How it Works

Our Method

1. Experimental Setup

Ocular administers a 20 minute visuospatial working memory task. For each trial, a patient is presented with 3 dot arrays, followed by a distractor image and probe. This serves as a mechanism to assess attentional performance.

2. Realtime Biometrics

In realtime, Ocular sends images of a patient's pupil through an optimized convolutional neural network (CNN) to detect the iris, produce a segmentation map of the pupil, and perform semantic segmentation, resulting in precise biometrics.

3. Machine/Deep Learning Backend

Biometric data is sent to a backend server and inputted into an optimized machine learning algorithm to analyze the millions of entries of time-series data.

4. Medical Advisement

A detailed summary of pupil dynamics is presented, along with biometrics and a probability of having ADHD. Moreover, an intelligent AI-based system outputs future steps for patients to take — whether a patient should see a neurologist, consult a medical practitioner, or seek immediate attention.

Capturing the Pupil

Ocular capturing and segmenting the iris and pupil in realtime.

Experimental Setup

Landing screen for administration of the 10 minute visuospatial working memory task.

Visuospatial Working Memory Task

A dot array being presented per the visuospatial memory task administered to patients.

Detailed Results

A comprehensive overview of biometrics given to patients, along with a risk score of having ADHD.