Visual acuity tests, commonly conducted in clinics and used for health screenings, are becoming more in demand due to an aging population. Current online self-eye check apps are limited as they don't accurately reflect true distance vision assessed in clinical settings. These tests, performed by trained personnel, are time-consuming and can cause delays in clinics. This project aims to develop an automated Visual Acuity (VA) station using AI technologies like speech-to-text and computer vision, hypothesizing that it can match the accuracy of manual assessments by clinic staff, thus potentially reducing waiting times and improving efficiency.

Visual acuity is done as a routine eye check for the majority of eye patients in the clinic. It is also done as a screening test for pre-employment health checks and health screening. Patients can be checked for refractive errors, on a community level or screened for eye diseases, for those with chronic medical conditions. With the increasing burden of aging population and eye conditions, the number of patients in eye clinics will increase.

There are a few existing online applications that allow self-eye checks, however there are limitations. They are usually done at an intermediate distance, i.e. distance from phone to eye and does not accurately represent true distance vision. Distance vision is typically set at 4- 6m in a clinical setting.

A visual acuity test is administered by specially trained healthcare personnel, such as optometrists and patient service assistants, which is often time-consuming and labour intensive, where one-on-one attention is required. In addition, vision is subjective and re-testing may be required at times to ensure accurate vision assessment.

As the visual acuity test is the first clinical station patient goes to after registration, this leads to a bottleneck in workflow causes delays in the subsequent services and eventually increases patient waiting times in the clinics.

This project aims to develop and validate an automated Visual Acuity (VA) station through speech-to-text and computer vision technology in comparison to existing manual VA assessments.

We hypothesize that we are able to use artificial intelligence to understand patient's speech and posture to automate the visual acuity test. We also hypothesize that the automated visual acuity test is comparable to having VA checked manually by a clinic staff.

Inclusion Criteria:

1. Patients age >21 and able to give consent
2. Patients who have at least counting finger vision
3. Patients who is able to speak in an audible and clear voice
4. Patients who is able to use a digital device independently (e.g. handphone)

Exclusion Criteria:

1. Patients on wheelchair/ walking aids
2. Patients with hearing difficulties
3. Patients with speech difficulties
4. Patients who have cognitive impairment
5. Patients who are hemiplegic/ motor dysfunction
6. Patients who have vision worse than counting fingers
7. Patients who are pregnant