OtherJournal of neuroengineering and rehabilitation2023

Towards PPG-based anger detection for emotion regulation.

Tuck-Voon How, Robin E A Green, Alex Mihailidis

PMID 37582733

WHAT IT FOUND

A smartwatch sensor failed to reliably detect anger in people with and without brain injury.

The signal varied too much day-to-day and person-to-person, making it an unreliable standalone marker for predicting emotional outbursts in a clinical setting.

Key findings

01An anger detection model using only PPG data achieved 56.8% mean accuracy, which the authors describe as only marginally better than chance.

02PPG signal characteristics were dominated by individual differences and day-to-day variations rather than emotional state, making the signal non-stationary.

03While anger could be separated from non-anger within single sessions for most participants, the specific features used for this separation changed drastically between days.

STILL TO COME

How it was doneWhat they found

Read the rest of this summary

You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.

Already have one?

What it does not show

The study only included 32 participants, with just 10 having a traumatic brain injury. Data was collected in a laboratory setting, which may not reflect the complexity of real-world environments. The study used a binary view of anger (anger vs. non-anger), ignoring the intensity or different types of anger. Participants were selected for a propensity to anger but did not necessarily have severe anger dyscontrol. The study did not control for factors like skin tone, body mass, or sensor placement tightness, which can affect PPG signals.

Declared interests

The study was funded by the Canadian Institutes of Health Research. No other conflicts of interest are reported in the provided text.

The easy way to misread this

Do not interpret the 56.8% accuracy as evidence that smartwatches can detect anger. This figure is only slightly better than random guessing, and the model failed to improve on the baseline for detecting the specific risk event of anger. The technology is not ready for clinical use.

Read it on PubMed →


The study

Participants
32
Certainty of evidence
Low

Browse

    Cite

    Tuck-Voon How, Robin E A Green, Alex Mihailidis Towards PPG-based anger detection for emotion regulation. Journal of neuroengineering and rehabilitation. 2023.

    Read the original