- Unrivalled performance: get accurate results in less than 3 seconds (see algorithm performance, below)
- An architecture tailored to any clinic
OphtAI uses “deep learning” artificial intelligence to provide recommendations on the need for patient referrals and the stage of diabetic retinopathy.
In less than three seconds, OphtAI analyzes the image of the retina, assessing its quality, mapping lesions, and evaluating the need of referring the patient to a specialist.
Depending on their responsibilities, users may complete the report with the patient’s clinical context, any other detected symptoms, and the recommended course of action.
- Sensitivity of 99,0 % with specificity of 87,0 %
- Specificity of 90,2 % with sensitivity of 96,8 %
With these operating points, OphtAI screens for diabetic retinopathy with unrivaled effectiveness: Fewer patients afflicted by the condition are sent home without a diagnosis (false negatives), and fewer healthy patients are referred to an ophthalmologist by mistake (false positives).
Our AI has been trained on close to 350,000 retinal images provided by a true network of tele-ophthalmology operated by AP-HP Paris Hospitals, and tested using several databases, including Messidor. Furthermore, the presence of diabetic retinopathy has been annotated by three ophthalmologists, and the chosen “ground truth” was defined by majority vote.
Our solution is constantly evolving. Soon, we will be releasing new versions that cover additional diseases. Get an exclusive first look.
Need more information ?
OphtAI is a Joint Venture specialized in Artificial Intelligence for Ophthalmology. Would you like to find out more about our solutions or meet our teams ?