Pupil reads the language of your gaze to screen for ADHD. Non-invasive, research-backed, and fast. We turn a short look at a screen into a clear signal, in minutes.
A calm, guided experience that takes under five minutes. No clinic visit, no waiting list.
Follow gentle dots on screen for under a minute while your webcam records your eyes.
Our vision pipeline extracts saccades, fixations and pupil dynamics frame by frame, then a transformer reads the temporal signature of attention.
Get an instant, explainable report with biomarkers, confidence, charts and clinical insights to share with a professional.
Record a short clip with your webcam or upload a video. Pupil sends it to the cloud and returns an attention analysis in seconds.
Enable your camera to record a short eye-tracking clip.
Sit about an arm's length away with your face evenly lit.
No eye video handy? Try a labeled sample from the training set:
Screening technology that makes ADHD detection accessible, fast, and genuinely non-invasive.
No blood tests, no brain scans. Simply watch dots on a screen while we read your eye movements in real time.
Complete a screening in under five minutes. Our AI processes thousands of data points and returns insights immediately.
No special hardware. Pupil runs on a standard laptop or phone webcam, so you can screen anywhere.
Trained on extensive eye-tracking datasets, our model spots the subtle gaze patterns that distinguish ADHD from neurotypical attention.
Watch your gaze trace live in the screening lab. See exactly what the AI sees as it analyzes your eye movements.
All processing happens locally. Your eye data never leaves your machine.
State-of-the-art AI architecture powering clinical-grade eye-tracking analysis.
Our camera system captures eye movements at high frequency. Pupil-detection algorithms extract gaze coordinates, pupil dilation and fixation stability in real time.
At the heart of Pupil is a Transformer encoder, the architecture behind modern AI breakthroughs. It learns temporal patterns in gaze sequences that traditional methods miss.
We pre-train on large-scale eye-tracking data (GazeBase: 12,000+ recordings) to learn general gaze dynamics, then fine-tune on ADHD-specific data to specialize in attention detection.
Pupil extracts clinically meaningful features that correlate with attention disorders:
Every frame is processed locally. Your eye video never leaves your machine, and your results stay yours.
Built on peer-reviewed science and validated against clinical diagnoses.
Eye movements are controlled by the same neural circuits behind attention and executive function, the core deficits in ADHD.
Pupil leverages transfer learning from GazeBase, one of the largest eye-tracking repositories, with 322 participants across multiple sessions.
The model learns robust representations of eye-movement dynamics before specializing on ADHD detection.
Our model is validated against clinical ADHD diagnoses, with ongoing studies to establish sensitivity and specificity benchmarks.
Pupil is designed to complement, not replace, professional clinical evaluation.
Large-scale, multi-stimulus, longitudinal eye-movement dataset.
View on figshare →Pupil-size dataset specifically collected for ADHD research.
View on figshare →Pupil is pioneering a new frontier in ADHD detection. It is an accessible, non-invasive screening tool that reads attention through the eyes.
Traditional ADHD diagnosis relies on subjective behavioral assessments and lengthy clinical evaluations. Pupil offers a different approach, using the eyes as a window into cognitive function. By analyzing how people track visual stimuli, our system detects patterns linked to attention disorders in minutes, not months.
Built by researchers passionate about bridging neuroscience and technology, Pupil aims to make early ADHD screening available to schools, clinics, and families worldwide.
Start a screening in minutes. No appointment, no waiting room. Just you and the science of the gaze.
Start a screening