Decades of decreasing cigarette smoking rates have stalled, and one in five U.S. adults continue to be active smokers (Dube et al., 2009). This high rate is due, in part, to the limited efficacy of existing smoking cessation treatments-for which maximum successful quit rates are around 35% (Fish et al., 2007). One approach that may improve success rates is personalizing treatment through adaptive interventions. Contrasting traditional treatments, these individualized and time-varying interventions tailor the type and dosage of a therapy to an individual's needs (Collins et al., 2004). For example, an abstaining smoker's current craving level may warrant medication dosage adjustment. Rapid and effective treatment adaptation, however, requires a comprehensive understanding of how intervention components affect the magnitude, speed, and shape of the resulting behavior change. Because traditional intervention models are static, their utility in the adaptive intervention design is liited (Riley et al., 2011). In this project, we use a novel engineering approach to quantify the dynamic features of smoking behavior change, and then connect these comprehensive descriptions of behavior change to design of optimized adaptive smoking interventions. First, we draw from dynamical systems modeling and system identification techniques, typically used by engineers in industrial settings, to model the dynamic behavioral relationships associated with quit success. To do so, we rely on dynamic models of behavior change according to mediational and self-regulatory mechanisms, which were developed previously by this group (Timms et al., 2012a), and intensive longitudinal data (ILD), behavioral data collected at frequent intervals over time, collected in a University of Wisconsin smoking cessation clinical trial of bupropion SR and counseling (Collins et al., 2006;McCarthy et al., 2008a). Because of this, we will also observe how bupropion and counseling affect the dynamic behavioral relationships following a quit attempt. Second, we extend the dynamic models developed in this group's preliminary work to be able to explicitly account for sources of inter- and intra-individual variability in their descriptions of dynamic behavior. In a case study of these expanded models, gene-environment interactions will be examined as a source of such variability. Finally, we will connect the comprehensive models of smoking behavior change produced in this study to actual intervention design. In doing so, we utilize control systems engineering principles, which offer methods to regulate a dynamical system in a practical manner;adhering to clinical constraints while maximizing resource utilization and minimizing resource waste (Rivera et al., 2007). This research will produce an algorithm that tailors intervention components, such as bupropion dosage and counseling frequency, to individual needs in real-time and in an optimal manner. Altogether, this project lays the conceptual and computational foundation for use of a novel engineering approach in smoker patient care.
This study will produce comprehensive descriptions of the process of smoking behavior change during cessation, quantifying the dynamic features of this behavior change and the effect of two common smoking cessation treatments on these dynamics. This project also lays the conceptual and computational foundation for use of a novel engineering approach in design of an optimized smoking intervention, which will have transformative implications to substance abuse healthcare more generally.
|Timms, Kevin P; Rivera, Daniel E; Collins, Linda M et al. (2014) A dynamical systems approach to understanding self-regulation in smoking cessation behavior change. Nicotine Tob Res 16 Suppl 2:S159-68|
|Riley, William T; Martin, Cesar A; Rivera, Daniel E (2014) The importance of behavior theory in control system modeling of physical activity sensor data. Conf Proc IEEE Eng Med Biol Soc 2014:6880-3|
|Timms, Kevin P; Rivera, Daniel E; Piper, Megan E et al. (2014) A Hybrid Model Predictive Control Strategy for Optimizing a Smoking Cessation Intervention. Proc Am Control Conf 2014:2389-2394|
|Timms, Kevin P; Martin, Cesar A; Rivera, Daniel E et al. (2014) Leveraging intensive longitudinal data to better understand health behaviors. Conf Proc IEEE Eng Med Biol Soc 2014:6888-91|
|Timms, Kevin P; Rivera, Daniel E; Collins, Linda M et al. (2014) Continuous-Time System Identification of a Smoking Cessation Intervention. Int J Control 87:1423-1437|
|Shiyko, Mariya P; Burkhalter, Jack; Li, Runze et al. (2014) Modeling nonlinear time-dependent treatment effects: an application of the generalized time-varying effect model (TVEM). J Consult Clin Psychol 82:760-72|
|Timms, Kevin P; Rivera, Daniel E; Collins, Linda M et al. (2013) Control Systems Engineering for Understanding and Optimizing Smoking Cessation Interventions. Proc Am Control Conf :1964-1969|