SacAI

Smartphone Eye Movement Biomarkers

Early cognitive signals, captured through the eyes.

Antisaccade testing and phone-based gaze analysis for MCI and Alzheimer’s research.

Run Test Explore Science

Why It Matters

A lightweight window into executive function.

Antisaccade performance reflects the ability to suppress a reflexive eye movement and redirect gaze. SacAI turns that task into a repeatable web experiment.

Problem

Early cognitive screening is often expensive, time-consuming, or difficult to repeat at scale.

Signal

Eye movement inhibition, latency, and error patterns can reveal subtle changes in cognition.

Approach

A web task, a phone video, and calibration data become structured gaze trajectories for analysis.

How It Works

From screen task to gaze CSV.

01

Sync

A white pulse aligns phone video with web events.

02

Calibrate

Five screen points map eye features into monitor coordinates.

03

Challenge

The subject looks away from the sudden stimulus during antisaccade trials.

04

Trace

Python analysis exports trial-level gaze paths and summaries.

Platform

Click through the pipeline.

Task

Measure inhibition.

Present a sudden cue, ask for the opposite gaze response, and record precise event timestamps.

Benefit

Designed for repeatable research.

No special hardware

Run the task on a monitor and record with a phone camera. No infrared eye tracker is required for early experiments.

Repeatable protocol

White-screen sync, five-point calibration, validation, and trial timing follow the same web-based sequence every time.

Trajectory-first analysis

Visualize gaze paths before reducing them into summary features, making failure modes easier to inspect.

Research-ready exports

Generate CSV outputs for statistics, model training, calibration validation, and trial-level review.

Team

Built by engineers close to clinical need.

Gijae Ra AI · System Development

Leads the eye-tracking task system, gaze analysis pipeline, and AI modeling direction.

  • KAIST mechanical engineering and biological sciences background
  • 3D navigation and bioengineering research experience
  • YOLOv8 lane detection and autonomous system development
  • Python, MATLAB, ROS, and C implementation experience
Jaehwan Jung Data Collection · Embedded Systems

Designs practical data collection workflows informed by direct care-center experience.

  • Hands-on experience with dementia care environments
  • Startup and prototype development experience
  • Morin lab and APTF lab research experience
  • Led an NYU Parkinson’s diagnostic device project
  • Python, MATLAB, ROS, and embedded system development

Contact

For research collaboration and pilot studies.

hwan050206@kaist.ac.kr rgj0410@kaist.ac.kr