Data, devices and diseases: How AI is reshaping patient care

Can AI-driven solutions realistically address India’s chronic shortage of healthcare capacity, bridge affordability gaps, and improve preventive care—or are expectations outpacing reality? These are some of the questions plaguing technologists and healthcare professionals today. A panel of them at our webinar recently, conducted in association with the MeitY Nasscom Centre of Excellence for IoT & AI, spoke of real-life applications that are already at work in hospitals and clinics, tackling everything from chronic diseases such as diabetes to the logistics of running a modern medical centre.
Their outlook was candid yet hopeful: generative AI will not solve every problem overnight, but it could fundamentally change how the country addresses capacity constraints, healthcare affordability, and preventive care.
One of the panellists, Vishnu Vardhan, an orthopaedic surgeon turned entrepreneur, has been employing large language models to scale up diabetes reversal programmes. Instead of focusing on one-off appointments, he collects continuous glucose monitoring data as well as details on patient lifestyles, diets, and genetic markers.

“We’re dealing with endless streams of complex data,” he said. “Doctors would spend most of their consultation simply flipping through charts. With AI, you can collate all that information, highlight the key concerns, and let doctors spend their time advising and counselling rather than scanning endless reports.” He believes this approach is essential in a country where diabetes is so widespread that no number of new hospitals or clinicians could cope with the demand on their own.
He also described a bold ambition: the concept of an autonomous, AI-driven hospital. “It’s being built in the United States for a huge sum,” he said. “Everything is automated, from robotic beds that monitor patient vitals in real time to large language models that interpret lab reports. It’s a multi-year project, but I’m convinced it can be adapted to India. Our infrastructure is different, our costs are different, but in principle, the concept remains valid: use AI to offer immediate assistance to clinicians and patients, reduce manual oversight for routine tasks, and free up staff to handle the human side of medicine.”
Need to unify data
Anil Kumar Madugulla, senior director and site leader at Baxter’s R&D centre in Bengaluru, pointed out that the power of generative AI lies in its ability to unify data from multiple sources. “Healthcare devices generate enormous volumes of information,” he said. “Infusion pumps, ventilators, monitors – each produces data in its own silo. The key to improving patient outcomes is integrating those streams so that doctors and nurses see one coherent picture of the patient’s status.”
While conventional AI can make sense of structured data, Madugulla indicated that LLMs bring reasoning and context to the next level. “We can now interpret clinical notes, lab reports, and real-time vitals together. That’s the kind of interoperability that’s previously been missing.”
Still, he cautioned that healthcare is a stringent, regulated space. “We can’t rely on half-baked or probabilistic approaches when a patient’s life is on the line,” he said. “We need robust validation, which is why you’ll see a lot of generative AI applications first appear in low-risk, administrative functions. But we’re already piloting solutions for sepsis management and therapy guidance. The ultimate aim is to spot red flags earlier. That’s a massive step forward, especially in an ICU, where seemingly small delays can have life-threatening consequences,” he said.
Meanwhile, Girish Raghavan, CTO for women’s health & X-Ray at GE Healthcare, has been overseeing projects that show how generative AI might alleviate bottlenecks in imaging. “We found that radiologists spend an inordinate amount of time just dictating or typing up preliminary observations,” he said. “With generative AI, the system can produce an initial report. The radiologist verifies and edits instead of starting from scratch, saving time and cost.” Raghavan noted that while this can work wonders for busy imaging centres in large cities, the bigger opportunity may lie in smaller towns and rural settings. “If we can create AI-assisted pathways, then a single radiologist could effectively serve multiple locations. That’s how you bring quality care to tier 2 and tier 3 cities,” he said.
Practical applications
Balaji Uppili, chief delivery officer at GAVS Technologies, stressed the importance of immediate, practical applications. “We’re already using AI to identify hospital-acquired infections by combining anonymised patient records with telemetry from patient beds,” he explained. “We’re seeing accuracy rates of about 75%, which is a big leap from where we were. Once hospital administrators see that, they become more open to expanding AI projects.” He also referenced how AI cuts operational overhead. “Take something like shipment logistics for a pharmaceutical giant: an AI model can gauge whether a package needs to be shipped by air or sea, saving both time and cutting carbon emissions. These might not be glamorous applications, but they save money and build confidence in AI as a whole,” he said.
To address the concern of biased outcomes – AI that might work well for certain populations but not others – he noted that diverse, high-quality data are essential. “If you only train on Western datasets,” he said, “you might be missing the local pathogens or genetic factors more prevalent in India. We need to ensure that each model is fed representative, relevant information.” For a country as vast and varied as India, that means collecting data from multiple regions and languages, a task that is much easier said than done.
Building an innovation ecosystem
Sanjeev Malhotra, CEO of the MeitY Nasscom Centre of Excellence for IoT & AI, is one of those working to overcome these obstacles. “We’re building an innovation ecosystem that connects startups, big healthcare firms, and government bodies,” he said. “Plenty of small companies have brilliant ideas, whether it’s a voice-based triaging tool or a chatbot that lets patients ask questions about their prescriptions in their local language. But they need regulatory guidance, access to domain experts, and sometimes pilot programmes in actual hospitals to test these ideas. We offer them that platform.”
He cited an example of an AI-based conversation assistant that prompts doctors with questions to ask while examining a patient. “It’s tested in a major hospital, and clinicians say it ensures they don’t miss something in a rushed consultation,” Malhotra said. “For the patient, it’s also a relief because they can ask any number of follow-up questions without feeling hurried.”
Malhotra further highlighted how local deployments can significantly improve basic healthcare access. “We’ve run pilot projects where AI chatbots handle first-level queries,” he said. “If a symptom triggers a red flag, the system immediately arranges a remote consultation with a human doctor. It’s a hybrid model, but crucially, it helps people in rural areas avoid unnecessary travel and expenses. You’re saving time for doctors, too, so they can focus on serious cases. That’s the real promise of AI – helping both patients and clinicians do more with less.”
He also stressed the importance of local partnerships to achieve results. “Every region has its own set of challenges, from TB to malaria,” he said. “We encourage startups to forge ties with local governments and adapt their tools to those realities. A solution that works in one state might need a tweak to succeed in another. By tailoring AI to local contexts, we ensure broader adoption and better health outcomes,” he said.
But regulation remains a chief concern in scaling these innovations. “We can’t simply deploy fully autonomous AI diagnostics,” Malhotra said. “The guidelines don’t allow for that yet – and rightly so, because we need to build trust through smaller steps. AI can guide a clinician’s decision or help patients understand a prescription, but the final say still rests with qualified medical professionals. Over time, as success stories accumulate and regulatory frameworks evolve, I see a clearer path to scaling AI-based care across India’s vast geography.”
Gradual build-up
One cannot avoid the question of whether any of this technology will truly reach the next billion Indians who are most in need of healthcare. That’s where the panellists sounded a collective note of realism. Vardhan explained that scaling up diabetes reversal programmes in rural areas demands not just AI but also adequate internet connectivity, a proper referral network for more complicated cases, and local staff trained to handle the system. “You do need a certain baseline infrastructure,” he said. “But if you have a smartphone and a data connection, you can do a lot of early diagnostics and continuous monitoring with AI as a backbone. It’s far cheaper than building a major hospital in every taluk.”
Raghavan said continuous incremental improvements can add up. “We won’t wake up tomorrow and find that every hospital has become a fully autonomous facility,” he said. “But piece by piece, you see billing, scheduling, service logs, and preliminary diagnoses being automated. That unburdens clinicians and staff, making more room for human empathy. As India’s population continues to urbanise, and as lifestyles drive higher incidence of chronic conditions, AI tools become all the more critical to keep up with the workload.”