TLDR: SigmaScheduling is a new method for mobile health interventions that dynamically schedules ‘decision points’ based on the predictability of an individual’s behavior. Unlike fixed-interval scheduling, it adjusts the timing of interventions—scheduling them closer to predicted behavior for predictable routines and earlier for less predictable ones. Evaluated with toothbrushing data from 68 participants, SigmaScheduling significantly increased the likelihood of interventions occurring before the target behavior, enhancing the effectiveness of personalized mobile health support.
Mobile health (mHealth) interventions, particularly those known as Just-in-Time Adaptive Interventions (JITAIs), are designed to provide support at crucial moments to help individuals adopt healthier behaviors. Imagine an app that reminds you to brush your teeth, or take medication, right when you need it most. The effectiveness of these interventions often hinges on delivering support precisely when it’s most impactful—shortly before a target behavior is likely to occur.
However, a significant challenge arises because people’s daily routines are not always predictable. The traditional approach to scheduling these ‘decision points’ (the moments when the mHealth system decides whether and how to intervene) is to set them at a fixed interval before a user’s self-reported expected behavior time. While this ‘one-size-fits-all’ method might work for individuals with highly consistent routines, it often falls short for those with irregular schedules. For instance, if someone’s toothbrushing time varies significantly day-to-day, a fixed reminder might arrive too late, after the behavior has already happened, rendering the intervention ineffective.
This is where a new method called SigmaScheduling comes into play. Proposed by researchers including Asim H. Gazi and Susan A. Murphy, SigmaScheduling offers a dynamic and personalized approach to scheduling these critical decision points. Instead of relying on a fixed interval, it intelligently adjusts the timing based on the *uncertainty* in predicting when a behavior will occur. If a person’s routine is very predictable, SigmaScheduling schedules the decision point closer to the anticipated behavior time. Conversely, if the timing is less certain, it schedules the decision point earlier, increasing the likelihood that the intervention will be delivered in a timely manner, before the behavior takes place.
The core idea behind SigmaScheduling is to use a measure of uncertainty (like a standard deviation) alongside the predicted behavior time. This allows the system to be more flexible. For example, in the context of the ‘Oralytics’ JITAI, an intervention designed to improve daily toothbrushing, the system needs to anticipate brushing events based on past patterns, as real-time detection isn’t always feasible. SigmaScheduling ensures that the intervention opportunity is preserved even when routines are variable.
To evaluate its effectiveness, SigmaScheduling was tested using real-world data from 68 participants in a 10-week trial of Oralytics. The study compared SigmaScheduling against the traditional fixed-interval method, using two different ways to predict brush times and quantify uncertainty: one based on user-provided times and past errors, and another utilizing online Bayesian machine learning. The results were compelling: SigmaScheduling significantly increased the probability that decision points occurred *before* brushing events, achieving this in at least 70% of cases. This improvement was particularly noticeable when a higher proportion of successful interventions was desired.
Also Read:
- Supporting Independent Living: A New Framework for Monitoring Daily Activities of Older Adults
- ExRec: A New Framework for Personalized Exercise Recommendations in Online Learning
In essence, SigmaScheduling’s strength lies in its personalization. It recognizes that not all individuals have the same level of predictability in their routines. By adapting the lead time for interventions, it ensures that support is delivered at the most opportune moment for each individual, maximizing the chances of influencing behavior. This advancement holds significant promise for the field of precision mHealth, especially for JITAIs targeting time-sensitive, habitual behaviors such as oral hygiene or dietary habits. While the current evaluation focused on oral self-care, future research aims to deploy SigmaScheduling in real-time and explore its application in other health domains. You can read the full research paper here.
