New Zealand-designed AI framework predicts eye diseases
Tuesday, 6 November 2018
Return to eHealthNews.nz home page New Zealand-designed artificial intelligence framework MedicMind, which enables the development and design of AI without coding, has been used to develop an advance set of algorithms to detect a range of common eye diseases.
The Dunedin-based research was part of an initiative to tackle growing ageing population and degenerative eye diseases, says oDocs Eye Care director Hong Sheng Chiong.
A total of 4,435 images were used for developing the AI and its underlying algorithm. The average accuracy was 80 per cent, with sensitivity of 75 per cent and specificity of 89 per cent.
The AI is capable of detecting common eye diseases such as diabetic retinopathy, age-related macular degeneration, glaucoma, and retinal vessel disease.
Age-related degenerative diseases such as macular degeneration, glaucoma, and diabetic retinopathy are on the rise, causing huge delays and backlogs to clinic appointments and clinical workload, says Hong.
“With an AI that is capable of triage and making initial predictive diagnosis, it would help clinicians with limited knowledge and experience in ophthalmology – eye care – to screen through retinal photographs efficiently.”
“This is a pilot study and its performance can be improved with a much larger dataset. We are calling for all New Zealand clinical and artificial intelligence scientists to join the movement to explore this field.”
The research and algorithm are not designed to replace more conventional approaches such as a visit to an optometrist or ophthalmologist. It is merely a way to improve efficiency and safety of patients’ eye care.
Source: oDocs media release, 5 November 2018
Sector updates are provided by organisations to eHealthNews.nz and have not necessarily been edited or checked for accuracy. Any queries should be directed to the organisation issuing the release.
Do you have an item to add to sector updates?
Email your information to us at updates@hinz.org.nz Return to eHealthNews.nz home page
|
Email to a Friend