Are Nomothetic or Ideographic Approaches Superior in Predicting Daily Exercise Behaviors?

Cheung, Ying Kuen; Hsueh, Pei-Yun Sabrina; Qian, Min; Yoon, Sunmoo; Meli, Laura; Diaz, Keith M.; Schwartz, Joseph E.; Kronish, Ian M.; Davidson, Karina W.

doi:10.3414/me16-02-0051

Cited by 29 publications

(22 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our findings complement a body of published work by other investigators who have used machine learning with GPS-based, EMA-based, or sensor-based inputs to predict drug use, 23 smoking, 24,25 exercising, 26 diet-related behaviors, [27][28][29][30] and mood changes, [31][32][33][34][35] on time scales ranging from hours to days. A closely related body of work used similar inputs for automated detection of current (not future) cigarette cravings, 36 food cravings, 37 stress, [38][39][40][41] drinking, 42 manic episodes, 43,44 and mood.…”

Section: Discussionsupporting

confidence: 79%

Prediction of stress and drug craving ninety minutes in the future with passively collected GPS data

et al. 2020

View full text Add to dashboard Cite

Just-in-time adaptive interventions (JITAIs), typically smartphone apps, learn to deliver therapeutic content when users need it. The challenge is to "push" content at algorithmically chosen moments without making users trigger it with effortful input. We trained a randomForest algorithm to predict heroin craving, cocaine craving, or stress (reported via smartphone app 3x/day) 90 min into the future, using 16 weeks of field data from 189 outpatients being treated for opioid-use disorder. We used only one form of continuous input (along with person-level demographic data), collected passively: an indicator of environmental exposures along the past 5 h of movement, as assessed by GPS. Our models achieved excellent overall accuracy-as high as 0.93 by the end of 16 weeks of tailoring-but this was driven mostly by correct predictions of absence. For predictions of presence, "believability" (positive predictive value, PPV) usually peaked in the high 0.70s toward the end of the 16 weeks. When the prediction target was more rare, PPV was lower. Our findings complement those of other investigators who use machine learning with more broadly based "digital phenotyping" inputs to predict or detect mental and behavioral events. When target events are comparatively subtle, like stress or drug craving, accurate detection or prediction probably needs effortful input from users, not passive monitoring alone. We discuss ways in which accuracy is difficult to achieve or even assess, and warn that high overall accuracy (including high specificity) can mask the abundance of false alarms that low PPV reveals.npj Digital Medicine (2020) 3:26 ; https://doi.

show abstract

Section: Discussionsupporting

confidence: 79%

Prediction of stress and drug craving ninety minutes in the future with passively collected GPS data

et al. 2020

View full text Add to dashboard Cite

show abstract

“…By coupling EMA with mobile devices, the TQL method can help make sense of the incoming streams of repeatedly collected exposure data (e.g., psychosocial stressors) in ecologically valid settings such as home and work. Compared to the survey-based EMA, the TQL-enhanced mobile EMA applications with more contextual information from real-world experience are expected to further reduce recall bias and improve the efficacy of N -of-1 study [51, 52]. Users and field validation studies would be needed to demonstrate the value of adaptive and interpretable feedback for improving patient experience, healthcare quality and care coordination in the care management flow.…”

Section: Discussionmentioning

confidence: 99%

An interpretable health behavioral intervention policy for mobile device users

Hu¹,

Hsueh²,

Chen³

et al. 2018

IBM J. Res. & Dev.

View full text Add to dashboard Cite

An increasing number of people use mobile devices to monitor their behavior, such as exercise, and record their health status, such as psychological stress. However, these devices rarely provide ongoing support to help users understand how their behavior contributes to changes in their health status. To address this challenge, we aim to develop an interpretable policy for physical activity recommendations that reduce a user’s perceived psychological stress, over a given time horizon. We formulate this problem as a sequential decision-making problem and solve it using a new method that we refer to as threshold Q-learning (TQL). The advantage of the TQL method over traditional Q-learning is that it is “doubly robust” and interpretable. This interpretability is achieved by making model assumptions and incorporating threshold selection into the learning process. Our simulation results indicate that the TQL method performs better than the Q-learning method given model misspecification. Our analyses are performed on data collected from 79 healthy adults over a 7 week period, where the data comprise physical activity patterns collected from mobile devices and self-assessed stress levels of the users. This work serves as a first step toward a computational health coaching solution for mobile device users.

show abstract

“…However, even Murphy et al have acknowledged that novel statistical methods are required to distinguish the effects of participant-determined features, dependent on personal agency, from those of routinely provided behavioural change intervention 35. This endogeneity problem is compounded when we acknowledge that awareness of being monitored could itself modify the experience and the behaviour—and how to fully address this in the analysis is not yet clear 37…”

Section: Limitations Of Current Study Designmentioning

confidence: 99%

Scientific challenges for precision public health

Kee

Taylor-Robinson

2020

J Epidemiol Community Health

View full text Add to dashboard Cite

The notion of ‘precision’ public health has been the subject of much debate, with recent articles coming to its defence following the publication of several papers questioning its value.Critics of precision public health raise the following problems and questionable assumptions: the inherent limits of prediction for individuals; the limits of approaches to prevention that rely on individual agency, in particular the potential for these approaches to widen inequalities; the undue emphasis on the supposed new information contained in individuals’ molecules and their ‘big data’ at the expense of their own preferences for a particular intervention strategy and the diversion of resources and attention from the social determinants of health.In order to refocus some of these criticisms of precision public health as scientific questions, this article outlines some of the challenges when defining risk for individuals; the limitations of current theory and study design for precision public health; and the potential for unintended harms.

show abstract

Are Nomothetic or Ideographic Approaches Superior in Predicting Daily Exercise Behaviors?

Cited by 29 publications

References 43 publications

Prediction of stress and drug craving ninety minutes in the future with passively collected GPS data

Prediction of stress and drug craving ninety minutes in the future with passively collected GPS data

An interpretable health behavioral intervention policy for mobile device users

Scientific challenges for precision public health

Contact Info

Product

Resources

About