Cocaine addiction is a highly complex clinical phenotype, one in which largely predictable and highly characteristic pharmacological effects of cocaine use/intoxication (e.g., physiological, behavioral, and subjective) take root in a unique physical, psychological and social person. The latter realities pose a clinical pharmacogenetic challenge with respect to parsing heterogeneity among addicted individuals and their treatment response. Thus, more modern methods of clinical phenotyping and treatment monitoring are required methods that go beyond static, cross- sectional, retrospective assessments of drug use and more holistically, accurately, and dynamically assess the addicted individual in a fine-grained and real-time fashion. With an eye towards developing such methods, the current application brings together two established research teams with complimentary interdisciplinary expertise in human cocaine self-administration and medications development (Malison, Yale) and mobile and pervasive wireless sensing, sensor data analysis/inference, and embedded and networked system technologies (Ganesan, UMass Amherst). Together, they propose to develop """"""""signatures"""""""" (i.e., inference algorithms based on low-level physiological sensor data as derived by Dynamic Bayesian Network analysis) that are both sensitive and specific for detecting cocaine use/intoxication.
Two specific aims are proposed, and include:
Specific Aim 1 : Inpatient human laboratory validation, outpatient real world refinement and exploratory clinical trial piloting of a remote wireless sensor network approach to detecting cocaine use/intoxication in human cocaine addicts (N=24);
and Specific Aim 2 : Designing an algorithm for reliably detecting cocaine in real-world settings, inference techniques for understanding the relationship between cocaine use and other contexts, and a system (deliverable) that can be easily replicated and used by other researchers for field studies (i.e., a toolkit that will include secur web-based configuration tools for study design, real-time data quality analysis, inference tools, tutorials, and cloud storage and computation resources). We believe that such a system has enormous long-term potential for realizing the opportunity instrinsically embedded within the clinical pharmacogenetic challenge, namely, the development of more highly personalized, and in turn, more effective, treatment interventions for individuals addicted to cocaine.
Cocaine addiction is serious public health problem for which better treatments are needed. Advances in wearable on-body sensors and smart phones are dramatically improving medicines'ability to collect more holistic, real-time data on individuals'health. The current application proposes to apply such technology to our understanding of cocaine addiction so that more effective and personalized treatment interventions for the disorder might be developed.
|Natarajan, Annamalai; Angarita, Gustavo; Gaiser, Edward et al. (2016) Domain Adaptation Methods for Improving Lab-to-field Generalization of Cocaine Detection using Wearable ECG. Proc ACM Int Conf Ubiquitous Comput 2016:875-885|
|Adams, Roy J; Saleheen, Nazir; Thomaz, Edison et al. (2016) Hierarchical Span-Based Conditional Random Fields for Labeling and Segmenting Events in Wearable Sensor Data Streams. Proc Int Conf Mach Learn 48:334-343|
|Natarajan, Annamalai; Gaiser, Edward; Angarita, Gustavo et al. (2014) Conditional Random Fields for Morphological Analysis of Wireless ECG Signals. ACM BCB 2014:370-379|
|Versteeg, Gijs A; Rajsbaum, Ricardo; Sanchez-Aparicio, Maria Teresa et al. (2013) The E3-ligase TRIM family of proteins regulates signaling pathways triggered by innate immune pattern-recognition receptors. Immunity 38:384-98|