.jpg)
.jpg)
.png)
Existing mental fatigue tools don't work outside a lab. Questionnaires are subjective. EEG and HRV require clinical setups. Neither scales, neither takes under five minutes, and neither gives you a number you can act on. NeuralPort had a different approach: gaze behaviour and blink patterns are direct markers of cognitive state and unlike self-report, they can't be consciously faked because your eyes does not lie.
A project that started as academic research at Osaka University, spun off into a company in 2020. By late 2022, they had a working technical prototype , proof the science could run on a VR headset.That's where I came in. I joined in December 2022, at beta launch.
The challenge was building a VR product that could extract that signal cleanly, at speed, in real-world conditions. That required UX that didn't get in the way.
Why understanding stress / tension some people call it matter ? Stress is a natural reaction of the body, resulting from demanding circumstances where the brain activates the “fight or flight” mode. In sports, one application of NeuralPort is to find the optimal level of stress that triggers the brain to increase the level of fight to be able to compete effectively. By identifying this level and correlating it with actual on-field performance, athletes can have personalized training that ensures they maintain that optimal stress level for peak performance before entering the field. At the moment, we’re using the Pico 4 Enterprise headset for Zen Eye Pro.
The brief was clean data collection. That meant earning trust from the first screen, an experience calm enough to relax into, clear enough that nothing prompted a second thought, short enough that there was no time to tense up. If the user isn't settled, the measurement is already compromised.
Delivered across the full pipeline as the XR UI/UX Designer working directly with a developer: interaction model and spatial UI designed in Figma, spatially prototyped in ShapesXR, handed off in Unity.
.png)
Design Principle
In most UX, you design for engagement. Here we design for the opposite. The more present the interface feels, the worse the product performs. Before any design decisions were made, the process started by mapping what the user would actually experience , not as a task flow, but as a series of physical moments.
.png)
Three friction points emerged from that design thinking the environment they entered, the time they had to think, and what they were shown at the end and with that we decided to make a user flow to remove any possible frictions and thinking from the experience.
A flow designed to prevent second-guessing
Home → Explanation → Measurement → Assessment. The measurement window was fixed by the science team ,700ms stimulus onset, 5 seconds active, 800ms close. My job was designing everything around that constraint: an Explanation screen tight enough that users arrived at the measurement phase settled, not mid-thought, and a transition out that didn't fracture the experience. Four steps because five would give users too long to think. Under two minutes end to end.
.png)
UI Approach : Gaze Interaction for User Comfort and Precision
To ensure comfortable gaze interactions, The ±60° comfort zone maps to the natural range of comfortable horizontal eye movement without head rotation ,beyond that, users compensate physically. Any strain or micro-adjustment changes their blink state. The 1.5m interaction radius keeps all elements within comfortable focus depth, removing any need to refocus or lean. Key Findings from Testing:To maintain comfort and prevent strain, the following should be avoided:
- Gaze angle greater than 10° above the horizon
- Gaze angle greater than 60° below the horizon
- Neck rotation exceeding 45° above the horizon
The result screen as an information design problem
The output is a brain fatigue percentage from a segmented AI model that behaves differently depending on which fatigue cohort the user falls into — at low fatigue, longer blinks indicate relaxation; at high fatigue, the same signal means worse cognitive load. The model segments before it interprets. A flat readout would have been wrong for a significant portion of users.
The design decisions visible in the screen: The label reads "Current Score"and not "Fatigue Score," not "Stress Level." A deliberate framing choice for a data point that will be actionable with overtime usage.
.png)
.png)
Prototype in ShapesXR
.png)
High Fidelity Product
.png)
.jpeg)
.png)
What the prototype manage to confirmed
Validated with 61 participants in real-world conditions (not lab) with a mean score of 44.20%, matching Japan's population average of 44%. The tool detected statistically significant fatigue reductions from both a 5-minute mindfulness session (−18.85%, p<0.001) and a 2-minute aromatherapy session (−14.47%, p<0.001). The fact that characteristic gaze patterns emerged without strict experimental control confirmed the approach works outside controlled environments , which was always the point. The paper is now published in VR Medical Science (VR医学), Vol. 25, No. 1, 2026. , here
900+ assessments
performance monitoring, 2026 season
clinical POC, paediatric patients, pre-operation anxiety reduction
Expanding into ZEN-EYE performance glasses
What I Bring to This Type of Brief
The ability to design in domains with no UX precedent , where the constraints come from the science, not only design system, and where the interface has to hold up under conditions the user can't control. If you're building XR where the stakes are real and the design problem isn't obvious yet , count me in to be part of that.
.png)
.png)