A Financial Incentives Program to Promote Smoking Cessation Among Recently Hospitalized Individuals: Feasibility and Acceptability Study.

BACKGROUND: Hospitalization is an opportunity to engage underserved individuals in tobacco treatment who may not otherwise have access to it. Tobacco treatment interventions that begin during hospitalization and continue for at least 1 postdischarge month are effective in promoting smoking cessation. However, there is low usage of postdischarge tobacco treatment services. Financial incentives for smoking cessation are an intervention in which participants receive incentives, such as cash payments or vouchers for goods, to encourage individuals to stop smoking or to reward individuals for maintaining abstinence. OBJECTIVE: We sought to determine the feasibility and acceptability of a novel postdischarge financial incentive intervention that uses a smartphone application paired to measurements of exhaled carbon monoxide (CO) concentration levels to promote smoking cessation in individuals who smoke cigarettes. METHODS: We collaborated with Vincere Health, Inc. to tailor their mobile application that uses facial recognition features, a portable breath test CO monitor, and smartphone technology to deliver financial incentives to a participant's digital wallet after the completion of each CO test. The program includes 3 racks. Track 1: Noncontingent incentives for conducting CO tests. Track 2: Combination of noncontingent and contingent incentives for CO levels <10 parts per million (ppm). Track 3: Contingent incentives only for CO levels <10 ppm. After obtaining informed consent, we pilot-tested the program from September to November 2020 with a convenience sample of 33 hospitalized individuals at Boston Medical Center, a large safety-net hospital in New England. Participants received text reminders to conduct CO tests twice daily for 30 days postdischarge. We collected data on engagement, CO levels, and incentives earned. We measured feasibility and acceptability quantitatively and qualitatively at 2 and 4 weeks. RESULTS: Seventy-six percent (25/33) completed the program and 61% (20/33) conducted at least 1 breath test each week. Seven patients had consecutive CO levels <10 ppm during the last 7 days of the program. Engagement with the financial incentive intervention as well as in-treatment abstinence was highest in Track 3 that delivered financial incentives contingent on CO levels <10 ppm. Participants reported high program satisfaction and that the intervention helped motivate smoking cessation. Participants suggested increasing program duration to at least 3 months and adding supplemental text messaging to increase motivation to stop smoking. CONCLUSIONS: Financial incentives paired to measurements of exhaled CO concentration levels is a novel smartphone-based tobacco cessation approach that is feasible and acceptable. Future studies should examine the efficacy of the intervention after it is refined to add a counseling or text-messaging component.

Assessment of IGRT variability for lung SBRT.

PURPOSE: The purpose of this project was to assess factors that may influence variability in the pre-treatment kilovoltage cone beam computed tomography (kV CBCT) image matching process for lung stereotactic body radiation therapy (SBRT). METHODS AND MATERIALS: Pre-treatment CBCT and planning CT data sets of previously-treated lung SBRT patients were gathered and anonymized from four radiotherapy centers in Alberta. Eight radiation therapists (RTTs) and four radiation oncologists (ROs) were recruited from the same four cancer centers for image matching. Identical data sets were provided to each user, but the order of image sets was randomized independently for each user to remove any learning bias. Inter-user variabilities were then investigated as functions of various factors, including image origin (source institution/machine), user's institution (local matching protocol), profession (RTT vs. RO), years of experience and image quality (presence/absence of added noise). RESULTS: Very little variation in image matching between different users was observed. The mean differences from the consensus means for different image sets were less than 1 mm in all directions, and cases that exceeded 3 mm (i.e. clinically significant differences) were extremely rare. Image origin, user's institution, and profession (RTT vs. RO) didn't lead to any meaningful clinical differences, while image quality didn't introduce any statistically significant differences. In addition, no discernible trend was seen between user's experience and deviation from the user mean. Overall, no meaningful differences in inter-user variabilities for the different factors investigated were found in this study. CONCLUSIONS: There appears to be an adequate standardization across the province of Alberta in terms of CBCT image matching process. No clinically significant differences were observed as functions of various factors investigated in this study. Consistency in matching between RTTs and ROs in this study suggests that RTTs do not need systematic RO approval of their lung CBCT match. It should be noted that RTTs at the centers in this study receive comprehensive training in CBCT-based image matching.

Dynamic Clustering Algorithms via Small-Variance Analysis of Markov Chain Mixture Models.

Bayesian nonparametrics are a class of probabilistic models in which the model size is inferred from data. A recently developed methodology in this field is small-variance asymptotic analysis, a mathematical technique for deriving learning algorithms that capture much of the flexibility of Bayesian nonparametric inference algorithms, but are simpler to implement and less computationally expensive. Past work on small-variance analysis of Bayesian nonparametric inference algorithms has exclusively considered batch models trained on a single, static dataset, which are incapable of capturing time evolution in the latent structure of the data. This work presents a small-variance analysis of the maximum a posteriori filtering problem for a temporally varying mixture model with a Markov dependence structure, which captures temporally evolving clusters within a dataset. Two clustering algorithms result from the analysis: D-Means, an iterative clustering algorithm for linearly separable, spherical clusters; and SD-Means, a spectral clustering algorithm derived from a kernelized, relaxed version of the clustering problem. Empirical results from experiments demonstrate the advantages of using D-Means and SD-Means over contemporary clustering algorithms, in terms of both computational cost and clustering accuracy.