Eyes Wise Open

Inspiration

Around 30% of the UK struggles with dry eye disease, and it is especially prevalent amongst office workers or people who spend a lot of their time looking at screens. A lot of people also don't know about it, or that they might be facing it as well! Amongst this demographic was us: STEM university students. Two of us have struggled with dry eyes. It is seriously exacerbated by the reduction in blinking that happens when we work on our computers all day and causes headaches, exhaustion, and burnout. And having to take breaks every 10-20 minutes when we were very busy was a challenge! So, we built Eyes Wise Open to make us blink more so that we can take fewer breaks and feel better and healthier.

What it does

Eyes Wise Open tracks the user's blinks to calculate their blinking rate. It can be used for a few minutes every day to track how blinking rate changes over time (so the progression of dry eyes over a month or year) or for a single longer session. In a single extended session, notifications will pop up over reasonable intervals to let the user know if their blinking rate is too low and if they need to blink more. Eyes Wise open also tracks whether a blink is incomplete (another symptom of dry eyes) and informs the user of this.

Individuals can also use Eyes Wise Open to view graphs of the change in their blinking rate over all of their sessions and compare it to the healthy levels (based on reviewed scientific papers and surveys) or compare a single one of their sessions to an average healthy level. Companies can use Eyes Wise Open to look at the dry eye progression of their employees (with their consent) and get suggestions on what future steps might look like.

How we built it

We shared the tasks and ideas between us and built the product with Claude Code in VS Code. Rebecca and Manavi are beginner developers, and this is their first hackathon. Shahida has more experience, so she guided them through the stages of a hackathon: scoping the idea, getting one working loop first, deploying early, then adding extras. Feedback from a Wise mentor changed our direction, from a break reminder to a one-time company eye health check.

The check. The front end is React, Vite and TypeScript. MediaPipe Face Landmarker runs in the browser, with its model files hosted by us, and counts blinks and flags partial ones. Each person calibrates first, and the app warns about low light or a lost face instead of guessing. Video is never recorded or uploaded. Only anonymous numbers leave the browser.

The employer side. People can choose to share an anonymous result. A small Express server stores these, and a manager view at /manager shows the team picture. A rule-based team agent (plain code, no model) gives its call on the team, its confidence and a recommended action for the priority the employer picks. It updates when new responses come in, when the priority changes, or when the employer corrects an assumption, and it shows what changed and why.

Evidence. Our reference blink ranges come from the American Academy of Ophthalmology and published studies, and we flag where published figures differ.

Sponsor tools

Gemini API (gemini-3.5-flash-lite): writes the short result summary and an opt-in trend explanation from anonymous numbers only. Each is labelled as written by Gemini or by a template.
Vultr: hosts the live site at eyeswiseopen.health, behind Caddy for HTTPS, with a GitHub Actions workflow that redeploys on every push.
incident.io (Adaptable Agents): the team agent summary / suggestions.

AI use. Claude Code (Sonnet 5.5 and Opus 5.5) wrote the first drafts of the code and ran builds. Google AI Studio only created the Gemini key. Every task is logged in AI_DISCLOSURE.md. [Add one sentence on what you changed or decided yourselves.]

Challenges we ran into

One of the biggest challenges we had was narrowing down the design of our exact product. Our focus remained on dry eyes throughout, but we contemplated building the product to automate the entirety of the DSE compliance process to make it more effective and actually look after employee health year round. However, the many components of DSE processes range from posture to fire safety and training required many features, and we ended up losing our focus in the haze. We lost track of who our product was intended for and how exactly it would work. If asked to describe our product in a single sentence, we wouldn't be able to. However, after discussions with Wise mentors and others around, we realised that our automation would not be a very effective solution unless built in collaboration with the specific company and decided to focus on the major issue of dry eyes specifically. Fortunately, we had enough time to build a much more coherent product that specialises on dry eyes and produces more comprehensive, understandable, and usable results!

Accomplishments that we're proud of

We're very proud of all the extensions we have integrated into the product. Eyes Wise Open uses Google Gemini API to produce simple explanations to help the user understand the product. We're very happy with it because while Eyes Wise Open can definitely be used by companies to help improve worker productivity and health, the Gemini-based explanations can help a user sort through all the overwhelming information that comes with learning about any health-related issue. The explanations do not provide any medical advice, making them safer, and can help the individual user (maybe a university student, like us!) understand their specific results.

Another extension we're very happy with is the Google product we used for our blink tracker, because it gives very accurate blink count rates and incomplete blink rate percentages. The product we used was MediaPipe Tasks Vision (FaceLandmarker), and we're happy with how seamlessly and bug-free it has been integrated into our product (per our testing!).

What we learned

We learnt to focus on one use case rather than scope. We changed our project trajectory mid-hakathon to focus on the dry eye issue rather than automating the DSE assements, this meant that our product will be suited for a wider audience from companies, individual workers seeking to imporove their eye health to university students. We also learnt to accomodate for edge cases such as poor lighting, users switching to websites mid-session or being in a call.

What's next for Eyes Wise Open

There are many opportunities to extend our work such as providing more personalised helath tips by incooperating local weather data (pollen for allergies) and travel plans such as flights. We also hope to help companies automate DSE check such as doing realtime posture and screen distance checks. Another feature we hope to add is a personalised chat bot that helps provide holistic, year-round eye care and also educational material on general eye health (such as complete vs incomplete blink, eye hygiene, how to manage stress and fatigue).