Retinal Analytics via Machine learning aiding Physics (RAMP)
Glaucoma occurs at least partially as result of the eye's mechanical behaviour, in particular the elevated intraocular pressure influencing the structural integrity of the sclera and the transport of proteins necessary for the well-being of the retinal cells. The explicit interdependencies from first principles physics to healthy vs degenerative conditions and, consecutively, successful treatment are largely unknown.
RAMP (Retinal Analytics via Machine learning aiding Physics) Intra-Create programme will look to solve the problem by using novel biomechanical analysis methods informed by data science, novel instrumentation and experiments in micro-structured physical emulators (phantoms). In addition, it will look to understand and quantify the links between first principles and pathophysiology which will be beneficial for both prevention and treatment of glaucoma. We will also anticipate major impact on other problems in ophthalmology and more generally on biology and medicine incorporating first principles aided by data science.
The primary objective of RAMP is to use machine learning-aided frameworks based on Optical Coherence Tomography (OCT) scans of the optic nerve head for glaucoma progression prediction. The project will last for 36 months. The main deliverable of our project will be the algorithm for predicting a patient's visual field progression from a set of ophthalmological and standard medical tests. This progression prediction outcome will assist with treatment decisions and reduce the risk of misdiagnosis. Our approach has the potential to simplify glaucoma diagnosis and reduce some glaucoma tests such as fundus imaging, while other important glaucoma tests (such as tonometry & gonioscopy) could still be used in parallel. Our approach might facilitate the learning process of more junior doctors and alleviate the "watch and wait" approach that is the easy fall back for "disc suspects" but which also clogs up clinics and urgently needs rationalizing, or at least risk stratifying. Note that, once diagnosed, a glaucoma patient needs periodic follow-up visits with average lifetime costs of up to S$80,000. In the UK, it has been estimated that if only 10% of the glaucoma population could receive earlier treatment, it would save the UK government between S$1 and S$2 billion annually. If our work is translated in the short term, it may have a considerable economic impact worldwide.