Hyderabad researchers develop offline AI to detect Glaucoma on smartphones

Hyderabad-based LVPEI and Remidio have developed Medios HI, an AI platform that detects glaucoma and other eye diseases using a smartphone without internet access. The system, recently approved by CDSCO, showed over 91% accuracy in clinical trials.

Hyderabad: A health worker in a remote village in Telangana with no internet connection can take a photo of a patient’s eye using his smartphone. In just a few seconds, the phone will tell the worker if the patient has glaucoma, diabetic retinopathy, or even age-related macular degeneration (AMD).

This is no longer a futuristic concept; it is a real medical technology, which is expected to clear all regulatory hurdles in the coming days.

---

The novel technology, developed through a high-impact collaboration between clinical experts at Hyderabad-based LV Prasad Eye Institute (LVPEI) and tech innovators at Remidio Innovative Solutions, is a massive leap for the concept of ‘offline’ artificial intelligence (AI).

While most of the AI platforms depend on powerful cloud servers and high-speed internet to analyse and come-up with results, the researchers from LVPEI and tech experts from Remidio have developed an AI that runs independently, without the need of the internet, to accurately diagnose glaucoma.

Dubbed as Medios HI (Humanising Intelligence), the AI platform runs entirely on the smartphone itself. Dr Sirisha Senthil, Head, Glaucoma, LVPEI and Dr Anand Sivaraman of Remidio collaborated in validating the software.

A few days ago, the AI platform received regulatory approval from the Central Drugs Standard Control Organisation (CDSCO). The AI doesn’t just look for blurriness of the eye; it analyses the optic nerve head and retinal nerve fiber layer using deep learning models trained on hundreds of thousands of real-world images.

In the clinical trials that were conducted at LVPEI, the AI system demonstrated over 91 per cent sensitivity, performing nearly as accurately as a human specialist sitting in a tertiary care center.

Under the leadership of Dr Sirisha, LVPEI provided access to a massive repository of high-quality, retinal images required to ‘train’ the AI algorithms. Later, clinical validation studies were taken up to ensure the AI’s accuracy matched that of experienced ophthalmologists.

While it is traditionally difficult to diagnose glaucoma, LVPEI’s clinical expertise was essential in teaching the AI to identify these specific structural changes in the eye.

The AI technology was also deployed across the LVPEI network of Vision Centers in rural Telangana and Andhra Pradesh, allowing researchers to test it in real world conditions and prove its ‘offline’ abilities on a smartphone.

A smartphone can now detect glaucoma

The phone is loaded with AI that can detect the ailment

No need for the smartphone to be connected to internet

The accuracy in detecting glaucoma, AMD and diabetic retinopathy is over 90 per cent

The AI platform, named Medios HI, was developed by Remidio and LVPEI