mCrave: Continuous Estimation of Craving During Smoking Cessation

Classification of Lapses in Smokers Attempting to Stop: A Supervised Machine Learning Approach Using Data From a Popular Smoking Cessation Smartphone App

INTRODUCTION

Smoking lapses after the quit date often lead to full relapse. To inform the development of real time, tailored lapse prevention support, we used observational data from a popular smoking cessation app to develop supervised machine learning algorithms to distinguish lapse from non-lapse reports.

AIMS AND METHODS

We used data from app users with ≥20 unprompted data entries, which included information about craving severity, mood, activity, social context, and lapse incidence. A series of group-level supervised machine learning algorithms (eg, Random Forest, XGBoost) were trained and tested. Their ability to classify lapses for out-of-sample (1) observations and (2) individuals were evaluated. Next, a series of individual-level and hybrid algorithms were trained and tested.

RESULTS

Participants (N = 791) provided 37 002 data entries (7.6% lapses). The best-performing group-level algorithm had an area under the receiver operating characteristic curve (AUC) of 0.969 (95% confidence interval [CI] = 0.961 to 0.978). Its ability to classify lapses for out-of-sample individuals ranged from poor to excellent (AUC = 0.482-1.000). Individual-level algorithms could be constructed for 39/791 participants with sufficient data, with a median AUC of 0.938 (range: 0.518-1.000). Hybrid algorithms could be constructed for 184/791 participants and had a median AUC of 0.825 (range: 0.375-1.000).

CONCLUSIONS

Using unprompted app data appeared feasible for constructing a high-performing group-level lapse classification algorithm but its performance was variable when applied to unseen individuals. Algorithms trained on each individual's dataset, in addition to hybrid algorithms trained on the group plus a proportion of each individual's data, had improved performance but could only be constructed for a minority of participants.

IMPLICATIONS

This study used routinely collected data from a popular smartphone app to train and test a series of supervised machine learning algorithms to distinguish lapse from non-lapse events. Although a high-performing group-level algorithm was developed, it had variable performance when applied to new, unseen individuals. Individual-level and hybrid algorithms had somewhat greater performance but could not be constructed for all participants because of the lack of variability in the outcome measure. Triangulation of results with those from a prompted study design is recommended prior to intervention development, with real-world lapse prediction likely requiring a balance between unprompted and prompted app data.

mRisk: Continuous Risk Estimation for Smoking Lapse from Noisy Sensor Data with Incomplete and Positive-Only Labels

Passive detection of risk factors (that may influence unhealthy or adverse behaviors) via wearable and mobile sensors has created new opportunities to improve the effectiveness of behavioral interventions. A key goal is to find opportune moments for intervention by passively detecting rising risk of an imminent adverse behavior. But, it has been difficult due to substantial noise in the data collected by sensors in the natural environment and a lack of reliable label assignment of low- and high-risk states to the continuous stream of sensor data. In this paper, we propose an event-based encoding of sensor data to reduce the effect of noises and then present an approach to efficiently model the historical influence of recent and past sensor-derived contexts on the likelihood of an adverse behavior. Next, to circumvent the lack of any confirmed negative labels (i.e., time periods with no high-risk moment), and only a few positive labels (i.e., detected adverse behavior), we propose a new loss function. We use 1,012 days of sensor and self-report data collected from 92 participants in a smoking cessation field study to train deep learning models to produce a continuous risk estimate for the likelihood of an impending smoking lapse. The risk dynamics produced by the model show that risk peaks an average of 44 minutes before a lapse. Simulations on field study data show that using our model can create intervention opportunities for 85% of lapses with 5.5 interventions per day.

Translation and psychometric validation of the Persian version of the Questionnaire on Smoking Urges for assessment of craving to smoke among university students

Background

Even though tobacco is one of the most preventable causes of death worldwide, it endangers more than 8 million people yearly. In this context, meta-analyses suggest that a significant part of the general Iranian population over 15 years of age smoke and that there is a need for good screening tools for smoking cravings and urges in Iran. The present study reported the translation and investigated the psychometric properties (i.e., factor structure, validity, and reliability) of the Persian version of the Questionnaire on Smoking Urges (QSU) with 12 items in the Iranian context.

Method

The translation process and content validity of the items were examined entirely in an expert panel using the Content Validity Index. The total sample of participants in which the translated version was tested consisted of 392 (172 female, 220 male, M_age = 22.31 years, SD = 2.90) university students who answered the QSU 12-item at the start of their participation in smoking cessation interventions. The QSU 12-item was firstly translated, then piloted using a subsample of 150 university students and finally validity and reliability of the instrument were investigated using a subsample of 242 participants. We tested the proposed models in the literature, that is, a 1-factor solution and a 2-factor solution with six items on each factor (Factor 1: desire/intention to smoke; Factor 2: relief of negative affect or withdrawal symptoms and anticipation of positive outcome). At last, we tested differences across differences in QSU-scores across different subgroups of individuals based on their demographics.

Results

The results suggested that, in contrast to past studies, a modified 2-factor model, using five items for Factor 1 and 7 items for Factor 2, was the best fitting model (CFI = .95, RMSEA = .09, CI = 90%). Additionally, the QSU 12-item Persian version showed good convergent and divergent validity, internal consistency (Factor 1 = .94, Factor 2 = .97), ICC (average measure ICC = .95, CI = 95%, F_{(391, 4301)} = 20.54, p < .001), concurrent validity (r = .71, p < .01), and discriminant validity (r = −.04, p > .05). Finally, subgroups based on gender, marital status, (un)employment, and educational level did not differed in their responses to the QSU 12-item.

Conclusion

The Persian version of the QSU 12-item has satisfactory psychometric properties and, with a slight modification, it can be considered as a reliable and valid method to estimate smoking urges in the Iranian population. Moreover, the QSU 12-item seems appropriate to measure urge for smoking among groups of individuals with different sociodemographic backgrounds. Importantly, the QSU 12-item differentiates individuals’ desire and intention to smoke from their anticipated relief of negative affect or withdrawal symptoms, which can be important for personalizing interventions targeting individuals who want to quit smoking.

Workshop on Implementation Science and Digital Therapeutics for Behavioral Health

Digital therapeutics can overcome many of the barriers to translation of evidence-based treatment for substance use, mental health, and other behavioral health conditions. Delivered via nearly ubiquitous platforms such as the web, smartphone applications, text messaging, and videoconferencing, digital therapeutics can transcend the time and geographic boundaries of traditional clinical settings so that individuals can access care when and where they need it. There is strong empirical support for digital therapeutic approaches for behavioral health, yet implementation science with regard to scaling use of digital therapeutics for behavioral health is still in its early stages. In this paper, we summarize the proceedings of a day-long workshop, "Implementation Science and Digital Therapeutics," sponsored and hosted by the Center for Technology and Behavioral Health at Dartmouth College. The Center for Technology and Behavioral Health is an interdisciplinary P30 Center of Excellence funded by the National Institute on Drug Abuse, with the mission of promoting state-of-the-technology and state-of-the-science for the development, evaluation, and sustainable implementation of digital therapeutic approaches for substance use and related conditions. Workshop presentations were grounded in current models of implementation science. Directions and opportunities for collaborative implementation science research to promote broad adoption of digital therapeutics for behavioral health are offered.

Low-Burden Mobile Monitoring, Intervention, and Real-Time Analysis Using the Wear-IT Framework: Example and Usability Study

BACKGROUND

Mobile health (mHealth) methods often rely on active input from participants, for example, in the form of self-report questionnaires delivered via web or smartphone, to measure health and behavioral indicators and deliver interventions in everyday life settings. For short-term studies or interventions, these techniques are deployed intensively, causing nontrivial participant burden. For cases where the goal is long-term maintenance, limited infrastructure exists to balance information needs with participant constraints. Yet, the increasing precision of passive sensors such as wearable physiology monitors, smartphone-based location history, and internet-of-things devices, in combination with statistical feature selection and adaptive interventions, have begun to make such things possible.

OBJECTIVE

In this paper, we introduced Wear-IT, a smartphone app and cloud framework intended to begin addressing current limitations by allowing researchers to leverage commodity electronics and real-time decision making to optimize the amount of useful data collected while minimizing participant burden.

METHODS

The Wear-IT framework uses real-time decision making to find more optimal tradeoffs between the utility of data collected and the burden placed on participants. Wear-IT integrates a variety of consumer-grade sensors and provides adaptive, personalized, and low-burden monitoring and intervention. Proof of concept examples are illustrated using artificial data. The results of qualitative interviews with users are provided.

RESULTS

Participants provided positive feedback about the ease of use of studies conducted using the Wear-IT framework. Users expressed positivity about their overall experience with the framework and its utility for balancing burden and excitement about future studies that real-time processing will enable.

CONCLUSIONS

The Wear-IT framework uses a combination of passive monitoring, real-time processing, and adaptive assessment and intervention to provide a balance between high-quality data collection and low participant burden. The framework presents an opportunity to deploy adaptive assessment and intervention designs that use real-time processing and provides a platform to study and overcome the challenges of long-term mHealth intervention.

SmokingOpp: Detecting the Smoking 'Opportunity' Context Using Mobile Sensors

Context plays a key role in impulsive adverse behaviors such as fights, suicide attempts, binge-drinking, and smoking lapse. Several contexts dissuade such behaviors, but some may trigger adverse impulsive behaviors. We define these latter contexts as 'opportunity' contexts, as their passive detection from sensors can be used to deliver context-sensitive interventions. In this paper, we define the general concept of 'opportunity' contexts and apply it to the case of smoking cessation. We operationalize the smoking 'opportunity' context, using self-reported smoking allowance and cigarette availability. We show its clinical utility by establishing its association with smoking occurrences using Granger causality. Next, we mine several informative features from GPS traces, including the novel location context of smoking spots, to develop the SmokingOpp model for automatically detecting the smoking 'opportunity' context. Finally, we train and evaluate the SmokingOpp model using 15 million GPS points and 3,432 self-reports from 90 newly abstinent smokers in a smoking cessation study.

Identifying Smoking Environments From Images of Daily Life With Deep Learning

IMPORTANCE

Environments associated with smoking increase a smoker's craving to smoke and may provoke lapses during a quit attempt. Identifying smoking risk environments from images of a smoker's daily life provides a basis for environment-based interventions.

OBJECTIVE

To apply a deep learning approach to the clinically relevant identification of smoking environments among settings that smokers encounter in daily life.

DESIGN, SETTING, AND PARTICIPANTS

In this cross-sectional study, 4902 images of smoking (n = 2457) and nonsmoking (n = 2445) locations were photographed by 169 smokers from Durham, North Carolina, and Pittsburgh, Pennsylvania, areas from 2010 to 2016. These images were used to develop a probabilistic classifier to predict the location type (smoking or nonsmoking location), thus relating objects and settings in daily environments to established smoking patterns. The classifier combines a deep convolutional neural network with an interpretable logistic regression model and was trained and evaluated via nested cross-validation with participant-wise partitions (ie, out-of-sample prediction). To contextualize model performance, images taken by 25 randomly selected participants were also classified by smoking cessation experts. As secondary validation, craving levels reported by participants when viewing unfamiliar environments were compared with the model's predictions. Data analysis was performed from September 2017 to May 2018.

MAIN OUTCOMES AND MEASURES

Classifier performance (accuracy and area under the receiver operating characteristic curve [AUC]), comparison with 4 smoking cessation experts, contribution of objects and settings to smoking environment status (standardized model coefficients), and correlation with participant-reported craving.

RESULTS

Of 169 participants, 106 (62.7%) were from Durham (53 [50.0%] female; mean [SD] age, 41.4 [12.0] years) and 63 (37.3%) were from Pittsburgh (31 [51.7%] female; mean [SD] age, 35.2 [13.8] years). A total of 4902 images were available for analysis, including 3386 from Durham (mean [SD], 31.9 [1.3] images per participant) and 1516 from Pittsburgh (mean [SD], 24.1 [0.5] images per participant). Images were evenly split between the 2 classes, with 2457 smoking images (50.1%) and 2445 nonsmoking images (49.9%). The final model discriminated smoking vs nonsmoking environments with a mean (SD) AUC of 0.840 (0.024) (accuracy [SD], 76.5% [1.6%]). A model trained only with images from Durham participants effectively classified images from Pittsburgh participants (AUC, 0.757; accuracy, 69.2%), and a model trained only with images from Pittsburgh participants effectively classified images from Durham participants (AUC, 0.821; accuracy, 75.0%), suggesting good generalizability between geographic areas. Only 1 expert's performance was a statistically significant improvement compared with the classifier (α = .05). Median self-reported craving was significantly correlated with model-predicted smoking environment status (ρ = 0.894; P = .003).

CONCLUSIONS AND RELEVANCE

In this study, features of daily environments predicted smoking vs nonsmoking status consistently across participants. The findings suggest that a deep learning approach can identify environments associated with smoking, can predict the probability that any image of daily life represents a smoking environment, and can potentially trigger environment-based interventions. This work demonstrates a framework for predicting how daily environments may influence target behaviors or symptoms that may have broad applications in mental and physical health.

Drinks & Crowds

The design of computational methods to recognize alcohol intake is a relevant problem in ubiquitous computing. While mobile crowdsensing and social media analytics are two current approaches to characterize alcohol consumption in everyday life, the question of how they can be integrated, to examine their relative value as informative of the drinking phenomenon and to exploit their complementarity towards the classification of drinking-related attributes, remains as an open issue. In this paper, we present a comparative study based on five years of Instagram data about alcohol consumption and a 200+ person crowdsensing campaign collected in the same country (Switzerland). Our contributions are two-fold. First, we conduct data analyses that uncover temporal, spatial, and social contextual patterns of alcohol consumption on weekend nights as represented by both crowdsensing and social media. This comparative analysis provides a contextual snapshot of the alcohol drinking practices of urban youth dwellers. Second, we use a machine learning framework to classify individual drinking events according to alcohol and non-alcohol categories, using images features and contextual cues from individual and joint data sources. Our best performing models give an accuracy of 82.3% on alcohol category classification (against a baseline of 48.5%) and 90% on alcohol/non-alcohol classification (against a baseline of 65.9%) using a fusion of image features and contextual cues in this task. Our work uncovers important patterns in drinking behaviour across these two datasets and the results of study are promising towards developing systems that use machine learning for self-monitoring of alcohol consumption.

Automated sensing of daily activity: A new lens into development

Rapidly maturing technologies for sensing and activity recognition can provide unprecedented access to the complex structure daily activity and interaction, promising new insight into the mechanisms by which experience shapes developmental outcomes. Motion data, autonomic activity, and "snippets" of audio and video recordings can be conveniently logged by wearable sensors (Lazer et al., 2009). Machine learning algorithms can process these signals into meaningful markers, from child and parent behavior to outcomes such as depression or teenage drinking. Theoretically motivated aspects of daily activity can be combined and synchronized to examine reciprocal effects between children's behaviors and their environments or internal processes. Captured over longitudinal time, such data provide a new opportunity to study the processes by which individual differences emerge and stabilize. This paper introduces the reader to developments in sensing and activity recognition with implications for developmental phenomena across the lifespan, sketching a framework for leveraging mobile sensors for transactional analyses that bridge micro- and longitudinal- timescales of development. It finishes by detailing resources and best practices to facilitate the next generation of developmentalists to contribute to this emerging area.

Predicting Smoking Events with a Time-Varying Semi-Parametric Hawkes Process Model

Health risks from cigarette smoking - the leading cause of preventable death in the United States - can be substantially reduced by quitting. Although most smokers are motivated to quit, the majority of quit attempts fail. A number of studies have explored the role of self-reported symptoms, physiologic measurements, and environmental context on smoking risk, but less work has focused on the temporal dynamics of smoking events, including daily patterns and related nicotine effects. In this work, we examine these dynamics and improve risk prediction by modeling smoking as a self-triggering process, in which previous smoking events modify current risk. Specifically, we fit smoking events self-reported by 42 smokers to a time-varying semi-parametric Hawkes process (TV-SPHP) developed for this purpose. Results show that the TV-SPHP achieves superior prediction performance compared to related and existing models, with the incorporation of time-varying predictors having greatest benefit over longer prediction windows. Moreover, the impact function illustrates previously unknown temporal dynamics of smoking, with possible connections to nicotine metabolism to be explored in future work through a randomized study design. By more effectively predicting smoking events and exploring a self-triggering component of smoking risk, this work supports development of novel or improved cessation interventions that aim to reduce death from smoking.

Combining ecological momentary assessment with objective, ambulatory measures of behavior and physiology in substance-use research

Whereas substance-use researchers have long combined self-report with objective measures of behavior and physiology inside the laboratory, developments in mobile/wearable electronic technology are increasingly allowing for the collection of both subjective and objective information in participants' daily lives. For self-report, ecological momentary assessment (EMA), as implemented on contemporary smartphones or personal digital assistants, can provide researchers with near-real-time information on participants' behavior and mood in their natural environments. Data from portable/wearable electronic sensors measuring participants' internal and external environments can be combined with EMA (e.g., by timestamps recorded on questionnaires) to provide objective information useful in determining the momentary context of behavior and mood and/or validating participants' self-reports. Here, we review three objective ambulatory monitoring techniques that have been combined with EMA, with a focus on detecting drug use and/or measuring the behavioral or physiological correlates of mental events (i.e., emotions, cognitions): (1) collection and processing of biological samples in the field to measure drug use or participants' physiological activity (e.g., hypothalamic-pituitary-adrenal axis activity); (2) global positioning system (GPS) location information to link environmental characteristics (disorder/disadvantage, retail drug outlets) to drug use and affect; (3) ambulatory electronic physiological monitoring (e.g., electrocardiography) to detect drug use and mental events, as advances in machine learning algorithms make it possible to distinguish target changes from confounds (e.g., physical activity). Finally, we consider several other mobile/wearable technologies that hold promise to be combined with EMA, as well as potential challenges faced by researchers working with multiple mobile/wearable technologies simultaneously in the field.