Misha Pavel
VerifiedNortheastern University · Electrical and Energy Engineering
Active 1978–2026
About
Misha Pavel holds a joint faculty appointment in the College of Computer & Information Science and Bouvé College of Health Sciences at Northeastern University. His background comprises electrical engineering, computer science, and experimental psychology. His research is focused on multiscale computational modeling of behaviors and their control, with applications ranging from elder care to augmentation of human performance. Professor Pavel is using model-based approaches to develop algorithms that transform unobtrusive monitoring from smart homes and mobile devices into useful and actionable knowledge for diagnosis and intervention. Under the auspices of the Northeastern-based Consortium on Technology for Proactive Care, he and his colleagues are targeting technological innovations to support the development of economically feasible, proactive, distributed, and individual-centered healthcare. His previous roles include director of the Smart and Connected Health Program at the National Science Foundation, chair of the Department of Biomedical Engineering at Oregon Health & Science University, and positions at AT&T Laboratories, Bell Laboratories, Stanford University, and New York University.
Research topics
- Computer Science
- Psychology
- Medicine
- Human–computer interaction
- Neuroscience
- Physical medicine and rehabilitation
- Applied psychology
- Artificial Intelligence
- Audiology
- World Wide Web
- Cognitive psychology
- Surgery
- Data science
- Knowledge management
- Cognitive science
- Nursing
- Multimedia
- Social psychology
- Pharmacology
- Database
Selected publications
OSF Preprints (OSF Preprints) · 2026-03-27
preprintOpen access1st authorCorrespondingPopular relapse prevention theories are represented using natural language descriptions and lack temporal information about how phenomena of interest (i.e., ‘relapse’, ‘prolapse’, ‘abstinence’) are dynamically caused over time and within individuals. We drew on the Theory Construction Methodology to develop a formal and computational model of relapse in smoking cessation. We used a participatory, iterative, multi-method approach involving an informal theory and computational model review, stakeholder interviews with researchers, people with lived experience, stop smoking practitioners, and policymakers (N = 15) and in silico simulations. We propose an initial within-person system dynamics model of relapse (‘COMPLAPSE’) in which biopsychosocial factors (e.g., stressors, cigarette cues, cravings, self-efficacy) are represented as time-varying inputs and state variables. These factors jointly determine the momentary preference for each behavioural option (i.e., smoke a cigarette, use a regulatory strategy, do nothing), with the probability of selecting each option (i.e., the output) generated by a softmax function. The simulations highlight the model’s ability to generate representational patterns of relapse, prolapse and abstinence, thus providing an early sense-check of its explanatory adequacy. In addition, local sensitivity analyses demonstrate that systematic variation of selected model parameters leads to expected qualitative shifts from, for example, prolapse to relapse. We discuss the implications of our work for relapse prevention theories and real-world applications, including the development and optimisation of technology-mediated just-in-time adaptive interventions for relapse prevention in smoking cessation.
Scientific Reports · 2026-01-10
articleOpen accessGrowing variable renewable energy and electrification of heating and transportation are intensifying the challenge of operating the electric grid. However, current demand response (DR) approaches compromise their efficacy by neglecting human-building interactions (HBIs). For example, utilities may increase thermostat setpoints on the hottest days of the year, reducing the strain on the grid but making occupants uncomfortable and frustrated. To better understand HBIs in residential buildings, 41 people in 20 homes in two climates participated in a 6-month study. Timestamps from app-based thermal comfort surveys and thermostat interactions were synchronized to time-series building systems data, resulting in the largest-of-its-kind HBI dataset. These survey data are compared to predictions from industry-standard thermal comfort models. Our analysis found that these models, developed under steady-state assumptions, tend to yield greater error magnitudes and/or biases when spatiotemporal temperature variations exceed 2°F, with several comparisons reaching statistical significance. The mean spatial variation within homes in the dataset was 4°F. Thermostat DR control would commonly exacerbate such temporal variation. The results highlight opportunities for improving DR load-control algorithms through a paradigm shift to modeling discomfort rather than comfort, increasing the use of low-cost sensors, and incorporating dynamic models of occupant behavior.
2025-12-12
articleOpen accessPopular relapse prevention theories are represented using natural language descriptions and lack temporal information about how phenomena of interest (i.e., ‘relapse’, ‘prolapse’, ‘abstinence’) are dynamically caused over time and within individuals. We drew on the Theory Construction Methodology to develop a formal and computational model of relapse in smoking cessation. We used a participatory, iterative, multi-method approach involving an informal theory and computational model review, stakeholder interviews with researchers, people with lived experience, stop smoking practitioners, and policymakers (N = 15) and in silico simulations. We propose an initial within-person system dynamics model of relapse (‘COMPLAPSE’) in which biopsychosocial factors (e.g., stressors, cigarette cues, cravings, self-efficacy) are represented as time-varying inputs and state variables. These factors jointly determine the momentary preference for each behavioural option (i.e., smoke a cigarette, use a regulatory strategy, do nothing), with the probability of selecting each option (i.e., the output) generated by a softmax function. The simulations highlight the model’s ability to generate representational patterns of relapse, prolapse and abstinence, thus providing an early sense-check of its explanatory adequacy. In addition, local sensitivity analyses demonstrate that systematic variation of selected model parameters leads to expected qualitative shifts from, for example, prolapse to relapse. We discuss the implications of our work for relapse prevention theories and real-world applications, including the development and optimisation of technology-mediated just-in-time adaptive interventions for relapse prevention in smoking cessation.
2025-04-02
preprintOpen accessBackground: Among smokers attempting to stop, lapses are common and typically lead to a return to regular smoking (‘relapse’). Under certain conditions, however, the person bounces back despite a few lapses (‘prolapse’), or lapses are avoided altogether (‘abstinence’). These well-established phenomena notwithstanding, addiction theories do not clearly articulate temporal information about how they are dynamically caused over time, within individuals. We used formal and computational modelling to develop an initial dynamic model of smoking lapse and relapse.Methods: Drawing on the Theory Construction Methodology, we used a participatory, iterative, multi-method approach, including: i) an informal theory review of popular lapse and relapse theories; ii) two linked stakeholder interviews with researchers, people with lived experience, stop smoking practitioners, and policymakers (N = 15); and iii) formal and computational modelling, including a series of in silico simulations. Results: We propose an initial dynamic model of smoking lapse and relapse (‘COMPLAPSE’) in which biopsychosocial factors (e.g., stressors, cigarette cues, the perceived permissibility of smoking, cravings) jointly contribute to the estimation of a momentary strategy preference, which, through a probabilistic function, leads to the selection of the behavioural option with the greatest value. Through leveraging formal and computational modelling, COMPLAPSE articulates the expected time scales and the relative weights of the variables which jointly influence the lapse and relapse process.Conclusions: We used a participatory, iterative, multi-method approach to develop an initial dynamic model of smoking lapse and relapse. We illustrate the model development process and discuss the next steps.
2025-12-12
articleOpen accessPopular relapse prevention theories are represented using natural language descriptions and lack temporal information about how phenomena of interest (i.e., ‘relapse’, ‘prolapse’, ‘abstinence’) are dynamically caused over time and within individuals. We drew on the Theory Construction Methodology to develop a formal and computational model of relapse in smoking cessation. We used a participatory, iterative, multi-method approach involving an informal theory and computational model review, stakeholder interviews with researchers, people with lived experience, stop smoking practitioners, and policymakers (N = 15) and in silico simulations. We propose an initial within-person system dynamics model of relapse (‘COMPLAPSE’) in which biopsychosocial factors (e.g., stressors, cigarette cues, cravings, self-efficacy) are represented as time-varying inputs and state variables. These factors jointly determine the momentary preference for each behavioural option (i.e., smoke a cigarette, use a regulatory strategy, do nothing), with the probability of selecting each option (i.e., the output) generated by a softmax function. The simulations highlight the model’s ability to generate representational patterns of relapse, prolapse and abstinence, thus providing an early sense-check of its explanatory adequacy. In addition, local sensitivity analyses demonstrate that systematic variation of selected model parameters leads to expected qualitative shifts from, for example, prolapse to relapse. We discuss the implications of our work for relapse prevention theories and real-world applications, including the development and optimisation of technology-mediated just-in-time adaptive interventions for relapse prevention in smoking cessation.
The temporal dynamics of the association between daily physical activity and life satisfaction
Refubium (Universitätsbibliothek der Freien Universität Berlin) · 2025-01-01
articleOpen accessSenior authorPurpose Life satisfaction (LS) is increasingly recognized as a crucial indicator and predictor of health and well-being across the lifespan. The impact of LS may be enhanced through physical activity (PA), although studies exploring the dynamic and bidirectional nature of the relationship are scarce. One principal goal of this project is to examine the dynamic, personalized interactions between LS and PA and exercise identity (the degree to which exercise is a fundamental aspect of one’s self-concept) in geographic areas with different air pollution loads. Method We used data from a 12-month prospective cohort study (N = 1314, mean age = 38.09 [12.55]; range 18-65) with four 2-week intensive measurement bursts to evaluate the bidirectional relationship between LS (assessed at the end of the day) and PA (assessed by Fitbit Charge 3 or 4 throughout the day). The sample included both active (runners; n = 747, 57%) and inactive (n = 567, 43%) individuals living in Moravia-Silesia and South Bohemia, geographic areas with different levels of air pollution. A dynamic Bayesian model based on an extension of the vector autoregressive model was used to estimate both lagged and contemporaneous associations between LS and PA. Results There were meaningful autoregressive effects of first order for both LS (β = 0.394) and PA (β = 0.316), and a within-person contemporaneous association between LS and PA (β = 0.087) that was also associated with temporal factors and trends (weekly and monthly seasonal variation, day in study), gender, age, and exercise identity. Conclusion This study highlights the importance of periodicity on 2 temporal scales for both PA and LS, with age and gender also playing crucial roles. The findings underscore the importance of tailored, context-aware interventions to sustain engagement and enhance well-being through PA.
The temporal dynamics of the association between daily physical activity and life satisfaction
Annals of Behavioral Medicine · 2025-01-01 · 1 citations
articleOpen accessSenior authorPURPOSE: Life satisfaction (LS) is increasingly recognized as a crucial indicator and predictor of health and well-being across the lifespan. The impact of LS may be enhanced through physical activity (PA), although studies exploring the dynamic and bidirectional nature of the relationship are scarce. One principal goal of this project is to examine the dynamic, personalized interactions between LS and PA and exercise identity (the degree to which exercise is a fundamental aspect of one's self-concept) in geographic areas with different air pollution loads. METHOD: We used data from a 12-month prospective cohort study (N = 1314, mean age = 38.09 [12.55]; range 18-65) with four 2-week intensive measurement bursts to evaluate the bidirectional relationship between LS (assessed at the end of the day) and PA (assessed by Fitbit Charge 3 or 4 throughout the day). The sample included both active (runners; n = 747, 57%) and inactive (n = 567, 43%) individuals living in Moravia-Silesia and South Bohemia, geographic areas with different levels of air pollution. A dynamic Bayesian model based on an extension of the vector autoregressive model was used to estimate both lagged and contemporaneous associations between LS and PA. RESULTS: There were meaningful autoregressive effects of first order for both LS (β = 0.394) and PA (β = 0.316), and a within-person contemporaneous association between LS and PA (β = 0.087) that was also associated with temporal factors and trends (weekly and monthly seasonal variation, day in study), gender, age, and exercise identity. CONCLUSION: This study highlights the importance of periodicity on 2 temporal scales for both PA and LS, with age and gender also playing crucial roles. The findings underscore the importance of tailored, context-aware interventions to sustain engagement and enhance well-being through PA.
Control Engineering Practice · 2025-06-29
articleHealth Psychology and Behavioral Medicine · 2025-09-18
articleOpen accessINTRODUCTION: Mobile health (mHealth) technologies such as wearable activity trackers (e.g. Fitbit) and digital applications (apps), can support behavior change in real-world contexts. Since effectiveness is dependent, in part, on participants' engagement with the digital technology (e.g. app page views) and the intervention components (e.g. anti-sedentary messages), there is a need for modeling approaches that support the investigation of engagement in digital interventions and the refinement of dynamic theories of behavior change. METHODS: Dynamic Bayesian Networks (DBN) were used to model the idiographic (individual) dynamic relationships between a participant's daily app engagement (page views), walking behavior, and intervention messages, accounting for context (e.g. temperature), and psychological variables (e.g. perceived restedness and perceived busyness). Additionally, we explored differences in the resulting DBN models between participants of Hispanic/Latino and non-Hispanic/Latino White backgrounds. RESULTS: Data from 10 participants in the HeartSteps II study (n = 5 Hispanic/Latinos and n = 5 non-Hispanic/Latino Whites) was used. Across participants (100%, n = 10), there was a strong positive effect of the number of messages/prompts received on their daily app page views with a predicted increase range of 12.84 (12.19-13.57) to 25.84 (24.28-27.59) app page views per day per message received. Among the majority of Hispanic/Latino participants (n = 4/5, 80%), there was a strong positive relationship between daily app page views and walking behavior with predictions ranging from a mean of 6.70 (6.37-7.05) to 10.93 (10.14-11.78) steps per minute of Fitbit wear time per app page view. Both groups showed idiographic differences in the effects of temperature and perceived busyness on walking behavior. CONCLUSION: The results demonstrate the benefits of DBNs to model the daily-level idiographic behavioral dynamics of engagement in digital intervention studies. This approach can be leveraged to support the refinement of dynamic theories of behavior change and improving personalized mHealth intervention strategies.
2025-10-15
reportOpen accessThis project developed a publicly available, high-fidelity dataset about the interactions among humans, homes, and heat pumps supporting grid interactive efficient buildings to balance demand on the grid with comfort for occupants. Laboratory measurements and simulations of the hardware capture the second-scale electric power dynamics of heat pumps providing grid services like load shifting and load shedding. Field measurements, behavior tracking, and qualitative surveys of people in their homes over multiple years—including experimentally adjusting the heating and cooling system to provide grid services—to capture the reciprocal effect of human behavior on grid services, and grid services on human comfort. Taken together, these data capture the complete Hardware and Human in the loop system for residential heat pumps, reducing large uncertainties in simulation for design, and models for control of heat pumps, and grid-interactive buildings.
Recent grants
NIH · $2.9M · 2019
Nonparametric Nonlinear Adaptive Detection and Estimation
NSF · $240k · 2006–2009
Frequent coauthors
- 69 shared
Holly Jimison
Northeastern University
- 35 shared
Deniz Erdoğmuş
Northeastern University
- 32 shared
Santosh Mathan
Honeywell (United States)
- 31 shared
Tamara L. Hayes
- 25 shared
Jeffrey Kaye
- 18 shared
Hynek Heřmanský
Johns Hopkins University
- 16 shared
Emiliano Santarnecchi
- 16 shared
Álvaro Pascual‐Leone
Hebrew SeniorLife
Awards & honors
- NSF-JST Planning Grant in Support of Super-Aging Societies (…
- NSF CAREER Award (2018)
- Resume-aware match score
- Save to shortlist
- AI-drafted outreach
See your match with Misha Pavel
PhdFit ranks faculty by your research interests, methods, and publications — grounded in their actual work, not templates.
- Free to start
- No credit card
- 30-second signup