Assessment & Research

Using Wearable Technology to Predict the Occurrence of Severe Behavior Problems among Neurodiverse Individuals: A Systematic Review

Romani et al. (2026) · Perspectives on Behavior Science 2026
★ The Verdict

Wearables can flash warnings seconds before severe behavior, but the science is still too thin for daily clinical use.

✓ Read this if BCBAs testing bio-signal tools in day or residential programs.
✗ Skip if Practitioners who need ready-to-use, evidence-based alert systems today.

01Research in Context

01

What this study did

Romani et al. (2026) hunted for papers that used smartwatches, chest bands, or other wearables to forecast severe behavior in neurodiverse people. They screened every study published up to 2026.

They kept 13 papers. The gadgets tracked heart rate, skin sweat, motion, or temperature. The goal was to see if the tech could give carers a few-second heads-up before a hit, bite, or scream.

02

What they found

The idea works in theory. Algorithms sometimes flagged rising stress a handful of seconds before the behavior.

But every study was tiny, used different cut-offs, and rarely checked if the alert was right outside the lab. The team calls the evidence 'early promise' only.

03

How this fits with other research

Liang et al. (2026) also used wearables on kids with NDDs, yet they measured movement, not behavior. They found the same weak spots: small samples and no two labs using the same recipe.

Kennedy (2025) showed sleep loss spikes challenging behavior in people with ID. That review is strong; Romani’s is shaky. The contrast warns us: wearables may predict, but fixing sleep gives faster, proven relief.

Mukherjee et al. (2021) reviewed infant wearables that predict cerebral palsy. Like Romani, they saw cool tech trapped by poor real-world tests. Together the three reviews say: hardware is ahead of validation.

04

Why it matters

Do not buy a smartwatch and assume it will keep your client safe. Use it as an extra probe, not a replacement for functional assessment. Track sleep first (H, 2025), then pilot wearables only if you can collect proof that the alerts match real outcomes.

Free CEUs

Want CEUs on This Topic?

The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.

Join Free →
→ Action — try this Monday

Run a one-week sleep log on any client with new spikes in aggression; fix sleep before you trial a new wrist sensor.

02At a glance

Intervention
not applicable
Design
systematic review
Population
mixed clinical
Finding
not reported

03Original abstract

Abstract Severe behavior problems (SBPs) exhibited by individuals with neurodevelopmental disabilities (NDD) can produce challenging and potentially dangerous situations. Although the field of behavior analysis has access to effective behavioral assessment and treatment methodologies, the risks associated with serving individuals with NDD engaging in SBPs remain high. Advances in wearable sensing, artificial intelligence, and machine learning offer potential support for behavior analysts working with individuals engaging in SBPs. Thus, researchers have begun studying physiological and behavioral signals (i.e., biometrics), such as heart rate or bodily motion, and their predictive relationship with SBPs. The current systematic literature review summarizes 13 peer-reviewed articles that studied predictive relations between biometrics and SBPs. We highlight commonalities, differences, and limitations among these studies. In particular, although some studies claim to predict the occurrence of SBPs over 30 s in advance of their occurrence, methodological concerns reduce the veracity of these claims. We propose short-term and long-term research questions to move this line of research forward.

Perspectives on Behavior Science, 2026 · doi:10.1007/s40614-026-00497-1