AI decision-support tools typically offer a fixed type of assistance, like AI recommendations and explanations, regardless of the specific decision, individual, or broader context. This fixed design has been shown to hinder both human-AI decision accuracy and human skill improvement in the task. We posit that AI assistance needs to be dynamic, changing in response to contextual factors (e.g., AI uncertainty, task difficulty), individual differences, and specified objectives (e.g., decision accuracy, skill improvement). To enable such adaptive support, we propose Reinforcement Learning (RL) as a general approach for modeling human-AI decision-making to optimize human-AI interaction for diverse objectives. RL enables optimizing various objectives in AI-assisted decision-making by tailoring and adaptively providing decision support to humans—the right type of assistance, to the right person, at the right time. We instantiated our approach with two objectives: human-AI accuracy on the decision-making task and human skill improvement (i.e., learning about the task) and learned decision support policies from previous human-AI interaction data. We compared the optimized policies against several baselines in AI-assisted decision-making. Across two experiments (N = 316 and N = 964), our results consistently demonstrated that people interacting with policies optimized for accuracy achieve significantly higher accuracy—and even human-AI complementarity—compared to those interacting with any other type of AI support. Our results further indicated that human learning was more difficult to optimize than accuracy. While the policies learned the best available actions to optimize learning, participants who interacted with learning-optimized policies showed significant learning improvement only at times. Our research (1) demonstrates offline RL to be a promising approach to model the dynamics of human-AI decision-making, leading to policies that may optimize various objectives and provide novel insights about the AI-assisted decision-making space, and (2) emphasizes the importance of considering skill improvement and other human-centric objectives beyond accuracy in AI-assisted decision-making, opening up the novel research challenge of optimizing human-AI interaction for such objectives.

A central challenge for digital health interventions aimed at improving habitual behaviors is deciding when to deliver an intervention prompt. For many daily habits, such as tooth brushing or eating, individuals tend to act around a usual time of day, but this timing is not fixed and can shift as routines evolve. When intervention timing is selected in advance and held constant throughout a study, it can gradually become misaligned with behavior, causing interventions to potentially arrive after the behavior has already occurred or too early to be effective. In this work, we address this habitual timing misalignment in digital health interventions by proposing an online decision-making framework that continuously adapts intervention timing as individual behavior patterns change. Rather than treating intervention timing as a static design choice, our framework adapts it over time and integrates it into a sequential process that determines both when and whether to deliver an intervention. Using data from a deployed oral health intervention trial as a case study, we evaluate our approach using both observed data and simulated settings to assess how well different intervention timing strategies align with the timing of brushing events. Across these evaluations, we measure performance using a coverage-based metric that captures whether an intervention is delivered sufficiently close to a subsequent brushing event. We find that adaptive intervention timing consistently improves coverage compared to fixed intervention times based on user-provided input. The proposed framework is currently deployed in an ongoing randomized controlled trial of a digital oral health intervention, with preliminary results that are consistent with and further support our prior evaluations.

Artificial intelligence systems increasingly shape how people think, feel, and act, particularly in high-impact domains such as healthcare, education, and social support. While significant attention has been paid to technical performance, bias, privacy, and security, the behavioural impacts of AI systems remain under-evaluated and under-governed. This Open Letter outlines why behavioural science should be treated as a core component of responsible AI practice. Even when technically accurate, AI systems can influence behaviour in ways that are harmful, misleading, or misaligned with people’s interests. These effects are often predictable consequences of repeated human–AI interaction, yet they are rarely assessed systematically. We describe where behavioural risks arise in practice and outline what good practice looks like, drawing on established behavioural science methods. We then propose practical steps for funders, researchers, and developers to embed behavioural science expertise and behavioural evaluation across the AI lifecycle, including behavioural testing pipelines, longitudinal evaluation, and interdisciplinary governance. Addressing behavioural impacts is essential to ensuring that AI systems are not only effective, but safe, trustworthy, and appropriate for real-world use.

Mobile-health interventions increasingly use online learning and decision making algorithms to personalize when to nudge users toward healthier behavior, but a poorly designed algorithm can burden and disengage participants. New algorithm design decisions should therefore be vetted against realistic simulated users before each real-life deployment. We propose a method to develop “JITAI-Twins”: digital twins of a target subpopulation for comparing candidate online algorithms before a just-in-time adaptive intervention (JITAI) deployment. The method builds on a conditional time-series diffusion model that is temporally consistent (future actions do not affect the generated past), and it supports repeated updating from three sources of information, in three steps: pre-training on a large observational dataset, fine-tuning on small prior intervention deployments in related populations, and inference-time calibration to the next target population from domain-scientist expertise. We validate the twin at each pre-deployment stage of the long-running HeartSteps series (v2 through v4) of physical-activity suggestion intervention deployments, treating each successive deployment as an upcoming study. The proposed method reproduces the target subpopulation's temporal and between-participant structure better than simpler simulators. These results suggest that our twin can be used to simulate a target deployment before it runs, the prerequisite for testing and informing online algorithm design decisions.

Monitoring just-in-time adaptive interventions (JITAIs) is important both during trialing and when the intervention is deployed in a broader healthcare program. While there is increasing interest in using artificial intelligence (AI) algorithms in JITAIs, these algorithms introduce additional complexity that requires additional monitoring. In this paper, we provide guidelines for monitoring online AI decision-making algorithms. Our guidelines include: (1) identifying potential issues, categorizing them by severity (red, yellow, and green), and (2) developing fallback methods (pre-specified procedures that are executed when an issue occurs). To make ideas concrete, we discuss algorithm monitoring systems in two case studies. In both, the monitoring systems detected real-time issues, and fallback methods both safeguarded participants and ensured quality data for post-deployment data analysis to further refine the JITAI. These guidelines and findings give teams the confidence to include online AI decision-making algorithms in JITAIs.

Timely decision making is critical to the effectiveness of mobile health (mHealth) interventions. At predefined timepoints called "decision points," intelligent mHealth systems such as just-in-time adaptive interventions (JITAIs) estimate an individual's biobehavioral context from sensor or survey data and determine whether and how to intervene. For interventions targeting habitual behavior (e.g., oral hygiene), effectiveness often hinges on delivering support shortly before the target behavior is likely to occur. Current practice schedules decision points at a fixed interval (e.g., one hour) before user-provided behavior times, and the fixed interval is kept the same for all individuals. However, this one-size-fits-all approach performs poorly for individuals with irregular routines, often scheduling decision points after the target behavior has already occurred, rendering interventions ineffective. In this paper, we propose SigmaScheduling, a method to dynamically schedule decision points based on uncertainty in predicted behavior times. When behavior timing is more predictable, SigmaScheduling schedules decision points closer to the predicted behavior time; when timing is less certain, SigmaScheduling schedules decision points earlier, increasing the likelihood of timely intervention. We evaluated SigmaScheduling using real-world data from 68 participants in a 10-week trial of Oralytics, a JITAI designed to improve daily toothbrushing. SigmaScheduling increased the likelihood that decision points preceded brushing events in at least 70% of cases, preserving opportunities to intervene and impact behavior. Our results indicate that SigmaScheduling can advance precision mHealth, particularly for JITAIs targeting time-sensitive, habitual behaviors such as oral hygiene or dietary habits.

Adaptive treatment assignment algorithms, such as bandit algorithms, are increasingly used in digital health intervention clinical trials. Frequently the data collected from these trials is used to conduct causal inference and related data analyses to decide how to refine the intervention, and whether to roll-out the intervention more broadly. This work studies inference for estimands that depend on the adaptive algorithm itself; a simple example is the mean reward under the adaptive algorithm. Specifically, we investigate the replicability of statistical analyses concerning such estimands when using data from trials deploying adaptive treatment assignment algorithms. We demonstrate that many standard statistical estimators can be inconsistent and fail to be replicable across repetitions of the clinical trial, even as the sample size grows large. We show that this non-replicability is intimately related to properties of the adaptive algorithm itself. We introduce a formal definition of a "replicable bandit algorithm" and prove that under such algorithms, a wide variety of common statistical estimators are guaranteed to be consistent and asymptotically normal. We present both theoretical results and simulation studies based on a mobile health oral health self-care intervention. Our findings underscore the importance of designing adaptive algorithms with replicability in mind, especially for settings like digital health where deployment decisions rely heavily on replicated evidence. We conclude by discussing open questions on the connections between algorithm design, statistical inference, and experimental replicability.

In the context of digital health, just-in-time adaptive interventions (JITAIs) are nascent precision medicine systems that can extend personalized health care support to everyday life. A challenge in designing JITAIs is that personalized support often involves sophisticated decision-making algorithms. These decision-making algorithms can require numerous nontrivial design decisions that must be made between successive JITAI deployments (eg, hyperparameter selection for an artificial intelligence algorithm). Making design decisions between deployments—rather than during deployment—ensures intervention fidelity and enhances the ability to replicate results. Yet, each deployment can be costly, precluding the use of A/B testing for every design decision. How should design decisions be made strategically between JITAI deployments? This paper introduces “digital twins for just-in-time adaptive interventions (JITAI-Twins)” to address this question. JITAI-Twins are “digital twins of a subpopulation” (term used in the 2023 National Academies workshop proceedings on digital twins). JITAI-Twins are used to virtually simulate the potential outcomes of a JITAI’s design decisions for an upcoming deployment. Based on simulation results, design decisions are made for the deployed JITAI. To continually improve the JITAI, data collected during deployment are used to update the JITAI-Twin—and this bidirectional feedback between deployments and simulation environments continues. JITAI-Twins are thus “fit-for-purpose” (term used in the National Academies 2024 consensus report on digital twins) instantiations of the digital twin concept. In this paper, we elucidate the specifics and design process of JITAI-Twins, with examples of prior use in clinical settings. JITAI-Twins highlight continuity over the course of a JITAI’s optimization and continual improvement, emphasizing the need for bidirectional feedback between versions of a simulation environment and a JITAI’s deployments.

Reinforcement learning (RL) has achieved remarkable success in real-world decision-making across diverse domains, including gaming, robotics, online advertising, public health, and natural language processing. Despite these advances, a substantial gap remains between RL research and its deployment in many practical settings. Two recurring challenges often underlie this gap. First, many settings offer limited opportunity for the agent to interact extensively with the target environment due to practical constraints. Second, many target environments often undergo substantial changes, requiring redesign and redeployment of RL systems (e.g., advancements in science and technology that change the landscape of healthcare delivery). Addressing these challenges and bridging the gap between basic research and application requires theory and methodology that directly inform the design, implementation, and continual improvement of RL systems in real-world settings. In this paper, we frame the application of RL in practice as a three-component process: (i) online learning and optimization during deployment, (ii) post- or between-deployment offline analyses, and (iii) repeated cycles of deployment and redeployment to continually improve the RL system. We provide a narrative review of recent advances that address the statistical challenges arising across these three components, including methods for enhancing sample efficiency during online deployment, maximizing data utility for post- or between-deployment inference, and designing sequences of deployments for continual improvement. We also outline future research directions in RL that are use-inspired – aiming for impactful application of RL in practice.

Cannabis use among emerging adults is increasing globally, posing significant health risks and creating a need for effective interventions. We present an exploratory analysis of the MiWaves pilot study, a digital intervention aimed at supporting cannabis use reduction among emerging adults (ages 18-25). Our findings indicate the potential of self-monitoring check-ins and trend visualizations in fostering self-awareness and promoting behavioral reflection in participants. MiWaves intervention message timing and frequency were also generally well-received by the participants. The participants' perception of effort were queried on intervention messages with different tasks, and our findings suggest that messages with tasks like exploring links and typing in responses are perceived as requiring more effort as compared to messages with tasks involving reading and acknowledging. Finally, we discuss the findings and limitations from this study and analysis, and their impact on informing future iterations on MiWaves.

Purpose: The widespread adoption of digital technologies offers unprecedented opportunities for delivering accessible and affordable health behavior interventions. Advances in mobile and wireless technologies have the capacity to collect rich and granular data about people’s states and contexts and utilize this information to intervene in (near) real-time in everyday life. Just-in-time adaptive interventions (JITAIs) can realize these opportunities by guiding how rapidly changing information about the person should be used in practice to decide whether, when and how to intervene. However, promoting health behavior change often requires interventions that harness and/or modify existing dyadic relationships, such as those involving parents, partners, care partners, or peers. To date, most JITAIs have been designed around a single individual rather than the coordinated delivery of support across two interdependent individuals. Methods: This manuscript establishes a common terminology to guide the development of dyadic JITAIs and advance innovation in the field of dyadic health behavior change. By synthesizing literature on JITAIs and dyadic interventions, we define dyadic JITAIs, clarify their scientific rationale, and describe their key components. We also offer guidelines for designing effective, resource-efficient dyadic JITAIs and identify directions for future research. Conclusion: Dyadic JITAIs have substantial potential to support health behavior change through shaping and harnessing close relationships. Realizing this potential will require new theories, experimental designs, and data analytic approaches to empirically inform their development.

The past decade has seen a surge in developing just-in-time adaptive interventions (JITAIs)—an intervention approach that leverages advancements in digital technologies to address the rapidly changing needs of individuals in daily life. This article provides an overview of the state of science on JITAI development and highlights important directions for future research. We explain what a JITAI is (and what it is not) and review the scientific and practical rationales underlying this approach. We also call attention to three key challenges relating to the development of JITAIs. The first challenge is that individuals may not be able to engage with (i.e., invest energy in) an intervention when they need it most in daily life. The second concerns the generally suboptimal engagement of individuals in interventions that leverage digital technologies as currently implemented. The third concerns the paucity of research on ways to harness the power of social relationships in JITAIs. We conclude that much research effort is needed to build more sophisticated and effective JITAIs.

Online artificial intelligence (AI) algorithms are an important component of digital health interventions. These online algorithms are designed to continually learn and improve their performance as streaming data is collected on individuals. Deploying online AI presents a key challenge: balancing adaptability of online AI with reproducibility. Online AI in digital interventions is a rapidly evolving area, driven by advances in algorithms, sensors, software, and devices. Digital health intervention development and deployment is a continuous process, where implementation - including the AI decision-making algorithm - is interspersed with cycles of re-development and optimization. Each deployment informs the next, making iterative deployment a defining characteristic of this field. This iterative nature underscores the importance of reproducibility: data collected across deployments must be accurately stored to have scientific utility, algorithm behavior must be auditable, and results must be comparable over time to facilitate scientific discovery and trustworthy refinement. This paper proposes a reproducible scientific workflow for developing, deploying, and analyzing online AI decision-making algorithms in digital health interventions. Grounded in practical experience from multiple real-world deployments, this workflow addresses key challenges to reproducibility across all phases of the online AI algorithm development life-cycle.

Medication adherence is critical for the recovery of adolescents and young adults (AYAs) who have undergone hematopoietic cell transplantation. However, maintaining adherence is challenging for AYAs after hospital discharge, who experience both individual (e.g. physical and emotional symptoms) and interpersonal barriers (e.g., relational difficulties with their care partner, who is often involved in medication management). To optimize the effectiveness of a three-component digital intervention targeting both members of the dyad as well as their relationship, we propose a novel Multi-Agent Reinforcement Learning (MARL) approach to personalize the delivery of interventions. By incorporating the domain knowledge, the MARL framework, where each agent is responsible for the delivery of one intervention component, allows for faster learning compared with a flattened agent. Evaluation using a dyadic simulator environment, based on real clinical data, shows a significant improvement in medication adherence (approximately 3%) compared to purely random intervention delivery. The effectiveness of this approach will be further evaluated in an upcoming trial.

We consider reinforcement learning (RL) for a class of problems with bagged decision times. A bag contains a finite sequence of consecutive decision times. The transition dynamics are non-Markovian and non-stationary within a bag. All actions within a bag jointly impact a single reward, observed at the end of the bag. For example, in mobile health, multiple activity suggestions in a day collectively affect a user's daily commitment to being active. Our goal is to develop an online RL algorithm to maximize the discounted sum of the bag-specific rewards. To handle non-Markovian transitions within a bag, we utilize an expert-provided causal directed acyclic graph (DAG). Based on the DAG, we construct states as a dynamical Bayesian sufficient statistic of the observed history, which results in Markov state transitions within and across bags. We then formulate this problem as a periodic Markov decision process (MDP) that allows non-stationarity within a period. An online RL algorithm based on Bellman equations for stationary MDPs is generalized to handle periodic MDPs. We show that our constructed state achieves the maximal optimal value function among all state constructions for a periodic MDP. Finally, we evaluate the proposed method on testbed variants built from real data in a mobile health clinical trial.

We consider after-study statistical inference for sequentially designed experiments wherein multiple units are assigned treatments for multiple time points using treatment policies that adapt over time. Our goal is to provide inference guarantees for the counterfactual mean at the smallest possible scale -- mean outcome under different treatments for each unit and each time -- with minimal assumptions on the adaptive treatment policy. Without any structural assumptions on the counterfactual means, this challenging task is infeasible due to more unknowns than observed data points. To make progress, we introduce a latent factor model over the counterfactual means that serves as a non-parametric generalization of the non-linear mixed effects model and the bilinear latent factor model considered in prior works. For estimation, we use a non-parametric method, namely a variant of nearest neighbors, and establish a non-asymptotic high probability error bound for the counterfactual mean for each unit and each time. Under regularity conditions, this bound leads to asymptotically valid confidence intervals for the counterfactual mean as the number of units and time points grows to $\infty$ together at suitable rates. We illustrate our theory via several simulations and a case study involving data from a mobile health clinical trial HeartSteps.

For the non-stationary multi-armed bandit (MAB) problem, many existing methods allow a general mechanism for the non-stationarity, but rely on a budget for the non-stationarity that is sub-linear to the total number of time steps T. In many real-world settings, however, the mechanism for the non-stationarity can be modeled, but there is no budget for the non-stationarity. We instead consider the non-stationary bandit problem where the reward means change due to a latent, auto-regressive (AR) state. We develop Latent AR LinUCB (LARL), an online linear contextual bandit algorithm that does not rely on the non-stationary budget, but instead forms good predictions of reward means by implicitly predicting the latent state. The key idea is to reduce the problem to a linear dynamical system which can be solved as a linear contextual bandit. In fact, LARL approximates a steady-state Kalman filter and efficiently learns system parameters online. We provide an interpretable regret bound for LARL with respect to the level of non-stationarity in the environment. LARL achieves sub-linear regret in this setting if the noise variance of the latent state process is sufficiently small with respect to T. Empirically, LARL outperforms various baseline methods in this non-stationary bandit problem.

Missing data in online reinforcement learning (RL) poses challenges compared to missing data in standard tabular data or in offline policy learning. The need to impute and act at each time step means that imputation cannot be put off until enough data exist to produce stable imputation models. It also means future data collection and learning depend on previous imputations. This paper proposes fully online imputation ensembles. We find that maintaining multiple imputation pathways may help balance the need to capture uncertainty under missingness and the need for efficiency in online settings. We consider multiple approaches for incorporating these pathways into learning and action selection. Using a Grid World experiment with various types of missingness, we provide preliminary evidence that multiple imputation pathways may be a useful framework for constructing simple and efficient online missing data RL methods.