Generative AI holds significant promise for India, given its cultural and language diversity. At the AWS Summit, Washington DC, Rowland Illing, the director and chief medical officer for global healthcare and non-profits, AWS, acknowledged the same.
“The appetite for generative AI in India is high. The technology can help with language and culturally sensitive information, addressing the breadth of languages in India. It’s exciting to see opportunities in the National Digital Health Mission as well,” he told AIM.
Illing’s sentiments resonate well since many Indian healthcare companies are actively tapping into the potential of generative AI. Today, every health company is turning into an AI company.
For example, Apollo partnered with Google Cloud to improve its AI-powered Clinical Intelligence Engine (CIE) using MedLM and retrieval-augmented generation (RAG), integrating extensive clinical data for improved decision-making.
While Narayana Health, in partnership with Microsoft, is making diagnostic accuracy and operational efficiency better, Manipal is personalising cancer care through imaging and predictive analytics with AI.
On the startup side, apart from Qure.ai, Khosla Ventures-backed Healthify provides custom diet and fitness plans via its generative AI assistant Ria. The team uses solution models based on diverse statistical models, including AWS, Meta, and OpenAI.
“I have a great respect for the Indian health-tech ecosystem. I think it’s an incredibly rich and exciting space to be in. So, we’ve got great health-tech companies that are already building on AWS across India,” said Illing, emphasising the work of Bengaluru-based startup Qure.ai in medical imaging AI.
AWS is closely working with many widespread Indian companies, such as the Center for Cellular and Molecular Biology in India, the Poshan Tracker, and the CoWIN vaccination platform. During the pandemic, CoWIN was built natively on AWS, supporting 2.2 billion vaccination doses across the country.
Beyond the functioning of AWS in India, Illing also shares another special connection with India. The Oxford University alumni completed his undergraduate paediatric training at St John’s Medical College and Hospital, Bengaluru, almost two decades ago.
Amazon Bedrock for Health
Meanwhile, Genomics England, a UK government-owned entity, is leveraging Claude on Amazon Bedrock, AWS’ fully managed generative AI service, to assist researchers in linking genetic variants to medical conditions. Illing told AIM that it resulted in the discovery of 20 new gene associations with learning disabilities, providing actionable insights for future research and treatment.
This research, based on peer-reviewed articles, aims to improve genetic tests and health, especially for intellectual disability, by quickly processing vast data to identify potential gene associations.
Genomics England used Anthropic’s LLM Claude, available through Amazon Bedrock, to make this happen. Interestingly, Amazon concluded a $4 billion investment in the AI startup in March of this year.
“We have a great relationship with Anthropic. We’re invested in them, and we get early releases of their models, like Claude 3.5 through Bedrock,” he added.
“Bedrock is a great vehicle because it allows users to access foundation models through APIs and provides functionalities like guardrails to ensure LLM answers are bound by constraints. Users can enrich foundation models with their data without sharing it with the model.”
However, this is just one example, as numerous companies leverage Bedrock’s diverse models from various AI providers like Anthropic, OpenAI, and Meta, to meet their unique needs, supporting AWS’ mission to democratise foundational models.
The model evaluator function allows users to compare the models based on speed, cost, performance, or accuracy. “We believe there won’t be one foundation model to rule them all but hundreds, if not thousands, giving users a choice in a secure way,” commented Illing.
“The current trend we are observing is the explosion of analytics and generative AI. This is an end-to-end data story, focusing on extracting meaningful insights from vast amounts of data, often multimodal. Generative AI plays a crucial role in understanding and analysing this data.”
At the AWS Summit Bengaluru in May, the company announced that Amazon Bedrock is now generally available in the Mumbai (Asia Pacific) region to allow large organisations across India to build and scale generative AI applications with enterprise-grade security and privacy.
Since 2017, AWS, the cloud subsidiary of Amazon, has trained over 5.5 million people in India on AI and cloud skills and over 8.3 million across the APAC and Japan regions.
With generative AI moving from proof of concept to deployment this year, the company plans to train about two million individuals in AI globally by 2025 through its “AI-Ready” initiative.
Shifting to Cloud
Apart from generative AI, there is a global shift happening towards cloud adoption in healthcare, especially after the COVID-19 pandemic, which needed remote work. Despite a decrease in global IT spending in 2020, cloud spending grew by over 6%, reaching $258 billion, with experts predicting the market will double in the coming years.
“Previously, the cloud was used for disaster recovery and backup, running things on-premises. However, with security being a top priority and concerns about outages, bad actors, and ransomware attacks, we are seeing a shift in organisations moving their production clinical environments to the cloud,” explained Illing.
This involves using clinical workloads with patient-identifiable data in a cloud-read-only mode initially, before running things on-premises. Over time, organisations have found that cloud-based systems often outperform their on-premises counterparts, leading to a complete transition to cloud-based production environments.
As per Nutanix’s report, the future of healthcare relies on moving away from legacy systems. Currently, 27% of healthcare companies use only traditional, non-cloud data centres and business systems, more than any other industry.
Several notable examples illustrate this trend. The world-renowned academic medical centre, Tufts Medicine and Danville-based Geisinger Health in the US, now run its electronic healthcare records (Epic) natively on AWS.
Similarly, New South Wales in Australia is migrating its entire eHealth system, including the single digital patient record for the state, onto AWS. Apart from new startups, legacy health giants like Pfizer, Amgen, and Merck, are also building drug discovery platforms on AWS.
More recently, AWS chartered into protein folding. It has partnered with the NY-based research team EvolutionaryScale to bring the latter’s ESM3 multimodal large language models available on AWS, aiding applications from drug discovery to carbon capture.
