An expandable approach for design and personalization of digital, just-in-time adaptive interventions

Experiences and Perceptions of Using Smartphone Ecological Momentary Assessment for Reporting Knee Osteoarthritis Pain and Symptoms

BACKGROUND

Knee osteoarthritis (OA) is a prevalent, painful, and disabling musculoskeletal condition. One method that could more accurately monitor the pain associated with knee OA is ecological momentary assessment (EMA) using a smartphone.

OBJECTIVES

The aim of this study was to explore participant experiences and perceptions of using smartphone EMA as a way of communicating knee OA pain and symptoms following participating in a 2-week smartphone EMA study.

MATERIALS AND METHODS

Using a maximum variation sampling method, participants were invited to share their thoughts and opinions in semistructured focus group interviews. Interviews were recorded and transcribed verbatim before thematic analysis using the general inductive approach.

RESULTS

A total of 20 participants participated in 6 focus groups. Three themes and 7 subthemes were identified from the data. Identified themes included: user experience of smartphone EMA, data quality of smartphone EMA, and practical aspects of smartphone EMA.

DISCUSSION

Overall, smartphone EMA was deemed as being an acceptable method for monitoring pain and symptoms associated with knee OA. These findings will assist researchers in designing future EMA studies alongside clinicians implementing smartphone EMA into practice.

PERSPECTIVE

This study highlights that smartphone EMA is an acceptable method for capturing pain-related symptoms and experiences of those expereiencing knee OA. Future EMA studies should ensure design features are considered that reduce missing data and limit the responder burden to improve data quality.

The interplay between social media use and problematic internet usage: Four behavioral patterns

Objective

The study aims to identify typical interplay between the use of social media apps on smartphones and Problematic Internet Usage (PIU).

Method

Our study utilizes data from a smartphone app that objectively monitors user usage, including the apps used and the start and finish times of each app session. This study included 334 participants who declared a need to be aware of their smartphone usage and control it. Problematic Internet Usage (PIU) was measured using the Problematic Internet Use Questionnaire-Short Form-6 (PIUQ-SF6). The total PIU score can range from 6 to 30, with a score above 15 indicating that a person is at risk of PIU. Time spent on Social Media (SM) apps of Facebook, WhatsApp, and Instagram, and whether people used each of these apps were studied along with the total PIU score. K-Prototype clustering was utilized for the analysis.

Results

Four distinct clusters, typifying the relationship between social media use and PIU, were identified. All the individuals in Cluster 1 (Light SM Use Cluster; Cluster size = 270, 80.84% of total dataset) spent between 0 and 109.01 min on Instagram, between 0 and 69.84 min on Facebook, and between 0 and 86.42 min on WhatsApp and its median PIU score was 17. Those who were in cluster 2 (Highly Visual SM Cluster; Cluster size = 23, 6.89% of total dataset) all used Instagram, and each member spent between 110 and 307.63 min on Instagram daily. The cluster median PIU score and average daily usage of Instagram were respectively 20 and 159.66 min. Those who were in Cluster 3 (Conversational SM Cluster; Cluster size = 19, 5.69% of total dataset) all used WhatsApp, and spent between 76.68 and 225.22 min on WhatsApp daily. The cluster median PIU score and average time spent per day on WhatsApp were 20 and 132.65 min, respectively. Those who were in Cluster 4 (Social Networking Cluster; (Cluster size = 22, 6.59% of total dataset) all used Facebook, and each spent between 73.09 and 272.85 min daily on Facebook. The cluster median PIU score and average time spent per day on Facebook were 18 and 133.61 min respectively.

Conclusion

The clusters indicate that those who use a particular social media app spend significantly less time on other social media apps. This indicates that problematic attachment to social media occurs primarily for one of three reasons: visual content and reels, conversations with peers, or surfing network content and news. This finding will help tailor interventions to fit each cluster, for example by strengthening interpersonal skills and resistance to peer pressure in the case of Cluster 3 and increasing impulse control in the case of Cluster 2.

Efficacy of an mHealth Behavior Change Intervention for Promoting Physical Activity in the Workplace: Randomized Controlled Trial

BACKGROUND

Insufficient physical activity (PA) is a well-established risk factor for several noncommunicable diseases such as cardiovascular diseases, cancer, diabetes, depression, and dementia. The World Health Organization (WHO) advises that individuals engage in 150 minutes of moderate PA per week or 75 minutes of intense PA per week. According to the WHO's latest report, 23% of adults fail to meet the minimum recommended level of PA. The percentage was even higher in a recent global study that showed 27% of adults were insufficiently active and reported a 5% increase in the prevalence trend of insufficient PA between 2001 and 2016. The study also showed the rate of insufficient PA among countries varied significantly. For instance, it was estimated that 40% were insufficiently active in the United States, and the percentage was even higher in Saudi Arabia (more than 50%). Governments are actively developing policies and methods to successfully establish a PA-inducing environment that encourages a healthy lifestyle in order to address the global steady decline in PA.

OBJECTIVE

The purpose of this study was to determine the effectiveness of mobile health (mHealth) interventions, particularly SMS text messaging interventions, to improve PA and decrease BMI in healthy adults in the workplace.

METHODS

In this parallel, 2-arm randomized controlled trial, healthy adults (N=327) were randomized to receive an mHealth intervention (tailored text messages combined with self-monitoring (intervention; n=166) or no intervention (control; n=161). Adults who were fully employed in an academic institution and had limited PA during working hours were recruited for the study. Outcomes, such as PA and BMI, were assessed at baseline and 3 months later.

RESULTS

Results showed significant improvement in PA levels (weekly step counts) in the intervention group (β=1097, 95% CI 922-1272, P<.001). There was also a significant reduction in BMI (β=0.60, 95% CI 0.50-0.69, P<.001).

CONCLUSIONS

Combining tailored text messages and self-monitoring interventions to improve PA and lower BMI was significantly effective and has the potential to leverage current methods to improve wellness among the public.

Investigating the Role of Context in the Delivery of Text Messages for Supporting Psychological Wellbeing

Without a nuanced understanding of users' perspectives and contexts, text messaging tools for supporting psychological wellbeing risk delivering interventions that are mismatched to users' dynamic needs. We investigated the contextual factors that influence young adults' day-to-day experiences when interacting with such tools. Through interviews and focus group discussions with 36 participants, we identified that people's daily schedules and affective states were dominant factors that shape their messaging preferences. We developed two messaging dialogues centered around these factors, which we deployed to 42 participants to test and extend our initial understanding of users' needs. Across both studies, participants provided diverse opinions of how they could be best supported by messages, particularly around when to engage users in more passive versus active ways. They also proposed ways of adjusting message length and content during periods of low mood. Our findings provide design implications and opportunities for context-aware mental health management systems.

An interpretable machine learning approach to multimodal stress detection in a simulated office environment

BACKGROUND AND OBJECTIVE

Work-related stress affects a large part of today's workforce and is known to have detrimental effects on physical and mental health. Continuous and unobtrusive stress detection may help prevent and reduce stress by providing personalised feedback and allowing for the development of just-in-time adaptive health interventions for stress management. Previous studies on stress detection in work environments have often struggled to adequately reflect real-world conditions in controlled laboratory experiments. To close this gap, in this paper, we present a machine learning methodology for stress detection based on multimodal data collected from unobtrusive sources in an experiment simulating a realistic group office environment (N=90).

METHODS

We derive mouse, keyboard and heart rate variability features to detect three levels of perceived stress, valence and arousal with support vector machines, random forests and gradient boosting models using 10-fold cross-validation. We interpret the contributions of features to the model predictions with SHapley Additive exPlanations (SHAP) value plots.

RESULTS

The gradient boosting models based on mouse and keyboard features obtained the highest average F1 scores of 0.625, 0.631 and 0.775 for the multiclass prediction of perceived stress, arousal and valence, respectively. Our results indicate that the combination of mouse and keyboard features may be better suited to detect stress in office environments than heart rate variability, despite physiological signal-based stress detection being more established in theory and research. The analysis of SHAP value plots shows that specific mouse movement and typing behaviours may characterise different levels of stress.

CONCLUSIONS

Our study fills different methodological gaps in the research on the automated detection of stress in office environments, such as approximating real-life conditions in a laboratory and combining physiological and behavioural data sources. Implications for field studies on personalised, interpretable ML-based systems for the real-time detection of stress in real office environments are also discussed.

Adaptive interventions for opioid prescription management and consumption monitoring

OBJECTIVES

While opioid addiction, treatment, and recovery are receiving attention, not much has been done on adaptive interventions to prevent opioid use disorder (OUD). To address this, we identify opioid prescription and opioid consumption as promising targets for adaptive interventions and present a design framework.

MATERIALS AND METHODS

Using the framework, we designed Smart Prescription Management (SPM) and Smart Consumption Monitoring (SCM) interventions. The interventions are evaluated using analytical modeling and secondary data on doctor shopping, opioid overdose, prescription quality, and cost components.

RESULTS

SPM was most effective (30-90% improvement, for example, prescriptions reduced from 18 to 1.8 per patient) for extensive doctor shopping and reduced overdose events and mortality. Opioid adherence was improved and the likelihood of addiction declined (10-30%) as the response rate to SCM was increased. There is the potential for significant incentives ($2267-$3237) to be offered for addressing severe OUD.

DISCUSSION

The framework and designed interventions adapt to changing needs and conditions of the patients to become an important part of global efforts in preventing OUD. To the best of our knowledge, this is the first paper on adaptive interventions for preventing OUD by addressing both prescription and consumption.

CONCLUSION

SPM and SCM improved opioid prescription and consumption while reducing the risk of opioid addiction. These interventions will assist in better prescription decisions and in managing opioid consumption leading to desirable outcomes. The interventions can be extended to other substance use disorders and to study complex scenarios of prescription and nonprescription opioids in clinical studies.

Leveraging Mobile Phone Sensors, Machine Learning and Explainable Artificial Intelligence to Predict Imminent Same-Day Binge Drinking Events to Support Just-In-Time Adaptive Interventions: A Feasibility Study

BACKGROUND

Digital Just-In-Time Adaptive Interventions (JITAIs) can reduce binge drinking events (BDEs: consuming 4+/5+ drinks per occasion for women/men) in young adults, but need to be optimized for timing and content. Delivering just-in-time support messages in the hours prior to BDEs could improve intervention impact.

OBJECTIVE

We determined the feasibility of developing a machine learning model to accurately predict future, that is, same-day, 1 to 6-hours prior BDEs using smartphone sensor data. We aimed to identify the most informative phone sensor features associated with BDEs on weekend and weekdays, respectively, to determine the key features that explain prediction model performance.

METHODS

We collected phone sensor data from 75 young adults (ages 21-25; mean =22.4, SD=1.9) with risky drinking behavior who reported drinking behavior over 14 weeks. Participants in this secondary analysis were enrolled in a clinical trial. We developed machine learning models testing different algorithms (e.g., XGBoost, decision tree) to predict same-day BDEs (versus low-risk drinking events and non-drinking periods) using smartphone sensor data (e.g., accelerometer, GPS). We tested various "prediction distance" time windows (more proximal: 1-hour; to distant: 6-hour) from drinking onset. We also tested various analysis time windows (i.e., amount of data to be analyzed), ranging from 1 to 12 hours prior to drinking onset, because this determines the amount of data that needs to be stored on the phone to compute the model. Explainable AI (XAI) was used to explore interactions between the most informative phone sensor features contributing to BDEs.

RESULTS

The XGBoost model performed best in predicting imminent same-day BDE, with 95.0% accuracy on weekends and 94.3% accuracy on weekdays (F1 score = 0.95 and 0.94, respectively). This XGBoost model needed 12- and 9-hours of phone sensor data at 3- and 6- hours prediction distance from the onset of drinking, on weekends and weekdays, respectively, prior to predicting same-day BDEs. The most informative phone sensor features for BDE prediction were time (e.g., time of day) and GPS-derived, such as radius of gyration (an indicator of travel). Interactions among key features (e.g., time of day, GPS-derived features) contributed to prediction of same-day BDE.

CONCLUSIONS

We demonstrated the feasibility and potential use of smartphone sensor data and machine learning to accurately predict imminent (same-day) BDEs in young adults. The prediction model provides "windows of opportunity" and with the adoption of XAI, we identified "key contributing features" to trigger JITAI prior to the onset of BDEs, with the potential to reduce the likelihood of BDEs in young adults.

CLINICALTRIAL

The Patient Experience of the Future is Personalized: Using Technology to Scale an N of 1 Approach

Personalized experiences are more effective at creating sustained behavior change. Digitally enabled personalized outreach can improve patient’s experience by providing relevant, meaningful calls to action at a time when labor-intensive human-to-human personalization is challenged by systemic health staffing shortages. Strategic use of digital tools to engage patients and supplement human-to-human care scale personalization to the benefit of patient and provider experience. Specifically, digital personalization can support: Identification of patients eligible for a procedure, service, or outreach Engaging patients with a personalized call to action Augmenting care through the use of digital tools, and Monitoring patient progress over time to ensure continued support. The technology to support a more personalized patient experience includes infrastructure to consolidate rich data, an intelligence capability to identify candidates for each call to action, and an engagement layer that presents patients with personalized output. Steps to develop and execute a personalization strategy are provided.

Digital headache self-management interventions for patients with a primary headache disorder: A systematic review of randomized controlled trials

OBJECTIVE

This article systematically reviews the empirical literature examining the efficacy of digital headache management interventions for patients with a primary headache disorder.

BACKGROUND

Digital headache management interventions provide opportunities to improve access to behavioral headache interventions to underserved groups.

METHODS

A systematic search of PubMed, Scopus, and EBSCO (PsycInfo, Education Research Complete, ERIC, Health Source: Nursing/Academic Edition, Psychology and Behavioral Sciences Collection) and reference review was conducted. Included studies had to recruit a sample with a primary headache diagnosis, be a randomized controlled trial including a digital component, assess a headache outcome (i.e., frequency, duration, severity, intensity, disability) or quality of life, and be published in English. Two authors independently extracted data for included studies. The methodological quality of studies was assessed using the revised Cochrane risk-of-bias tool.

RESULTS

Thirteen studies with unique interventions met inclusion criteria. More than half of the studies were pilots; however, nearly 70% (9/13) demonstrated significant between-group or within-group improvements on one or more headache-related outcomes. All interventions included some form of relaxation training and the majority were delivered via interactive website. While fewer than half the studies report participant race and/or ethnicity, of those that do, 83% (5/6) reported a predominately White/Caucasian sample.

CONCLUSIONS

Efficacy testing of digital headache interventions is in its infancy with the majority of these studies relying on pilot studies with small samples comprised of homogenous patient populations. Interactive websites were the most common digital medium to deliver digital headache management interventions and have demonstrated promising results. Further testing using large-scale randomized controlled trials and exploration of other digital tools is warranted. Future studies with more diverse samples are needed to inform health equity of digital headache interventions.

The role of socio-demographic factors and physical functioning in the intra- and interpersonal variability of older adults' sedentary time: an observational two-country study

BACKGROUND

Insight into the variability of older adults' sedentary time is needed to inform future interventions. The aim of this study was to examine the intra- and interpersonal variability in sedentary time, and the moderating role of socio-demographics, physical functioning and geographical location in this variability.

METHODS

Cross-sectional data from 818 community-dwelling older adults (mean age: 74.8 years; 61.1%women) of the Active Lifestyle and the Environment in Chinese Seniors and Belgian Environmental Physical Activity Study in Seniors were used. An interview questionnaire was administered to collect socio-demographic information. The Short Physical Performance Battery was performed to evaluate physical functioning, and Actigraph GT3X( +) accelerometers were used to estimate sedentary time. Linear mixed models with random intercepts at the neighborhood, person and day levels examined the variability in sedentary time, and the moderating role of socio-demographics, physical functioning and geographical location within this variability.

RESULTS

Most of the variance in accelerometry-assessed sedentary time was due to intrapersonal variability across periods of the day (72.4%) followed by interpersonal variability within neighborhoods (25.6%). Those who were older, men, lived in Hong Kong, and experienced a lower level of physical functioning were more sedentary than their counterparts. Sedentary time increased throughout the day, with highest levels of sedentary time observed between 6:00 and 9:00 pm. The patterns of sedentary time across times of the day differed by gender, educational attainment, age, physical functioning and/or geographical location. No significant differences were detected between week and weekend day sedentary time.

CONCLUSIONS

The oldest old, men, and those with functional limitations are important target groups for sedentary behavior interventions. As sedentary time was the highest in the evening future sedentary behavior intervention should pay particular attention to the evening hours. The variations in diurnal patterns of sedentary time between population subgroups suggest that personalized just-in-time adaptive interventions might be a promising strategy to reduce older adults' sedentary time.

Improving Outcomes Through Personalized Recommendations in a Remote Diabetes Monitoring Program: Observational Study

Background

Diabetes management is complex, and program personalization has been identified to enhance engagement and clinical outcomes in diabetes management programs. However, 50% of individuals living with diabetes are unable to achieve glycemic control, presenting a gap in the delivery of self-management education and behavior change. Machine learning and recommender systems, which have been used within the health care setting, could be a feasible application for diabetes management programs to provide a personalized user experience and improve user engagement and outcomes.

Objective

This study aims to evaluate machine learning models using member-level engagements to predict improvement in estimated A_1c and develop personalized action recommendations within a remote diabetes monitoring program to improve clinical outcomes.

Methods

A retrospective study of Livongo for Diabetes member engagement data was analyzed within five action categories (interacting with a coach, reading education content, self-monitoring blood glucose level, tracking physical activity, and monitoring nutrition) to build a member-level model to predict if a specific type and level of engagement could lead to improved estimated A_1c for members with type 2 diabetes. Engagement and improvement in estimated A_1c can be correlated; therefore, the doubly robust learning method was used to model the heterogeneous treatment effect of action engagement on improvements in estimated A_1c.

Results

The treatment effect was successfully computed within the five action categories on estimated A_1c reduction for each member. Results show interaction with coaches and self-monitoring blood glucose levels were the actions that resulted in the highest average decrease in estimated A_1c (1.7% and 1.4%, respectively) and were the most recommended actions for 54% of the population. However, these were found to not be the optimal interventions for all members; 46% of members were predicted to have better outcomes with one of the other three interventions. Members who engaged with their recommended actions had on average a 0.8% larger reduction in estimated A_1c than those who did not engage in recommended actions within the first 3 months of the program.

Conclusions

Personalized action recommendations using heterogeneous treatment effects to compute the impact of member actions can reduce estimated A_1c and be a valuable tool for diabetes management programs in encouraging members toward actions to improve clinical outcomes.

A personalized mobile app for physical activity: An experimental mixed-methods study

Objectives

To investigate the feasibility of the be.well app and its personalization approach which regularly considers users’ preferences, amongst university students.

Methods

We conducted a mixed-methods, pre-post experiment, where participants used the app for 2 months. Eligibility criteria included: age 18–34 years; owning an iPhone with Internet access; and fluency in English. Usability was assessed by a validated questionnaire; engagement metrics were reported. Changes in physical activity were assessed by comparing the difference in daily step count between baseline and 2 months. Interviews were conducted to assess acceptability; thematic analysis was conducted.

Results

Twenty-three participants were enrolled in the study (mean age = 21.9 years, 71.4% women). The mean usability score was 5.6 ± 0.8 out of 7. The median daily engagement time was 2 minutes. Eighteen out of 23 participants used the app in the last month of the study. Qualitative data revealed that people liked the personalized activity suggestion feature as it was actionable and promoted user autonomy. Some users also expressed privacy concerns if they had to provide a lot of personal data to receive highly personalized features. Daily step count increased after 2 months of the intervention (median difference = 1953 steps/day, p-value <.001, 95% CI 782 to 3112).

Conclusions

Incorporating users’ preferences in personalized advice provided by a physical activity app was considered feasible and acceptable, with preliminary support for its positive effects on daily step count. Future randomized studies with longer follow up are warranted to determine the effectiveness of personalized mobile apps in promoting physical activity.

REinforcement learning to improve non-adherence for diabetes treatments by Optimising Response and Customising Engagement (REINFORCE): study protocol of a pragmatic randomised trial

INTRODUCTION

Achieving optimal diabetes control requires several daily self-management behaviours, especially adherence to medication. Evidence supports the use of text messages to support adherence, but there remains much opportunity to improve their effectiveness. One key limitation is that message content has been generic. By contrast, reinforcement learning is a machine learning method that can be used to identify individuals' patterns of responsiveness by observing their response to cues and then optimising them accordingly. Despite its demonstrated benefits outside of healthcare, its application to tailoring communication for patients has received limited attention. The objective of this trial is to test the impact of a reinforcement learning-based text messaging programme on adherence to medication for patients with type 2 diabetes.

METHODS AND ANALYSIS

In the REinforcement learning to Improve Non-adherence For diabetes treatments by Optimising Response and Customising Engagement (REINFORCE) trial, we are randomising 60 patients with suboptimal diabetes control treated with oral diabetes medications to receive a reinforcement learning intervention or control. Subjects in both arms will receive electronic pill bottles to use, and those in the intervention arm will receive up to daily text messages. The messages will be individually adapted using a reinforcement learning prediction algorithm based on daily adherence measurements from the pill bottles. The trial's primary outcome is average adherence to medication over the 6-month follow-up period. Secondary outcomes include diabetes control, measured by glycated haemoglobin A1c, and self-reported adherence. In sum, the REINFORCE trial will evaluate the effect of personalising the framing of text messages for patients to support medication adherence and provide insight into how this could be adapted at scale to improve other self-management interventions.

ETHICS AND DISSEMINATION

This study was approved by the Mass General Brigham Institutional Review Board (IRB) (USA). Findings will be disseminated through peer-reviewed journals, clinicaltrials.gov reporting and conferences.

TRIAL REGISTRATION NUMBER

Clinicaltrials.gov (NCT04473326).

Cyberbullying Prevention for Adolescents: Iterative Qualitative Methods for Mobile Intervention Design

BACKGROUND

Cybervictimization among adolescents is associated with multiple negative mental health consequences. Although pediatricians often screen for cyberbullying, validated and acceptable programs to reduce the frequency and impact of adolescent cybervictimization are lacking.

OBJECTIVE

This study uses agile qualitative methods to refine and evaluate the acceptability of a mixed-modality intervention, initiated within the context of usual pediatric care, for adolescents with a history of cyberharassment and cyberbullying victimization.

METHODS

Three groups of adolescents were successively recruited from an urban primary care clinic to participate in three consecutive iterations (1, 2, and 3) of the program, which consisted of a brief in-clinic intervention followed by 8 weeks of daily, automated SMS text messaging. After 2 weeks of messaging, iteration 1 (I1) participants completed semistructured interviews regarding intervention experiences. Participant feedback was evaluated via framework matrix analysis to guide changes to the program for iteration 2 (I2). Feedback from 2-week interviews of I2 participants was similarly used to improve the program before initiating iteration 3 (I3). Participants in all 3 iterations completed the interviews after completing the program (8 weeks). Daily response rates assessed participant engagement, and satisfaction questionnaires assessed acceptability.

RESULTS

A total of 19 adolescents (aged 13-17 years) reporting past-year cybervictimization were enrolled: 7 in I1, 4 in I2, and 8 in I3. Demographic variables included the following: a mean age of 15 (SD 1.5) years; 58% (11/19) female, 42% (8/19) male, 63% (12/19) Hispanic, 37% (7/19) non-Hispanic, 79% (15/19) people of color, and 21% (4/19) White. A total of 73% (14/19) self-identified as having a low socioeconomic status, and 37% (7/19) self-identified as lesbian, gay, or bisexual. The average past 12-month cybervictimization score at baseline was 8.2 (SD 6.58; range 2-26). Participant feedback was used to iteratively refine intervention content and design. For example, participants in I1 recommended that the scope of the intervention be expanded to include web-based conflicts and drama, rather than narrowly focusing on cyberbullying prevention. On the basis of this feedback, the I2 content was shifted toward more general de-escalation skills and bystander empowerment. Overall, 88.34% (940/1064) of the daily queries sent to participants across all 3 iterations received a reply. Participant satisfaction improved considerably with each iteration; 0% (0/7) of I1 participants rated the overall quality of Intervention to Prevent Adolescent Cybervictimization with Text message as excellent, compared to 50% (2/4) of I2 participants and 86% (6/7) of I3 participants. Engagement also improved between the first and third iterations, with participants replying to 59.9% (235/392) of messages in I1, compared to 79.9% (358/488) of messages in I3.

CONCLUSIONS

This study shows the value of structured participant feedback gathered in an agile intervention refinement methodology for the development of a technology-based intervention targeting adolescents.

Personalized mobile technologies for lifestyle behavior change: A systematic review, meta-analysis, and meta-regression

Given that the one-size-fits-all approach to mobile health interventions have limited effects, a personalized approach might be necessary to promote healthy behaviors and prevent chronic conditions. Our systematic review aims to evaluate the effectiveness of personalized mobile interventions on lifestyle behaviors (i.e., physical activity, diet, smoking and alcohol consumption), and identify the effective key features of such interventions. We included any experimental trials that tested a personalized mobile app or fitness tracker and reported any lifestyle behavior measures. We conducted a narrative synthesis for all studies, and a meta-analysis of randomized controlled trials. Thirty-nine articles describing 31 interventions were included (n = 77,243, 64% women). All interventions personalized content and rarely personalized other features. Source of data included system-captured (12 interventions), user-reported (11 interventions) or both (8 interventions). The meta-analysis showed a moderate positive effect on lifestyle behavior outcomes (standardized difference in means [SDM] 0.663, 95% CI 0.228 to 1.10). A meta-regression model including source of data found that interventions that used system-captured data for personalization were associated with higher effectiveness than those that used user-reported data (SDM 1.48, 95% CI 0.76 to 2.19). In summary, the field is in its infancy, with preliminary evidence of the potential efficacy of personalization in improving lifestyle behaviors. Source of data for personalization might be important in determining intervention effectiveness. To fully exploit the potential of personalization, future high-quality studies should investigate the integration of multiple data from different sources and include personalized features other than content.

A reinforcement learning based algorithm for personalization of digital, just-in-time, adaptive interventions

Suboptimal health related behaviors and habits; and resulting chronic diseases are responsible for majority of deaths globally. Studies show that providing personalized support to patients yield improved results by preventing and/or timely treatment of these problems. Digital, just-in-time and adaptive interventions are mobile phone-based notifications that are being utilized to support people wherever and whenever necessary in coping with their health problems. In this research, we propose a reinforcement learning-based mechanism to personalize interventions in terms of timing, frequency and preferred type(s). We simultaneously employ two reinforcement learning models, namely intervention-selection and opportune-moment-identification; capturing and exploiting changes in people's long-term and momentary contexts respectively. While the intervention-selection model adapts the intervention delivery with respect to type and frequency, the opportune-moment-identification model tries to find the most opportune moments to deliver interventions throughout a day. We propose two accelerator techniques over the standard reinforcement learning algorithms to boost learning performance. First, we propose a customized version of eligibility traces for rewarding past actions throughout an agent's trajectory. Second, we utilize the transfer learning method to reuse knowledge across multiple learning environments. We validate the proposed approach in a simulated experiment where we simulate four personas differing in their daily activities, preferences on specific intervention types and attitudes towards the targeted behavior. Our experiments show that the proposed approach yields better results compared to the standard reinforcement learning algorithms and successfully capture the simulated variations associated with the personas.

Precision clinical trials: a framework for getting to precision medicine for neurobehavioural disorders

The goal of precision medicine (individually tailored treatments) is not being achieved for neurobehavioural conditions such as psychiatric disorders. Traditional randomized clinical trial methods are insufficient for advancing precision medicine because of the dynamic complexity of these conditions. We present a pragmatic solution: the precision clinical trial framework, encompassing methods for individually tailored treatments. This framework includes the following: (1) treatment-targeted enrichment, which involves measuring patients' response after a brief bout of an intervention, and then randomizing patients to a full course of treatment, using the acute response to predict long-term outcomes; (2) adaptive treatments, which involve adjusting treatment parameters during the trial to individually optimize the treatment; and (3) precise measurement, which involves measuring predictor and outcome variables with high accuracy and reliability using techniques such as ecological momentary assessment. This review summarizes precision clinical trials and provides a research agenda, including new biomarkers such as precision neuroimaging, transcranial magnetic stimulation-electroencephalogram digital phenotyping and advances in statistical and machine-learning models. Validation of these approaches - and then widespread incorporation of the precision clinical trial framework - could help achieve the vision of precision medicine for neurobehavioural conditions.

Key facets to build up eHealth and mHealth interventions to enhance physical activity, sedentary behavior and nutrition in healthy subjects - an umbrella review

BACKGROUND

Electronic (eHealth) and mobile (mHealth) health interventions can provide a large coverage, and are promising tools to change health behavior (i.e. physical activity, sedentary behavior and healthy eating). However, the determinants of intervention effectiveness in primary prevention has not been explored yet. Therefore, the objectives of this umbrella review were to evaluate intervention effectiveness, to explore the impact of pre-defined determinants of effectiveness (i.e. theoretical foundations, behavior change techniques, social contexts or just-in-time adaptive interventions), and to provide recommendations for future research and practice in the field of primary prevention delivered via e/mHealth technology.

METHODS

PubMed, Scopus, Web of Science and the Cochrane Library were searched for systematic reviews and meta-analyses (reviews) published between January 1990 and May 2020. Reviews reporting on e/mHealth behavior change interventions in physical activity, sedentary behavior and/or healthy eating for healthy subjects (i.e. subjects without physical or physiological morbidities which would influence the realization of behaviors targeted by the respective interventions) were included if they also investigated respective theoretical foundations, behavior change techniques, social contexts or just-in-time adaptive interventions. Included studies were ranked concerning their methodological quality and qualitatively synthesized.

RESULTS

The systematic search revealed 11 systematic reviews and meta-analyses of moderate quality. The majority of original research studies within the reviews found e/mHealth interventions to be effective, but the results showed a high heterogeneity concerning assessment methods and outcomes, making them difficult to compare. Whereas theoretical foundation and behavior change techniques were suggested to be potential positive determinants of effective interventions, the impact of social context remains unclear. None of the reviews included just-in-time adaptive interventions.

CONCLUSION

Findings of this umbrella review support the use of e/mHealth to enhance physical activity and healthy eating and reduce sedentary behavior. The general lack of precise reporting and comparison of confounding variables in reviews and original research studies as well as the limited number of reviews for each health behavior constrains the generalization and interpretation of results. Further research is needed on study-level to investigate effects of versatile determinants of e/mHealth efficiency, using a theoretical foundation and additionally explore the impact of social contexts and more sophisticated approaches like just-in-time adaptive interventions.

TRIAL REGISTRATION

The protocol for this umbrella review was a priori registered with PROSPERO: CRD42020147902 .

Bayesian strategy selection identifies optimal solutions to complex problems using an example from GP prescribing

Complex health problems require multi-strategy, multi-target interventions. We present a method that uses machine learning techniques to choose optimal interventions from a set of possible interventions within a case study aiming to increase General Practitioner (GP) discussions of physical activity (PA) with their patients. Interventions were developed based on a causal loop diagram with 26 GPs across 13 clinics in Geelong, Australia. GPs prioritised eight from more than 80 potential interventions to increase GP discussion of PA with patients. Following a 2-week baseline, a multi-arm bandit algorithm was used to assign optimal strategies to GP clinics with the target outcome being GP PA discussion rates. The algorithm was updated weekly and the process iterated until the more promising strategies emerged (a duration of seven weeks). The top three performing strategies were continued for 3 weeks to improve the power of the hypothesis test of effectiveness for each strategy compared to baseline. GPs recorded a total of 11,176 conversations about PA. GPs identified 15 factors affecting GP PA discussion rates with patients including GP skills and awareness, fragmentation of care and fear of adverse outcomes. The two most effective strategies were correctly identified within seven weeks of the algorithm-based assignment of strategies. These were clinic reception staff providing PA information to patients at check in and PA screening questionnaires completed in the waiting room. This study demonstrates an efficient way to test and identify optimal strategies from multiple possible solutions.