Dr. Egon Dejonckheere
Egon does research in Emotion, Clinical Psychology and Abnormal Psychology. His most recent publication is 'The Bipolarity of Affect and Depressive Symptoms', featured in Journal of Personality and Social Psychology: Personality Processes and Individual Differences.
What is a just-in-time adaptive intervention?
Just-in-time adaptive interventions (JITAIs) are intervention designs that deliver real-time, context-aware support by adapting the timing, type, and intensity of intervention to an individual’s changing internal state and external context, with the aim of supporting behavior change and improving health outcomes.
The defining feature of a JITAI is dynamic, real-time adaptation. Unlike static interventions, such as stand-alone mobile apps that deliver the same content to all users, or conventional time-based and tailored approaches, such as Ecological Momentary Interventions that may customize support to relatively stable participant characteristics at baseline, JITAIs continuously update the content and level of support throughout the intervention period.
This adaptation is driven by incoming data on the individual’s current state and context, often collected through repeated self-report (Ecological Momentary Assessment / Experience Sampling) or passive mobile or wearable sensing. These data are used to determine whether to intervene, when to intervene, and what to deliver at each moment. The process is governed by pre-specified decision rules that map current states onto a set of intervention options, including the option to withhold support when intervention is unlikely to be beneficial.
JITAIs are widely used in digital health and behavior change research. Their study and design are often multidisciplinary and draw on behavioral science, mobile health (mHealth) research, statistics, and computer science.
The defining feature of a JITAI is dynamic, real-time adaptation. Unlike static interventions, such as stand-alone mobile apps that deliver the same content to all users, or conventional time-based and tailored approaches, such as Ecological Momentary Interventions that may customize support to relatively stable participant characteristics at baseline, JITAIs continuously update the content and level of support throughout the intervention period.
This adaptation is driven by incoming data on the individual’s current state and context, often collected through repeated self-report (Ecological Momentary Assessment / Experience Sampling) or passive mobile or wearable sensing. These data are used to determine whether to intervene, when to intervene, and what to deliver at each moment. The process is governed by pre-specified decision rules that map current states onto a set of intervention options, including the option to withhold support when intervention is unlikely to be beneficial.
JITAIs are widely used in digital health and behavior change research. Their study and design are often multidisciplinary and draw on behavioral science, mobile health (mHealth) research, statistics, and computer science.
How JITAIs decide when to intervene
A central assumption underlying just-in-time adaptive interventions (JITAIs) is that intervention effectiveness depends not only on what is delivered, but also on when it is delivered, with the goal of maximizing impact while minimizing unnecessary or ill-timed interventions. Three key concepts used to guide this timing are vulnerability, opportunity, and receptivity.
Vulnerability
Vulnerability refers to moments when a person is at increased risk for an undesirable outcome, such as feeling stressed, experiencing craving, or engaging in unhealthy behavior. These moments are not constant, but fluctuate over time depending on both stable factors, such as personality or life circumstances, and transient factors, such as current mood or situation.
For example, someone trying to reduce alcohol use may generally be doing well, but become vulnerable in the evening when feeling tired or when being in a social setting where others are drinking. Mobile JITAIs aim to detect these moments and provide support before the situation escalates.
For example, someone trying to reduce alcohol use may generally be doing well, but become vulnerable in the evening when feeling tired or when being in a social setting where others are drinking. Mobile JITAIs aim to detect these moments and provide support before the situation escalates.
Opportunity
Opportunity refers to moments when an intervention is likely to be effective because the person is in a position to benefit from it. These are often “teachable moments” of reinforcement learning where behavior is ongoing and can still be influenced.
For example, a prompt encouraging a short walk is more effective when someone has been sitting for a long time, rather than hours later. Similarly, providing feedback immediately after a behavior increases the chance that the person can learn from it and adjust future actions.
For example, a prompt encouraging a short walk is more effective when someone has been sitting for a long time, rather than hours later. Similarly, providing feedback immediately after a behavior increases the chance that the person can learn from it and adjust future actions.
Receptivity
Receptivity refers to whether a person is (physically) able and willing to engage with an intervention at a given moment. Even if vulnerability is high or a good opportunity is present, an intervention may not work if the person is busy, distracted, or not open to receiving support.
For example, a notification delivered during a meeting or while driving is unlikely to be acted upon. Repeatedly delivering interventions at such moments can reduce engagement and contribute to intervention fatigue over time.
For example, a notification delivered during a meeting or while driving is unlikely to be acted upon. Repeatedly delivering interventions at such moments can reduce engagement and contribute to intervention fatigue over time.
In practice, JITAIs aim to deliver interventions at moments when vulnerability is elevated or a clear opportunity for learning or behavior change is present, and only when receptivity is sufficient to support engagement. These considerations are formalized through the core components of the JITAI framework.
The core components of a JITAI design
To translate the timing logic of vulnerability, opportunity and receptivity into a working intervention, just-in-time adaptive interventions (JITAIs) are structured around six key characteristics that define how support is delivered and adapted over time. These components specify the outcomes the intervention aims to influence, the data used to guide decisions, the moments at which decisions are made, and the actions that can be taken. Together, they provide a framework for linking real-time data to adaptive intervention delivery in a systematic and reproducible way.
Distal outcome
The distal outcome is the ultimate goal of the intervention, such as preventing relapse, improving mental health, or increasing physical activity. It represents the primary endpoint the JITAI is designed to influence and should guide all design decisions.
Proximal outcome
Proximal outcomes are short-term, actionable targets that the intervention aims to influence in the moment. These often reflect mechanisms that lead to the distal outcome (i.e., intermediate behavioral processes), such as reducing stress, lowering craving, or increasing motivation. By repeatedly influencing proximal outcomes, JITAIs aim to produce longer-term change.
Tailoring variables
Tailoring variables are the inputs used to inform intervention decisions. These variables capture the individual’s current state and context, such as affect, behavior, location, social setting, or activity.
They can be collected through active assessments, such as self-report via ecological momentary assessment, or through passive sensing, such as GPS, accelerometery, or device usage. The choice of tailoring variables is critical, as they determine how well the system can detect moments of vulnerability or opportunity and how precisely interventions can be targeted.
They can be collected through active assessments, such as self-report via ecological momentary assessment, or through passive sensing, such as GPS, accelerometery, or device usage. The choice of tailoring variables is critical, as they determine how well the system can detect moments of vulnerability or opportunity and how precisely interventions can be targeted.
Decision points
Decision points are moments at which the system evaluates whether an intervention should be delivered. These can occur at fixed times, at regular intervals, or event-contingent, such as changes in behavior or context.
Because states of vulnerability and opportunity can emerge rapidly and unpredictably, decision points in JITAIs often occur frequently and are closely tied to real-time monitoring. Not every decision point results in an intervention. At each point, the system evaluates whether delivering support isappropriate, allowing for both action and inaction.
Because states of vulnerability and opportunity can emerge rapidly and unpredictably, decision points in JITAIs often occur frequently and are closely tied to real-time monitoring. Not every decision point results in an intervention. At each point, the system evaluates whether delivering support isappropriate, allowing for both action and inaction.
Decision rules
Decision rules specify how tailoring variables are translated into intervention decisions. They define whether an intervention should be delivered at a given decision point and, if so, which intervention option should be selected.
These rules operationalize the adaptive nature of mobile health interventions by linking real-time data to action in a systematic way. For example, a rule may specify that if a risk score exceeds a certain threshold, a specific intervention is delivered; otherwise, no action is taken.
These rules operationalize the adaptive nature of mobile health interventions by linking real-time data to action in a systematic way. For example, a rule may specify that if a risk score exceeds a certain threshold, a specific intervention is delivered; otherwise, no action is taken.
Intervention options
Intervention options refer to the set of actions that can be delivered to the individual at any given decision point. These may include prompts, behavioral suggestions, feedback, psycho-educational content, or brief therapeutic exercises.
Intervention options can vary in type, intensity, timing, and delivery mode. Importantly, JITAIs explicitly include the option to provide no intervention when it is unlikely to be beneficial. This reflects a key design principle: support should only be delivered when it is expected to add value.
Intervention options can vary in type, intensity, timing, and delivery mode. Importantly, JITAIs explicitly include the option to provide no intervention when it is unlikely to be beneficial. This reflects a key design principle: support should only be delivered when it is expected to add value.
Potential challenges of JITAI designs
While just-in-time adaptive interventions (JITAIs) offer a powerful framework for delivering timely and personalized support, their implementation in real-world settings is not without challenges. These challenges arise from the need to combine continuous data collection, real-time decision-making, and sustained user engagement in dynamic environments. Understanding these limitations is essential for designing interventions that are not only adaptive in theory, but also effective and usable in practice.
The engagement paradox
A fundamental challenge in JITAI design is that the moments when intervention is most needed, namely periods of peak vulnerability, are often the moments when individuals are least able or willing to engage with it. High cognitive load, emotional dysregulation, or unfavorable contextual demands can simultaneously increase the need for support while reducing receptivity.
As a result, mobile health interventions delivered at these critical moments may be ignored, delayed, or processed superficially. In some cases, poorly timed interventions may even have unintended negative effects, for example by increasing frustration or interrupting ongoing activities. This creates a central design tension: JITAIs must identify moments of need, while also ensuring that support is delivered when individuals are able to engage with it effectively.
As a result, mobile health interventions delivered at these critical moments may be ignored, delayed, or processed superficially. In some cases, poorly timed interventions may even have unintended negative effects, for example by increasing frustration or interrupting ongoing activities. This creates a central design tension: JITAIs must identify moments of need, while also ensuring that support is delivered when individuals are able to engage with it effectively.
Burden and data quality
The effectiveness of a JITAI depends directly on the quality of its tailoring variables. When these are based on repeated self-report, they can impose substantial burden, leading to missed assessments, delayed responses, or careless responding over time. As a result, the data used to guide intervention decisions may become noisy or unreliable.
When tailoring variables are derived from passive sensing, different challenges arise. In real-world deployment, sensor data are often incomplete or unreliable due to device non-wear, battery depletion, connectivity issues, or differences in proprietary algorithms across devices. In addition, many internal psychological states cannot be directly observed, creating a “mind-body gap” where inferences from sensor data remain imperfect.
Because mobile health interventions rely on these inputs to determine when and how to intervene, poor data quality directly undermines adaptivity. Decisions based on missing or erroneous data may be mistimed or inappropriate, and in some cases worse than no intervention at all.
When tailoring variables are derived from passive sensing, different challenges arise. In real-world deployment, sensor data are often incomplete or unreliable due to device non-wear, battery depletion, connectivity issues, or differences in proprietary algorithms across devices. In addition, many internal psychological states cannot be directly observed, creating a “mind-body gap” where inferences from sensor data remain imperfect.
Because mobile health interventions rely on these inputs to determine when and how to intervene, poor data quality directly undermines adaptivity. Decisions based on missing or erroneous data may be mistimed or inappropriate, and in some cases worse than no intervention at all.
Changing behavior and context over time
A key challenge in JITAI design is that the relations between variables are not stable over time. Behavior and context can shift as individuals learn, adapt, or encounter new situations, meaning that patterns that are predictive early in an intervention may become less relevant later. For example, a trigger for stress or craving may initially signal vulnerability but lose its predictive value as coping strategies develop. Similarly, the effectiveness of specific intervention options may change over time as the person gradually acquires new skills.
As a result, fixed decision rules can become outdated, leading to less effective or poorly timed interventions. JITAIs therefore need to account for ongoing change, either by updating rules or by using adaptive approaches.
As a result, fixed decision rules can become outdated, leading to less effective or poorly timed interventions. JITAIs therefore need to account for ongoing change, either by updating rules or by using adaptive approaches.
Interdisciplinary design and complex setup
Designing a JITAI requires integrating multiple components, including measurement, decision-making, and intervention delivery. This brings together expertise from behavioral science, statistics, computer science, and human-computer interaction, making both design and implementation inherently complex.
In addition, JITAI development often involves multiple stakeholders, such as researchers, clinicians, developers, and end users. Aligning theoretical goals, technical constraints, and user needs is not straightforward, and misalignment can lead to interventions that are either difficult to implement or poorly suited to real-world use.
As a result, effective JITAI design requires close collaboration across disciplines and stakeholders, as well as iterative development to ensure that the system remains both scientifically grounded and practically usable.
In addition, JITAI development often involves multiple stakeholders, such as researchers, clinicians, developers, and end users. Aligning theoretical goals, technical constraints, and user needs is not straightforward, and misalignment can lead to interventions that are either difficult to implement or poorly suited to real-world use.
As a result, effective JITAI design requires close collaboration across disciplines and stakeholders, as well as iterative development to ensure that the system remains both scientifically grounded and practically usable.
Privacy and ethical considerations
Mobile health interventions often rely on continuous monitoring of behavior and context, including sensitive data such as location, activity patterns, or device use. This raises important concerns around data privacy, security, and responsible data handling. Collecting high-frequency, real-world data increases the risk of unintended disclosure or misuse, particularly when data streams are combined.
Beyond data protection, JITAIs also raise questions about transparency and autonomy. Because intervention decisions are made by the system, users may not fully understand why certain interventions are delivered or withheld. This can affect trust and perceived control.
Ethical JITAI design therefore requires clear consent procedures, data minimization, and mechanisms that give users insight and control over how data are used and how interventions are triggered.
Beyond data protection, JITAIs also raise questions about transparency and autonomy. Because intervention decisions are made by the system, users may not fully understand why certain interventions are delivered or withheld. This can affect trust and perceived control.
Ethical JITAI design therefore requires clear consent procedures, data minimization, and mechanisms that give users insight and control over how data are used and how interventions are triggered.
Evaluating the effectiveness of a JITAI
Evaluating the effectiveness of just-in-time adaptive interventions (JITAIs) requires methods that account for their dynamic and adaptive nature. Because interventions are delivered repeatedly and tailored over time, different designs are needed to assess effects at multiple levels, from momentary responses to long-term outcomes.
Micro-randomized trials
Micro-randomized trials (MRTs) are specifically designed to evaluate time-varying interventions such as JITAIs. In an MRT, individuals are repeatedly randomized at each decision point to either receive an intervention or not. This allows researchers to estimate the immediate, or proximal, effect of an intervention at the moment it is delivered.
Because randomization occurs many times within individuals, MRTs provide detailed insight into when and under which conditions an intervention is effective. For example, they can be used to test whether a prompt reduces stress in the next hour, and whether this effect depends on context or prior state.
Unlike traditional trial designs, MRTs focus on within-person, momentary effects rather than overall outcomes. This makes them particularly well suited for evaluating the core logic of JITAIs, where timing and context are central.
Because randomization occurs many times within individuals, MRTs provide detailed insight into when and under which conditions an intervention is effective. For example, they can be used to test whether a prompt reduces stress in the next hour, and whether this effect depends on context or prior state.
Unlike traditional trial designs, MRTs focus on within-person, momentary effects rather than overall outcomes. This makes them particularly well suited for evaluating the core logic of JITAIs, where timing and context are central.
Sequential multiple assignment randomized trials
Sequential multiple assignment randomized trials (SMARTs) are used to evaluate and optimize adaptive intervention strategies over longer time scales. In a SMART design, participants are randomized at key decision points during the intervention, for example based on their response to earlier treatment.
This allows researchers to compare different sequences of interventions, such as whether individuals who do not respond to an initial strategy benefit more from intensifying support or switching to a different approach. As such, SMARTs focus on identifying effective adaptation strategies rather than momentary intervention effects.
Compared to MRTs, which operate at the level of individual decision points, SMARTs evaluate broader treatment pathways over time. They are particularly useful for informing how JITAIs should adapt across phases of an intervention.
This allows researchers to compare different sequences of interventions, such as whether individuals who do not respond to an initial strategy benefit more from intensifying support or switching to a different approach. As such, SMARTs focus on identifying effective adaptation strategies rather than momentary intervention effects.
Compared to MRTs, which operate at the level of individual decision points, SMARTs evaluate broader treatment pathways over time. They are particularly useful for informing how JITAIs should adapt across phases of an intervention.
Randomized controlled trials
Randomized controlled trials (RCTs) remain the standard approach for evaluating the overall effectiveness of an intervention. In an RCT, participants are typically randomized once to receive either the intervention or a control condition, and outcomes are assessed over a predefined period.
In the context of JITAIs, RCTs are used to determine whether the intervention as a whole leads to improvements in distal outcomes, such as symptom reduction or behavior change. However, they provide limited insight into how and when the intervention produces these effects.
Unlike MRTs and SMARTs, RCTs do not capture the time-varying and adaptive nature of JITAIs. While they are essential for establishing overall efficacy, they are less suited for understanding the underlying mechanisms or optimizing the timing and delivery of interventions.
In the context of JITAIs, RCTs are used to determine whether the intervention as a whole leads to improvements in distal outcomes, such as symptom reduction or behavior change. However, they provide limited insight into how and when the intervention produces these effects.
Unlike MRTs and SMARTs, RCTs do not capture the time-varying and adaptive nature of JITAIs. While they are essential for establishing overall efficacy, they are less suited for understanding the underlying mechanisms or optimizing the timing and delivery of interventions.
Examples of JITAIs
Just-in-time Adaptive Interventions (JITAIs) can be used across many domains of psychology and behavior change. They are especially useful when support needs to be timed to a person’s current situation, such as their location, activity level, emotional state, risk context, or recent behavior. Below we discuss three evidence-based examples that show how different JITAIs can look in practice.
Increasing physical activity
One well-known JITAI application is the use of a mobile health intervention to encourage physical activity. Instead of sending the same reminder at the same time each day, the system can use information such as time of day, recent activity, weather, and the person’s availability to decide when an activity suggestion might be useful.
In a micro-randomized trial, researchers found that contextually tailored activity suggestions increased subsequent walking compared with moments when no suggestion was sent. This example shows how JITAIs can support everyday behavior change by intervening when action is feasible.
In a micro-randomized trial, researchers found that contextually tailored activity suggestions increased subsequent walking compared with moments when no suggestion was sent. This example shows how JITAIs can support everyday behavior change by intervening when action is feasible.
Supporting smoking cessation
JITAIs have also been used to support people who are trying to quit smoking. In this type of mHealth intervention, smartphone-based systems can combine self-report, contextual information, and tailored messages to deliver support when lapse risk is expected to be higher.
In a randomized controlled trial among adults with low income, a tailored just-in-time smoking-cessation app was compared with a standard smoking-cessation app. The JITAI approach was designed to provide more personalized and timely support around smoking triggers, cravings, and lapse risk. The study found that the tailored JITAI improved smoking-cessation outcomes compared with the comparison app, suggesting that just-in-time personalization may be especially useful when people face changing risk contexts in daily life.
In a randomized controlled trial among adults with low income, a tailored just-in-time smoking-cessation app was compared with a standard smoking-cessation app. The JITAI approach was designed to provide more personalized and timely support around smoking triggers, cravings, and lapse risk. The study found that the tailored JITAI improved smoking-cessation outcomes compared with the comparison app, suggesting that just-in-time personalization may be especially useful when people face changing risk contexts in daily life.
Reducing depressive rumination
JITAIs may also be effective to target depressive rumination in people with clinical depression. Rumination is especially well suited to a just-in-time approach because it often unfolds in episodes, can be triggered by daily events, and may become more difficult to interrupt once it has continued for some time.
In a pilot randomized controlled trial, researchers tested a mobile rumination-focused CBT intervention for patients in therapy for clinical depression. Participants were prompted by smartphone to report rumination-related symptoms several times per day, and the intervention was personalized to their own rumination timing patterns. When rumination was detected, the system delivered mobile CBT materials designed to interrupt the episode and reduce carry over into later rumination.
Compared with a no-treatment control condition, participants receiving the JITAI reported larger reductions in the number of rumination episodes and in the average time spent ruminating. This example shows how JITAIs can target short-lived psychological processes or symptoms as they occur in daily life (i.e., proximal outcomes), rather than treating depression a general syndrom.
In a pilot randomized controlled trial, researchers tested a mobile rumination-focused CBT intervention for patients in therapy for clinical depression. Participants were prompted by smartphone to report rumination-related symptoms several times per day, and the intervention was personalized to their own rumination timing patterns. When rumination was detected, the system delivered mobile CBT materials designed to interrupt the episode and reduce carry over into later rumination.
Compared with a no-treatment control condition, participants receiving the JITAI reported larger reductions in the number of rumination episodes and in the average time spent ruminating. This example shows how JITAIs can target short-lived psychological processes or symptoms as they occur in daily life (i.e., proximal outcomes), rather than treating depression a general syndrom.
Roots and history of JITAI
Just-in-time Adaptive Interventions (JITAIs) emerge at the intersection of behavioral science, computer science, and mobile technology. They were developed to provide tailored support in real-world settings, addressing the limitations of static, one-size-fits-all interventions.
A key foundation lies in the adaptive treatment strategy literature, developed by Susan Murphy and colleagues in the early 2000s, which introduced methods for adjusting support based on individual response over time. Around the same period, approaches such as EMA or ESM had already demonstrated that it was feasible to measure people’s experiences repeatedly in their daily lives, enabling the use of real-time data not only to understand behavior, but also to guide intervention.
The widespread adoption of smartphones further accelerated this development by enabling continuous data collection through mobile sensing and providing a scalable platform for delivering interventions in real time. The term “just-in-time adaptive intervention” was later introduced by Inbal Nahum-Shani and colleagues in 2015, bringing these ideas together into a unified design framework. Since then, the field has grown rapidly, with advances in experimental designs, statistical methods, and real-world applications across awide range of domains, including mental health, physical activity, and workplace well-being.
A key foundation lies in the adaptive treatment strategy literature, developed by Susan Murphy and colleagues in the early 2000s, which introduced methods for adjusting support based on individual response over time. Around the same period, approaches such as EMA or ESM had already demonstrated that it was feasible to measure people’s experiences repeatedly in their daily lives, enabling the use of real-time data not only to understand behavior, but also to guide intervention.
The widespread adoption of smartphones further accelerated this development by enabling continuous data collection through mobile sensing and providing a scalable platform for delivering interventions in real time. The term “just-in-time adaptive intervention” was later introduced by Inbal Nahum-Shani and colleagues in 2015, bringing these ideas together into a unified design framework. Since then, the field has grown rapidly, with advances in experimental designs, statistical methods, and real-world applications across awide range of domains, including mental health, physical activity, and workplace well-being.
Dr. Egon Dejonckheere
Egon Dejonckheere is an Assistant Professor at Tilburg University, affiliated with KU Leuven, specializing in ambulatory assessment and intensive longitudinal research. His work combines methodological innovation to improve these methods with applications in mental health and emotion dynamics, and includes dozens of smartphone-based studies in daily life.
FAQ
01.
What is a Just-in-Time Adaptive Intervention (JITAI)?
A Just-in-Time Adaptive Intervention (JITAI) is a type of intervention that delivers support in real time by adapting to a person’s current state and context. It uses data such as self-reports or sensor inputs to decide when and how to intervene, with the goal of providing the right support at the right moment.
02.
What is the difference between JITAI and EMI?
Ecological Momentary Interventions (EMIs) deliver interventions in real-world settings, often based on time or simple triggers. JITAIs extend this approach by continuously adapting when and how support is delivered based on incoming data and decision rules, making them more dynamic and personalized.
03.
What data do JITAIs use?
JITAIs typically use a combination of self-report and passive sensing data. Self-report is often collected through methods such as Ecological Momentary Assessment (EMA) or Experience Sampling Methodology (ESM), while passive data can include location, activity, or device usage. These data are used as tailoring variables to guide intervention decisions.
04.
What is the difference between vulnerability, opportunity, and receptivity in JITAI design?
In JITAIs, vulnerability refers to moments of increased risk (e.g., stress or craving), opportunity to moments where intervention can be effective (e.g., a teachable moment), and receptivity to whether a person is able and willing to engage. In practice, interventions are delivered when vulnerability or opportunity is present, but only if receptivity is sufficient.
05.
Are JITAIs effective?
JITAIs have shown promise across a range of domains, including mental health, substance use, and physical activity. Their effectiveness depends on how well they identify the right moments to intervene and how appropriate the delivered support is. Specialized study designs, such as micro-randomized trials, are often used to evaluate their (proximal) effects.
06.
How are JITAIs evaluated?
JITAIs are evaluated using a combination of study designs. Micro-randomized trials (MRTs) assess momentary effects of interventions, sequential multiple assignment randomized trials (SMARTs) evaluate adaptive intervention strategies over time, and randomized controlled trials (RCTs) assess overall effectiveness on long-term outcomes.
07.
What are the main challenges of JITAIs?
Key challenges include identifying the right moment to intervene, maintaining user engagement, dealing with missing or noisy data, and ensuring privacy and ethical data use. In addition, designing JITAIs requires integrating insights from multiple disciplines, making development complex.