Connected eyes
summary
Machine learning models trained on roughly 404 patients, 747 procedure instances predict uncorrected visual acuity at day 1, week 1 and month 1 after LASIK and are deployed as a cloud web service for clinics to call per patient. Models use preoperative exams, demographics and surgical parameters; age is the strongest predictor. Reported accuracy is about 0.07 logMAR day‑1 error, 0.06 week‑1, and the workflow also extends to forecasting astigmatism (114k visits, 51k patients), corneal graft survival and diabetic retinopathy screening.
AI-written summary · about
chapters (8)
- 0:00Start
- 1:33Predictive Eye Care
- 2:07Thin Flap Is Created
- 2:17Gently Folded Back
- 2:29Flap is Folded Back
- 4:30The Lasik ML Pipeline
- 6:22Machine Learning for Corneal Transplants
- 8:17Diabetic Retinopathy Prediction
related talks
05mAI for Ophthalmology
01h 03mUsing Machine Learning and AI to serve people and communities
23mEngineering AI-Powered Predictive Healthcare Systems
17mAI Innovation in Health
21mThe eyes of industry 4.0 - Vision intelligence and machine learning
24mDeveloping Innovative Systems Through Responsible AI In Healthcare
36mResponsible AI
30mTurning Glasses into Smart Assistants: Open Source Vision Language Models