exerted on the knee joint are promising developments for interventions that can adapt the minimum and maximum load limit to each participant and situation [-]. Wearable joint monitoring technologies may also help indirectly assess varying levels of joint stiffness.
For example, by quantifying the joint range of motion [] or assessing relationships between gait biomechanics and fluctuations in KOA symptoms, valuable data can be gained []. Additionally, EMA could be used for a targeted assessment of specific symptoms, such as morning stiffness. Pain can also be assessed as user input through a mix of event-based and random sampling using EMAs []. This would allow to tailor the JITAI to, for example, a user specific self-reported amount of pain, or knee load under consideration of certain conditions and allow for a highly individualized approach.
A further important intervention method found by this review are educational prompts. These prompts could contain informative facts to educate the user on how to move and exercise appropriately [] tailored to, for example, their PA, SB, pain, and stiffness. While conventional JITAIs mainly aim to increase PA through classical exercise, JITAIs that focus on the therapy of specific diseases, such as KOA, could use therapeutic exercises and tasks to promote health behavior. For example, it has already been shown that a combination of muscle strengthening and range-of-motion exercises resulted in a significant improvement in osteoarthritis symptoms in the long term [,]. Moreover, muscle strengthening was associated with reduced pain and increased function in patients with KOA []. These exercises could be encouraged in appropriate moments through the use of a JITAI. In the deployment of digital interventions and JITAIs, it is imperative to account for the technological prerequisites and potential barriers for participants and practitioners. This consideration becomes especially significant when the intended audience includes older patients, exemplified by those with KOA, as underscored in the research conducted by Hinman et al [].
Designing a JITAI for the Use Case of KOASeveral important decisions need to be made to design a JITAI for KOA. First, the outcome parameters need to be selected. Second, the device for JITAI delivery and assessment of tailoring variables needs to be chosen. Third, tailoring variables, decision points, intervention options, and decision rules need to be defined.
Derived from previous research and the results of this review, the aim of a promising JITAI is to target the proximal outcomes PA and SB in accordance with the recommendations of KOA therapy []. This should result in positive developments in the distal outcomes of pain, physical function, stiffness, and QoL. Here, smartphones, accelerometers, or KOA-specific measurement technologies can be used for the intervention and the evaluation of effectiveness. In addition to device-based measured PA and SB, EMAs can be used to gather information through user input about loads of previous activities or provide inferences about parameters such as pain and stiffness. The potential of EMAs for understanding individual behavior is indispensable [,] and 2 studies including EMA showed significant pain reduction in participants [,]. In addition, the use of sensors for load management would be very useful to provide device-based feedback of knee load and adapt the intervention accordingly to prevent an overloading of the knee joint [,]. Further approaches could also include smartphone data about the weather, calendar, or the user’s location to improve the opportune-moment identification [].
The time span in which the decision points are to be made should be of shorter duration, as emerged from the results. Decision points every minute proved to be useful and offer quick adaptation to changing situations. That is especially important to avoid too high knee load in near real time and to break up SB. Prompts to boost overall PA and deliver knowledge could use longer decision points to avoid overburdening the participants [].
The best evidence regarding intervention options exists for prompts to break prolonged SB or physical inactivity. That is also a relevant parameter for KOA [] and could be linked with positive reinforcement messages or booster messages, which showed success in the study by Martin et al []. Additional intervention options include maintaining an optimal amount of PA while avoiding overloading of the osteoarthritis-affected knee. This could be achieved by continuously tracking and displaying the estimated knee load (eg, using knee-specific feedback technology [,]) on the smartphone and prompting participants in near real time if the knee load reaches a critical individual level. To avoid insufficient PA of participants, daily PA goals could be used. Here, daily prompts with positive reinforcement messages could prompt participants in the morning if they failed to achieve the PA goal of the previous day.
In the next step, the decision rules can be formulated. On the basis of the study by Thomas and Bond [], interrupting SB after 30 minutes is promising, which is also linked to improved metabolic health []. Here, it is important to consider that the prompt really appears in an opportune moment [] to avoid overburdening participants, especially if such a short time frame is chosen. One option to solve this is to allow users to choose time frames when such triggers occur, enable them to mute the triggers for a certain period, or use more complex decision-making algorithms based on machine learning principles that implement the availability of the user (eg, including calendar). Rules for accompanying positive reinforcement messages and educational information about why PA is useful in the topic of KOA should be formulated to have a maximum number of messages per day. There is also the possibility of including movement recommendations to go for a walk in the nearest park or to explore new places. Decision rules for the optimal amount of PA should only send prompts if participants are closing in on their critical knee load level or if inappropriate kinds of PA such as reoccurring jumps are detected to avoid making participants hesitant about PA. These prompts should always be followed by an explanation for the prompt and recommendation
留言 (0)