Google licenses its AI model to help Indian firms detect early blindness in diabetic patients

Google unveils AI-led collaborations for India’s health, sustainability and agriculture sectors

Google on October 17 unveiled new partnerships and collaborations across India's healthcare, sustainability, and agriculture sectors that will utilise the company's artificial intelligence (AI) research and models.

"At Google, we're not just building AI, but are helping shape a future where AI benefits everyone. Collaborating with key Indian organisations, our focused research in India across language understanding, healthcare, agriculture and sustainability is helping tackle many of the country’s unique challenges and creating AI-led solutions that will improve billions of lives," said Dr Manish Gupta, Research Director, Google DeepMind.

The tech giant is partnering with Bengaluru-based Forus Health and AuroLab, the manufacturing division of Madurai-based Aravind Eye Hospital, to help scale screening for diabetic retinopathy, an eye condition that can cause vision loss and blindness in people who have diabetes.

As part of this partnership, Google is licensing its diabetic retinopathy AI model to Forus Health and AuroLab along with Thailand's Perceptra to support around six million AI-assisted screenings for diabetic retinopathy in resource-constrained communities in India and Thailand over the next 10 years, at no cost to patients.

"Each of these partners will own, operate, deploy and maintain the AI model, and they will work with their local regulatory bodies to obtain necessary approvals and clearances to expand the program further at a population scale" said Sunny Virmani, Group Product Manager for Health AI Research at Google in a media briefing.

Google's AI model aims to address preventable blindness amongst patients with diabetes, helping clinicians with early detection at scale (Image: Google)

The model aims to address preventable blindness among patients with diabetes, helping clinicians with early detection at scale. Early research and deployment of the model was conducted in India and it has already helped support more than 600,000 screenings in clinics around the world, the company said.

"It has been great to see the multiple phases of Google’s diabetic retinopathy AI model - from development to validation to deployment. At AuroLab, we are hopeful that this AI model will play a crucial role in preventing unnecessary vision loss for millions of people," said RD Sriram, Managing Director, AuroLab in a statement.

Forus Health founder K Chandrasekhar said he is confident this tie-up will "impact millions of lives" and fulfil their mission of eliminating preventable blindness.

Read: Google's AI model to help researchers identify TB, other diseases from the sound of cough

Collaboration with Saahas Zero Waste

Saahas Zero Waste (SZW), a Bengaluru-based environmental and social enterprise, is leveraging Google's open-source machine-learning computer vision model CircularNet to sort plastic waste and improve recycling, thereby reducing strain on landfills from recyclable waste.

CircularNet is powered by TensorFlow, the company’s free and open-source software library for machine learning and artificial intelligence, which has been trained on global datasets.

It provides pixel-level instance segmentation, serving as a vision-based quality control system for Saahas Zero Waste. This system checks the quality and quantity of waste before it is sorted, baled, and sent to recycling centers, as well as when it is received at those centers.

Read: Google woos India's booming AI developer community with new tools, access to latest models

In a pilot of CircularNet’s AI model at its material recycling facility, Saahas Zero Waste estimates that it realised around 85 percent accuracy in detection of plastic waste. The startup hopes that the higher quality recyclable materials that are recovered from this process could translate into a 10-12 percent improvement in revenue generation.

Additionally, Saahas Zero Waste estimates that roughly 90 percent of recyclable waste could be diverted from landfills.

"Our collaboration with Saahas Zero Waste demonstrates the potential impact of circular ecosystems, where valuable resources are reused instead of being discarded...By enabling more efficient recycling and a truly circular economy for plastics, together we can work towards a more sustainable future for communities in India and across the world," said Sujit Sanjeev, Lead, CircularNet, Google.

Arun Murugesh, VP Sales and Marketing, Saahas Zero Waste said, "Saahas Zero Waste is committed to the evolving waste management ecosystem in India, aiming to responsibly manage over 500 tonnes of waste daily by 2026. Google’s CircularNet model is playing an important role by supporting automated AI-powered quality control systems that can be deployed efficiently at scale."

Google is also opening up access for developers to its Agricultural Landscape Understanding (ALU) Research API that uses remote sensing and AI to provide landscape insights at individual farm level across India.

The API was first announced at the company’s I/O Connect Bengaluru developer conference in July. It aims to address various challenges faced by farmers, such as accessing subsidies and capital to improve yields and market access.

ALU API, which was developed through collaborations between Google’s AnthroKrishi team and researchers and state and union governments across India, combines high-resolution satellite imagery with advanced machine learning models to identify fields, water bodies and vegetation boundaries and their acreage.

These capabilities can enable the development of precision agriculture tools, optimising resource allocation, and improving farm management practices. It can also assists more data-driven and efficient decisions across India's agri ecosystem.