Can a Smartwatch Detect Anxiety?
A wrist skin sensor sorted calm from distress with about 89% accuracy in 50 people. Excitement against anxiety was not tested in any study found.
Evidence rating for the proposition “Electrodermal activity rises with acute sympathetic arousal induced by psychological stressors”: Emerging
It can detect arousal, but telling anxiety apart from excitement is not proven. Wearable skin sensors measure sympathetic arousal, which rises with emotional and cognitive states alike. Tests have compared calm with distress or stress with baseline. A wrist skin conductance device sorted calm from distress with about 89% accuracy in 50 people viewing standard pictures. No study found here compared excitement with anxiety. In a 237-person Fitbit Sense 2 study, elevated stress alone went with higher skin temperature, not higher skin conductance.
What does a wearable stress sensor actually measure?
Skin conductance, which follows sweat gland activity. The sympathetic nervous system drives it. A review of the brain mechanisms behind it calls electrodermal activity a sensitive index of changes in sympathetic arousal, integrated with emotional and cognitive states.
Arousal is not one feeling. The same signal can rise for effort, attention or emotion.
Heart rate and skin conductance also do not move as one. A brain imaging study of 18 people under social threat found brain patterns predicted each signal (r = 0.54 for heart rate, 0.58 for skin conductance). The patterns predicted each other only weakly (r = 0.21 and 0.22). The authors conclude that physiological measures of threat are not interchangeable.
A review of stress biomarkers in animals and humans found that most physiological changes during emotional stress do not track how intense the stressor is. Heart rate and blood pressure respond quickly but are poor gauges of intensity; skin conductance and pupil dilation looked more promising (Neuroscience and Biobehavioral Reviews, 2021).
2 studies
- Electrodermal activity is described as a sensitive psychophysiological index of changes in autonomic sympathetic arousal that are integrated with emotional and cognitive states, with brain lesion and imaging studies clarifying the regions involved.
- In 18 people under social threat, multivariate brain patterns predicted heart rate (r = 0.54) and skin conductance level (r = 0.58). Each pattern predicted the other signal only weakly (r = 0.21 and 0.22).
Has anyone tested whether it can tell excitement from anxiety?
Not in the studies found here. The tests compared calm with distress, or stress with baseline.
A wrist-worn skin conductance device was tested in 50 participants viewing pictures from the International Affective Picture System. It sorted calm from distress with about 89% accuracy.
A second study followed 41 patients before surgery. An algorithm that detected motion artifacts had 97.83% accuracy on new users. Classifying low, moderate and high stress reached 85.06% on new users.
Both tests used a stressor and a control. Neither included a positive high-arousal state such as excitement. The result says the device can tell a calm person from a distressed one. It does not say it can tell two aroused people apart.
2 studies
- A wrist-worn electrodermal device sorted calm from distress with about 89% accuracy in 50 participants viewing standardized pictures.
- In 41 patients before surgery, a wrist wearable with an artifact detector reached 97.83% artifact-detection accuracy and 85.06% accuracy grading low, moderate and high stress on new users.
What does everyday wearable data show?
It shows group differences in mental health symptoms, not moment-to-moment emotion. A prospective study analyzed 237 people who wore a Fitbit Sense 2 with its electrodermal sensor on for four weeks.
People with elevated depressive and anxiety symptoms had higher tonic EDA, skin temperature and heart rate than people without them. They did not move more. People with elevated stress alone had only higher skin temperature.
Differences in EDA were largest in the early morning. The authors suggest electrodermal sensors may be useful for measuring physiological correlates of symptoms in daily life.
1 study
- In 237 people wearing a Fitbit Sense 2 for four weeks, elevated depressive and anxiety symptoms went with higher tonic EDA, skin temperature and heart rate, while elevated stress alone went with only higher skin temperature.
Can heat or movement look like arousal?
Yes. Thermoregulation changes skin conductance independently of emotion. A 2026 study used the WESAD dataset to correct EDA for temperature. The corrected features improved separation of stress-related from heat-related activity. The abstract gives no accuracy figure.
Movement is a second source of error. A study of 10 people recorded EDA from both hands while they moved only one. A machine learning model detected movement artifacts with 83.85% accuracy (standard deviation 4.91%). Wrist recordings in daily life carry that noise.
The wider question of stress scores is in whether a stress score measures stress.
What does this not settle?
It does not show whether any consumer device can separate excitement from anxiety. No study found here tested that comparison.
The 50-person test used still pictures. The pre-surgery study used one hospital's patients. The 237-person study compared symptom groups and did not label individual moments.
The accuracy figures come from a 50-person picture experiment and a 41-patient pre-surgery study using research or prototype wrist devices. None tested positive high-arousal states such as excitement, and none tested a consumer stress alert against the user's own emotion in real time.
Common questions
Can a wearable tell excitement from anxiety?
Not shown. Skin conductance reflects sympathetic arousal, which rises with emotional and cognitive states alike. The tests found compared calm with distress or stress with baseline, and none compared excitement with anxiety.
How accurate is wearable stress detection from skin conductance?
A wrist device sorted calm from distress with about 89% accuracy in 50 people viewing standard pictures. In 41 pre-surgery patients, grading low, moderate and high stress reached 85.06% on new users.
Does everyday wearable data separate stress from anxiety?
Partly, by group. In 237 Fitbit Sense 2 wearers, elevated depressive and anxiety symptoms went with higher tonic EDA, skin temperature and heart rate, while elevated stress alone went with only higher skin temperature.
Can heat or movement trigger a stress reading?
In skin conductance, yes. Temperature alters the signal independently of emotion, and movement adds artifacts that a machine learning model detected with 83.85% accuracy in a 10-person test.
Do heart rate and skin conductance rise together?
Only partly. In 18 people under social threat, brain patterns predicting one signal predicted the other weakly (r = 0.21 and 0.22).
Sources
- Electrodermal responses: what happens in the brain.
- Multivariate Brain Prediction of Heart Rate and Skin Conductance Responses to Social Threat.
- Electrodermal Activity Sensor for Classification of Calm/Distress Condition.
- Electrodermal Activity Based Pre-surgery Stress Detection Using a Wrist Wearable.
- Evidence of differences in diurnal electrodermal, temperature and heart rate patterns by mental health status in free-living data [author affiliations include Google and Verily Life Sciences; funding listed from Google]
- Electrodermal Temperature-Adjusted Electrodermal Activity (EDA) for Stress Detection in Virtual Reality.
- A Preliminary Study on Automatic Motion Artifact Detection in Electrodermal Activity Data Using Machine Learning.
- Focusing attention on biological markers of acute stressor intensity: Empirical evidence and limitations.