CP 5: Affective Science and Smoking Cessation: Real-time Real-world Assessment

Collaborating Investigator:

Dr. Cho Lam (PI) & Dr. David Wetter, University of Utah

Funding Status:

R01CA224537

NIH/NCI

1/19/18 – 12/31/22

Associated with:

TR&D1

Publications
Significance
Approach & Push Pull Relationship

TR&D1: Discovery

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

Authors:

Md Azim Ullah, Soujanya Chatterjee, Christopher P Fagundes, Cho Lam, Inbal Nahum-Shani, James M Rehg, David W Wetter, Santosh Kumar

Publication Venue:

ACM on Interactive, Mobile, Wearable, and Ubiquitous Technologies (IMWUT)

Publication Date:

September 7, 2022

Keywords:

behavioral intervention, human-centered computing, risk prediction, smoking cessation, ubiquitous and mobile computing design and evaluation methods, wearable sensors

View Full Paper

Related Projects:

CP 1, CP 5

Estimation of the continuous risk state may be critical for delivering temporally-precise interventions and treatment adaptations in cessation programs. Continuous sensor data collected from wearables and smartphones to capture risk factors of adverse behaviors in the natural environment are usually noisy and incomplete. For adverse behavioral events such as a smoking lapse, capturing the precise timing of each smoking lapse may not be feasible, as sensors may not be worn at the time of a lapse or the lapse events may not be accurately detected due to the imperfection of machine learning models that are used to detect smoking events via hand-to-mouth gestures. Therefore, only a few positive events (i.e., smoking lapse in a cessation attempt) are available. Confirmed negative labels can be assigned to a block of sensor data corresponding to a prediction window only if the entire time period is confirmed to have no high-risk moment. As not all high-risk moments may result in a lapse, labeling a block of sensor data to the negative class is difficult for such events. We addressed each of these challenges in developing the mRisk model. Specifically, we encoded sensor data as events to handle noise and missingness, modeled the historical influence of recent psychological, behavioral, and environmental events via deep learning model and addressed the issue of lack of negative labels and only a small subset of positive labels by using a positive-unlabeled framework with a novel loss function.

Abstract:

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.

TL;DR:

We present a model for identifying ideal moments for intervention by passively detecting risk of an imminent adverse behavior.

Smoking cessation is a cornerstone of cancer risk reduction, as 30% of all cancers are directly attributable to tobacco. Although half of all current smokers attempt to quit each year, over 90% of those quit attempts fail. Numerous conceptual models and robust empirical evidence suggests that affect is a potent determinant of smoking lapse. For example, higher lapse risk and lower rates of long-term abstinence are associated with acute experiences of negative emotions such as anger, sadness, and guilt, while acute experiences of positive emotions such as happiness, pride, and gratitude have the opposite effect. A key factor in abstinence is the self-regulatory capacity (SRC), which is an individual’s ability to exert control over behavior, thoughts, and emotions. It is based on the constellation and temporal dynamics of emotion, context, and other factors. Unfortunately, very little is known about the temporal dynamics and interactions of emotion, context, and SRC in real time under field conditions, as well as how these complex combinations influence lapse risk. This lack of knowledge severely hampers both our conceptual models and our ability to develop just-in-time, adaptive interventions (JITAIs) that effectively target the real-time, real-world self-regulatory mechanisms underlying successful cessation. The three goals of CP5 are to 1) leverage sensor-derived biomarkers to obtain objective real-time assessments of lapse, SRC, and geographic place (i.e. context); 2) integrate self-reports of distinct positive and negative emotions with sensor-derived biomarkers to model the time-varying dynamics of affect and the impact of context on affective experience; and 3) develop predictive models of lapse risk based on affect, SRC, and context which can form the basis for future JITAIs.

While CP1 is investigating methods to engage newly abstinent smokers in self-regulatory intervention using biomarkers of SRC, self-efficacy, stress, and other contexts, CP5 is analyzing the mobile sensor and self-report data collected in its observational study of smokers to develop biomarker of SRC and predictive models of lapse risk from mHealth biomarkers of affect, SRC, and other contexts.

CP5 is recruiting daily smokers interested in quitting who wear the AutoSense chestband and MotionSense wrist sensors for one week prior to quitting and two weeks post-quit. They also carry a study-provided smartphone. From the sensor data, CP5 is obtaining sensor-derived biomarkers of stress, geoexposure, and smoking, and EMA’s provide additional assessment of emotions and context. CP5 will work with Aims 1, 2, and 3 of TR&D1. An initial model for SRC under development by CP5 does not explicitly account for uncertainty in sensor-derived biomarkers or in EMA-based assessments resulting from missingness. A push activity will utilize uncertainty models developed in Aim 1 of TR&D1 to represent and quantify uncertainty in sensor- and EMA-derived biomarkers. In a pull activity, we will jointly characterize the resulting uncertainty in the model for SRC. This will enable researchers to take measurement of uncertainty into account when analyzing the linkage between SRC and distinct emotions, EMA-derived context, sensor-derived states of stress and craving, and geographic context. In addition, uncertainty representation for the derived SRC biomarker can inform downstream analysis of the link between SRC and lapse risk. CP5 will work with TR&D1 Aim 2 to identify timing triggers from affect, context, SRC, and other lapse risks from the collected data, to guide the delivery of temporally-precise interventions. CP5 will utilize tools developed in Aim 2 for identifying specific temporal patterns that predict future lapse risk. Analysis of the effectiveness of these triggers on collected data will lead to a pull activity in which the identified temporal patterns are distilled into a composite risk score which can be computed from sensor data in real-time and characterizes the time-varying risk level. The resulting time triggers and risk scores will be utilized in the MRT which will be deployed by CP1. CP5 will utilize methods from TR&D1 Aim 3 for modeling the dynamic relationships between positive and negative discrete emotions, and biomarkers for stress, craving, and context that inform the temporal variations in SRC and lapse risk. Initially, the models will describe the lead-lag relationships between covariates. When sufficient sensor data has been collected, a pull activity with Aim 3 of TR&D1 will jointly develop deep feature learning methods based on variational autoencoders that can learn nonlinear temporal features to improve the accuracy in predicting the time-varying risk of lapse. This can help discover the optimal content and timing for temporally-precise interventions. The iterations between the two teams in testing the initial model, revising the model, and testing again will eventually lead to methods for all three aims that will be directly applicable to smoking cessation research using mHealth biomarkers.