Real-Time Learning from an Expert in Deep Recommendation Systems with Application to mHealth for Physical Exercises.

Recommendation systems play an important role in today's digital world. They have found applications in various areas such as music platforms, e.g., Spotify, and movie streaming services, e.g., Netflix. Less research effort has been devoted to physical exercise recommendation systems. Sedentary lifestyles have become the major driver of several diseases as well as healthcare costs. In this paper, we develop a recommendation system to recommend daily exercise activities to users based on their history, profiles and similar users. The developed recommendation system uses a deep recurrent neural network with user-profile attention and temporal attention mechanisms. Moreover, exercise recommendation systems are significantly different from streaming recommendation systems in that we are not able to collect click feedback from the participants in exercise recommendation systems. Thus, we propose a real-time, expert-in-the-loop active learning procedure. The active learner calculates the uncertainty of the recommendation system at each time step for each user and asks an expert for recommendation when the certainty is low. In this paper, we derive the probability distribution function of marginal distance, and use it to determine when to ask experts for feedback. Our experimental results on a mHealth and MovieLens datasets show improved accuracy after incorporating the real-time active learner with the recommendation system.

Toward a Psychology of Deep Reinforcement Learning Agents Using a Cognitive Architecture.

We argue that cognitive models can provide a common ground between human users and deep reinforcement learning (Deep RL) algorithms for purposes of explainable artificial intelligence (AI). Casting both the human and learner as cognitive models provides common mechanisms to compare and understand their underlying decision-making processes. This common grounding allows us to identify divergences and explain the learner's behavior in human understandable terms. We present novel salience techniques that highlight the most relevant features in each model's decision-making, as well as examples of this technique in common training environments such as Starcraft II and an OpenAI gridworld.

Resting-state heart rate variability after stressful events as a measure of stress tolerance among elite performers.

Introduction: A common trait of elite performers is their ability to perform well when stressed by strong emotions such as fear. Developing objective measures of stress response that reliably predict performance under stress could have far-reaching implications in selection and training of elite individuals and teams. Prior data suggests that (i) Heart rate and heart rate variability (HR/HRV) are associated with stress reaction, (ii) Higher basal sympathetic tone prior to stressful events is associated with higher performance, and (iii) Elite performers tend to exhibit greater increase in parasympathetic tone after a stressful event. Methods: The current study assesses the predictive utility of post-stressful event HR/HRV measures, an under-studied time point in HR/HRV research, in the context of military personnel selection. Specifically, we examined the relationship between a comprehensive set of HR/HRV measures and established questionnaires related to stress tolerance, experimental evaluation of executive function during stress induction, and ecologically valid selection assessment data from a week-long Special Operations Forces selection course (N = 30). Results: We found that post-stressful event HR/HRV measures generally had strong correlations with the neuroticism facet of the NEO personality inventory as well as the general and distress facets of the defensive reactivity questionnaire. HR/HRV measures correlated reliably with a change in executive function measured as a decrease in verbal fluency with exposure to a well-validated stressor. Finally, we observed a divergent pattern of correlation among elite and non-elite SOF candidates. Specifically, among elite candidates, parasympathetic nervous system (PNS) measures correlated positively and sympathetic nervous system (SNS) measures correlated negatively with evaluation of stress tolerance by experts and peers. This pattern was not present in non-elite candidates. Discussion: Our findings demonstrate that post-stressful event HR/HRV data provide an objective non-invasive method to measure the recovery and arousal state in direct reaction to the stressful event and can be used as metrics of stress tolerance that could enhance selection of elite individuals and teams.

A computational cognitive model of behaviors and decisions that modulate pandemic transmission: Expectancy-value, attitudes, self-efficacy, and motivational intensity.

We present a computational cognitive model that incorporates and formalizes aspects of theories of individual-level behavior change and present simulations of COVID-19 behavioral response that modulates transmission rates. This formalization includes addressing the psychological constructs of attitudes, self-efficacy, and motivational intensity. The model yields signature phenomena that appear in the oscillating dynamics of mask wearing and the effective reproduction number, as well as the overall increase of rates of mask-wearing in response to awareness of an ongoing pandemic.

Managing evidence in food safety and nutrition.

Evidence ('data') is at the heart of EFSA's 2020 Strategy and is addressed in three of its operational objectives: (1) adopt an open data approach, (2) improve data interoperability to facilitate data exchange, and (3) migrate towards structured scientific data. As the generation and availability of data have increased exponentially in the last decade, potentially providing a much larger evidence base for risk assessments, it is envisaged that the acquisition and management of evidence to support future food safety risk assessments will be a dominant feature of EFSA's future strategy. During the breakout session on 'Managing evidence' of EFSA's third Scientific Conference 'Science, Food, Society', current challenges and future developments were discussed in evidence management applied to food safety risk assessment, accounting for the increased volume of evidence available as well as the increased IT capabilities to access and analyse it. This paper reports on presentations given and discussions held during the session, which were centred around the following three main topics: (1) (big) data availability and (big) data connection, (2) problem formulation and (3) evidence integration.

Leveraging Self-Affirmation to Improve Behavior Change: A Mobile Health App Experiment.

BACKGROUND: mHealth interventions can help to improve the physical well-being of participants. Unfortunately, mHealth interventions often have low adherence and high attrition. One possible way to increase adherence is instructing participants to complete self-affirmation exercises. Self-affirmation exercises have been effective in increasing many types of positive behaviors. However, self-affirmation exercises often involve extensive essay writing, a task that is not easy to complete on mobile platforms. OBJECTIVE: This study aimed to adapt a self-affirmation exercise to a form better suited for delivery through a mobile app targeting healthy eating behaviors, and to test the effect of differing self-affirmation doses on adherence to behavior change goals over time. METHODS: We examined how varied self-affirmation doses affected behavior change in an mHealth app targeting healthy eating that participants used for 28 days. We divided participants into the 4 total conditions using a 2×2 factorial design. The first independent variable was whether the participant received an initial self-affirmation exercise. The second independent variable was whether the participant received ongoing booster self-affirmations throughout the 28-day study. To examine possible mechanisms through which self-affirmation may cause positive behavior change, we analyzed three aspects of self-affirmation effects in our research. First, we analyzed how adherence was affected by self-affirmation exercises. Second, we analyzed whether self-affirmation exercises reduced attrition rates from the app. Third, we examined a model for self-affirmation behavior change. RESULTS: Analysis of 3556 observations from 127 participants indicated that higher doses of self-affirmation resulted in improved adherence to mHealth intervention goals (coefficient 1.42, SE 0.71, P=.04). This increased adherence did not seem to translate to a decrease in participant attrition (P value range .61-.96), although our definition of attrition was conservative. Finally, we examined the mechanisms by which self-affirmation may have affected intentions of behavior change; we built a model of intention (R2=.39, P<.001), but self-affirmation did not directly affect final intentions (P value range .09-.93). CONCLUSIONS: Self-affirmations can successfully increase adherence to recommended diet and health goals in the context of an mHealth app. However, this increase in adherence does not seem to reduce overall attrition. The self-affirmation exercises we developed were simple to implement and had a low cost for both users and developers. While this study focused on an mHealth app for healthy eating, we recommend that other mHealth apps integrate similar self-affirmation exercises to examine effectiveness in other behaviors and contexts.

Implementation Intention and Reminder Effects on Behavior Change in a Mobile Health System: A Predictive Cognitive Model.

BACKGROUND: Implementation intentions are mental representations of simple plans to translate goal intentions into behavior under specific conditions. Studies show implementation intentions can produce moderate to large improvements in behavioral goal achievement. Human associative memory mechanisms have been implicated in the processes by which implementation intentions produce effects. On the basis of the adaptive control of thought-rational (ACT-R) theory of cognition, we hypothesized that the strength of implementation intention effect could be manipulated in predictable ways using reminders delivered by a mobile health (mHealth) app. OBJECTIVE: The aim of this experiment was to manipulate the effects of implementation intentions on daily behavioral goal success in ways predicted by the ACT-R theory concerning mHealth reminder scheduling. METHODS: An incomplete factorial design was used in this mHealth study. All participants were asked to choose a healthy behavior goal associated with eat slowly, walking, or eating more vegetables and were asked to set implementation intentions. N=64 adult participants were in the study for 28 days. Participants were stratified by self-efficacy and assigned to one of two reminder conditions: reminders-presented versus reminders-absent. Self-efficacy and reminder conditions were crossed. Nested within the reminders-presented condition was a crossing of frequency of reminders sent (high, low) by distribution of reminders sent (distributed, massed). Participants in the low frequency condition got 7 reminders over 28 days; those in the high frequency condition were sent 14. Participants in the distributed conditions were sent reminders at uniform intervals. Participants in the massed distribution conditions were sent reminders in clusters. RESULTS: There was a significant overall effect of reminders on achieving a daily behavioral goal (coefficient=2.018, standard error [SE]=0.572, odds ratio [OR]=7.52, 95% CI 0.9037-3.2594, P<.001). As predicted by ACT-R, using default theoretical parameters, there was an interaction of reminder frequency by distribution on daily goal success (coefficient=0.7994, SE=0.2215, OR=2.2242, 95% CI 0.3656-1.2341, P<.001). The total number of times a reminder was acknowledged as received by a participant had a marginal effect on daily goal success (coefficient=0.0694, SE=0.0410, OR=1.0717, 95% CI -0.01116 to 0.1505, P=.09), and the time since acknowledging receipt of a reminder was highly significant (coefficient=-0.0490, SE=0.0104, OR=0.9522, 95% CI -0.0700 to -0.2852], P<.001). A dual system ACT-R mathematical model was fit to individuals' daily goal successes and reminder acknowledgments: a goal-striving system dependent on declarative memory plus a habit-forming system that acquires automatic procedures for performance of behavioral goals. CONCLUSIONS: Computational cognitive theory such as ACT-R can be used to make precise quantitative predictions concerning daily health behavior goal success in response to implementation intentions and the dosing schedules of reminders.

A computational cognitive model of self-efficacy and daily adherence in mHealth.

Mobile health (mHealth) applications provide an excellent opportunity for collecting rich, fine-grained data necessary for understanding and predicting day-to-day health behavior change dynamics. A computational predictive model (ACT-R-DStress) is presented and fit to individual daily adherence in 28-day mHealth exercise programs. The ACT-R-DStress model refines the psychological construct of self-efficacy. To explain and predict the dynamics of self-efficacy and predict individual performance of targeted behaviors, the self-efficacy construct is implemented as a theory-based neurocognitive simulation of the interaction of behavioral goals, memories of past experiences, and behavioral performance.

A Group-Based Mobile Application to Increase Adherence in Exercise and Nutrition Programs: A Factorial Design Feasibility Study.

BACKGROUND: Novel methods of promoting self-monitoring and social support are needed to ensure long-term maintenance of behavior change. In this paper, we directly investigate the effects of group support in an exercise and nutrition program delivered by an mHealth application called Fittle. OBJECTIVE: Our first specific study aim was to explore whether social support improved adherence in wellness programs. Our second specific study aim was to assess whether media types (ePaper vs mobile) were associated with different levels of compliance and adherence to wellness programs. The third aim was to assess whether the use of an mHealth application led to positive changes to participants' eating behavior, physical activity, and stress level, compared to traditional paper-based programs. METHODS: A 2 × 2 (eg, Media: Mobile vs ePaper × Group Type: Team vs Solo) factorial design feasibility study was conducted. A sample of 124 volunteers who were interested in improving eating behavior, increasing physical activity, or reducing stress participated in this study. The study duration was 8 weeks. All groups were self-directed with no ongoing human input from the research team. RESULTS: Participants in ePaper conditions had higher attrition rates compared to participants in Mobile conditions, χ3(2)=9.96, P=.02 (N=124). Participants in Mobile conditions reported their compliance with a much higher frequency closer to the time of challenge activity completion (2-sample Kolmogorov-Smirnov test comparing distributions was highly significant-KS=0.33, P<.001 [N=63]). Participants in ePaper conditions had a much higher frequency of guessing while reporting as compared with those in Mobile conditions-χ1(2)=25.25, P<.001 (N=63). Together, these findings suggest that the mobile app allowed a more accurate method to report and track health behaviors over a longer period than traditional ePaper-based diaries or log books. There was a significant difference in the overall compliance score for Mobile-Solo (Mean [SD] 0.30 [0.39]) and Mobile-Team (Mean [SD] 0.49 [0.35]) conditions (t50.82=1.94, P=.05). This suggests that working in a team increased participants' overall compliance within Fittle. Survival analysis showed that participants assigned to Team conditions are 66% more likely to engage longer with mHealth app-based intervention than those assigned to the Solo condition. Overall, participants across all groups reported some positive changes in eating behavior, physical activity, and stress level; however, participants in the Mobile-Solo condition reported higher perceived stress levels at the end of the study. CONCLUSIONS: The team-based Fittle app is an acceptable and feasible wellness behavior change intervention and a full randomized controlled trial to investigate the efficacy of such an intervention is warranted.

From good intentions to healthy habits: towards integrated computational models of goal striving and habit formation.

Computational models were developed in the ACT-R neurocognitive architecture to address some aspects of the dynamics of behavior change. The simulations aim to address the day-to-day goal achievement data available from mobile health systems. The models refine current psychological theories of self-efficacy, intended effort, and habit formation, and provide an account for the mechanisms by which goal personalization, implementation intentions, and remindings work.

Acceptability of a team-based mobile health (mHealth) application for lifestyle self-management in individuals with chronic illnesses.

With increased incidence of chronic illnesses arising due to unhealthy lifestyle habits, it is increasingly important to leverage technology applications to promote and sustain health behavior change. We developed a smartphone-based application, NutriWalking (NW), which recommends personalized daily exercise goals and promotes healthy nutritional habits in small peer teams. Here, we demonstrate an early study of usability and acceptability of this app in patients with type 2 Diabetes Mellitus and Depression. Our goal was to evaluate the potential of NW as a self-management support tool. Findings point to design considerations for team-based self-management tools delivered via mHealth platforms.

Reinforcement learning and counterfactual reasoning explain adaptive behavior in a changing environment.

Animals routinely adapt to changes in the environment in order to survive. Though reinforcement learning may play a role in such adaptation, it is not clear that it is the only mechanism involved, as it is not well suited to producing rapid, relatively immediate changes in strategies in response to environmental changes. This research proposes that counterfactual reasoning might be an additional mechanism that facilitates change detection. An experiment is conducted in which a task state changes over time and the participants had to detect the changes in order to perform well and gain monetary rewards. A cognitive model is constructed that incorporates reinforcement learning with counterfactual reasoning to help quickly adjust the utility of task strategies in response to changes. The results show that the model can accurately explain human data and that counterfactual reasoning is key to reproducing the various effects observed in this change detection paradigm.

A functional model of sensemaking in a neurocognitive architecture.

Sensemaking is the active process of constructing a meaningful representation (i.e., making sense) of some complex aspect of the world. In relation to intelligence analysis, sensemaking is the act of finding and interpreting relevant facts amongst the sea of incoming reports, images, and intelligence. We present a cognitive model of core information-foraging and hypothesis-updating sensemaking processes applied to complex spatial probability estimation and decision-making tasks. While the model was developed in a hybrid symbolic-statistical cognitive architecture, its correspondence to neural frameworks in terms of both structure and mechanisms provided a direct bridge between rational and neural levels of description. Compared against data from two participant groups, the model correctly predicted both the presence and degree of four biases: confirmation, anchoring and adjustment, representativeness, and probability matching. It also favorably predicted human performance in generating probability distributions across categories, assigning resources based on these distributions, and selecting relevant features given a prior probability distribution. This model provides a constrained theoretical framework describing cognitive biases as arising from three interacting factors: the structure of the task environment, the mechanisms and limitations of the cognitive architecture, and the use of strategies to adapt to the dual constraints of cognition and the environment.

Soccer science and the Bayes community: exploring the cognitive implications of modern scientific communication.

Science is a form of distributed analysis involving both individual work that produces new knowledge and collaborative work to exchange information with the larger community. There are many particular ways in which individual and community can interact in science, and it is difficult to assess how efficient these are, and what the best way might be to support them. This paper reports on a series of experiments in this area and a prototype implementation using a research platform called CACHE. CACHE both supports experimentation with different structures of interaction between individual and community cognition and serves as a prototype for computational support for those structures. We particularly focus on CACHE-BC, the Bayes community version of CACHE, within which the community can break up analytical tasks into "mind-sized" units and use provenance tracking to keep track of the relationship between these units.

Investigating the Roles of Knowledge and Cognitive Abilities in Older Adult Information Seeking on the Web.

This study investigated the influences of knowledge, particularly Internet, Web browser, and search engine knowledge, as well as cognitive abilities on older adult information seeking on the Internet. The emphasis on aspects of cognition was informed by a modeling framework of search engine information-seeking behavior. Participants from two older age groups were recruited: twenty people in a younger-old group (ages 60-70) and twenty people in an older-old group (ages 71-85). Ten younger adults (ages 18-39) served as a comparison group. All participants had at least some Internet search experience. The experimental task consisted of six realistic search problems, all involving information related to health and well-being and which varied in degree of complexity. The results indicated that though necessary, Internet-related knowledge was not sufficient in explaining information-seeking performance, and suggested that a combination of both knowledge and key cognitive abilities is important for successful information seeking. In addition, the cognitive abilities that were found to be critical for task performance depended on the search problem's complexity. Also, significant differences in task performance between the younger and the two older age groups were found on complex, but not on simple problems. Overall, the results from this study have implications for instructing older adults on Internet information seeking and for the design of Web sites.

Rational analyses of information foraging on the web.

This article describes rational analyses and cognitive models of Web users developed within information foraging theory. This is done by following the rational analysis methodology of (a) characterizing the problems posed by the environment, (b) developing rational analyses of behavioral solutions to those problems, and (c) developing cognitive models that approach the realization of those solutions. Navigation choice is modeled as a random utility model that uses spreading activation mechanisms that link proximal cues (information scent) that occur in Web browsers to internal user goals. Web-site leaving is modeled as an ongoing assessment by the Web user of the expected benefits of continuing at a Web site as opposed to going elsewhere. These cost-benefit assessments are also based on spreading activation models of information scent. Evaluations include a computational model of Web user behavior called Scent-Based Navigation and Information Foraging in the ACT Architecture, and the Law of Surfing, which characterizes the empirical distribution of the length of paths of visitors at a Web site.