From cancer detection to treatment, how AI is transforming health care

Artificial intelligence is changing the DNA of health care with the promise of a 'healthier human'. The second of a three-part series

In early 2024, Marly Garnreiter, a 27-year-old French woman, began experiencing persistent night sweats and itchy skin. Her doctors dismissed it as stress. Her blood report came back normal. Out of curiosity, she asked ChatGPT about it. The artificial intelligence (AI) chatbot suggested she might have blood cancer. Nearly a year later, in April 2025, doctors confirmed she had Hodgkin’s lymphoma, a cancer that affects the lymphatic system.

 

Stories like Garnreiter’s are making headlines repeatedly: AI catching what humans miss. 

 

While anecdotal, they point to a shift underway in health care — one where AI doesn’t replace doctors but sharpens their eyes and expands their reach. And also, frees up the time doctors and nurses spend on tasks like writing patient summaries, current patient condition, treatment plan, and so on. Discharge summaries and reports, too, are being prepared by AI and sent to patients, after a human has crosschecked them for errors.

 

“The future lies in human-AI collaboration, where technology enhances our diagnostic and therapeutic capabilities,” says Bharat Aggarwal, principal director, radiology services, Max Super Speciality Hospital, Saket, Delhi. The hospital chain is implementing AI in ‘shadow mode’, running predictions alongside human clinicians to verify accuracy before actual deployment.

 

“AI is reshaping the healthcare landscape at a brisk pace,” says Venkatesh Krishnan, executive director, public sector, health care, education, Microsoft India and South Asia. “Globally, 79 per cent of health care and lifes­ciences companies are leveraging AI to drive better outcomes.” In India, he says, this momentum is visible as healthcare systems navigate workfo­rce challenges, rising patient expectations, and need for personalised care.

 

Apollo Hospitals is using Azure OpenAI to streamline clinical documentation audits. The national telemedicine platform, eSanjeevani, built on Microsoft Azure, has facilitated several million remote consultations — many in rural areas.

 

AI scores in its ability to look deeper into lab samples. Its models rely on millions of past samples trained to detect anomalies a human can miss. Be it blood tests, medical scans — X-rays, MRIs, CTs, ultrasounds — intelligent algorithms can decode patterns far more efficiently than even seasoned radiologists or pathologists. This is where AI holds its most transformative promise, revolutionising health care at an unprecedented scale.

 

AI is being embedded across the value chain — from early screening and diagnosis to treatment planning, nursing handovers, medical sales, and remote consultations. 

 

Vipin Gudwani, global leader, health care and manufacturing, AIonOS, says, “The goal eventually is how AI can help drive us towards a ‘healthier human’.” Rather than just treatment or cures, this shift towards well-being will focus on prevention, faster clinical trials for new medicines and personalised AI advice, he adds. AIonOS is a Singapore-based startup founded by former Tech Mahindra CEO CP Gurnani and InterGlobe’s Rahul Bhatia. 

ALSO READ PART I: AI in currency: How it is quietly rewriting the rules of modern banking

 

Seeing what humans can’t

 

Health care has long been data-rich but insight-poor. Hospitals generate mountains of lab reports and clinical notes — but distilling those into actionable insight has been a challenge. 

 

Now, AI-powered large language models (LLMs) trained on millions of anonymised health records are bridging that gap.

 

For instance, an LLM can scan a patient’s blood test and detect subtle patterns across cholesterol, glucose and inflammatory markers that mig­ht suggest early-stage cardiovascular disease — patterns that would­n’t necessarily raise red flags in isolation. This gives doctors an advanced early-warning system, trans­­forming routine diag­n­o­stics into proactive care.

 

Take radiology. AI, trained on millions of images, can spot a nodule the size of a pencil tip and even predict if it’s malignant or benign. 

 

At Max Hospitals, AI embedded in bedside portable X-ray machines (used in critical-care situations) flags signs of pneumonia, pneumothorax (leaking air in the lungs), or fluid accumulation in the lungs. “AI tools act as pre-readers,” says Aggarwal. 

 

All new medical machines (including X-ray, CT scanners, MRI, ultrasound) that Max is buying include an embedded AI reconstruction model and computer vision-based AI to interpret the patient’s medical condition in real time.

 

“Any alert given by AI is validated and vetted by a radiologist, without which treatment is not given,” Aggarwal says, adding, “AI is assisting but not entirely replacing the opinion of a human expert.”

 

Healthcare professionals have been slow to shift to tech tools, but the Covid-19 pandemic has changed that. AI tools helped identify people at risk of getting infected and also speeded up vaccine development to fight the virus.

 

“During Covid, people realised that without tech, they were struggling to provide healthcare services,” says Ankur Dhandharia, health care partner, EY-Parthenon India. That, he adds, helped accelerate the adoption of tech — “both in the backend (administrative work) and frontend (patient care), but tech budgets for hospitals remain small — around 0.5 per cent of the revenue.”

 

Now, some of the early gains (in speeding up services) are prompting a wider use of AI across hospitals — to detect tuberculosis, oral cancer, and fractures — especially in underserved areas where specialists are scarce. 

ALSO READ PART III: Buy-buy AI: Why India's retailers are betting big on AI-powered growth

 

For instance, the Ministry of Hea­lth and Family Welfare is leveraging AI to improve public health services across India. AI tools for diabetic retinopathy identification, an abnormal chest X-ray classifier model, and various other medical conditions are in early stages of development.

 

Under the government’s tuberculosis elimination programme, Cough Against TB, AI is used to screen for pulmonary TB in the community settings. In the deployed areas, AI has shown an additional yield of 12-16 per cent in TB reported cases, which may have been missed if patients were screened using conventional methods. 

 

Researchers are also training AI models to detect patterns in immune system receptors that flag autoimmune diseases like lupus and type-1 diabetes earlier than current diagnostic tools allow.

 

Startups, too, are using AI to detect anomalies. Hyderabad-based Salcit Technologies, for instance, is testing Google’s HeAR model — trained on 300 million cough recordings — to detect TB using just a cough sound. HeAR, or Health Acoustic Representations, has been trained to discern patterns within cough sounds, creating a medical audio analysis. 

 

Enter a medical assistant

 

The early benefits of AI use in hospitals are making it easier to shift more tasks to software to help assist doctors and nurses. 

 

Apollo Hospitals, with its Clinical Intelligence Engine built on Google Cloud’s Vertex AI, analyses medical data from over 14 million patients to support doctors in diagnosis and treatment. Manipal Hospitals has slashed nurse handover time through AI automation, freeing caregivers to spend more time with patients. Its e-pharmacy processing time has dropped from 15 minutes to five. Naren Kachroo, head of go-to-market for generative AI at Google Cloud India, says, “AI will radically transform the way we prevent, diagnose and treat diseases.”

 

Behind the scenes, hospitals are embedding AI into their information systems, revenue cycle management, accounting, supply chains, and inventory planning. Max Healthcare and other hospitals are using GenAI for checking symptom, radiology pre-reads, and automated documentation. “We’ve seen up to 30 per cent efficiency gains in administrative tasks,” says Dhandharia.

 

According to EY’s 2025 report, “The AIdea of India,” 66 per cent of healthcare companies in India are piloting GenAI, while 25 per cent in pharma have moved to full production. Use cases range from medical documentation assistants to AI-driven R&D for drug discovery.

 

“In hospitals, medical documentation is a breakout use case,” says Dhandharia. “GenAI listens to doctor-patient conversations, converts them into clinical summaries, and integrates into the electronic medical record (EMR).” 

 

Stopping short of going too far

 

While geeks want to sell the idea of AI health care, doctors are cautious in going overboard with tech tools. 

 

As Aggarwal says, “We view use of AI tools with cautious optimism. These can provide valuable preliminary insights, but may inadvertently encourage self-medication or delay in seeking professional opinion.” He reaffirms that they should supplement, not replace, medical consultation.

 

Besides, there are regulatory guardrails. Every clinical AI tool must be approved by the Delhi-based Central Drugs Standard Control Organisation (CDSCO) — the country’s equivalent of the US Food and Drug Administration (FDA).

 

There are also concerns about AI bias. 

 

A review by the medical journal Lancet showed that AI skin cancer algorithms lacked training data for dark-skinned individuals. Some stu­dies have found that AI models used gender (having been trained only on male or female data) or race as decision-making shortcuts, leading to misdiagnosis. Such AI models may not be fit for India as they may not have been trained on Indian datasets.

 

AI systems must be trained on diverse datasets and deployed under vigilant human supervision, much like drugs, which are put through long and rigorous trials before they are cleared for use. “Transparency, oversight, and accountability are essential — not optional,” says Aggarwal. These principles help ensure that human judgement remains central, especially in high-stakes domains like healthcare, where patients seek relief, not just algorithmic diagnosis.

 

In radiology, AI tools highlight abnormalities, improving detection and reducing false negatives. In oncology, they’re helping with predictive modelling and drug selection. In pathology, models read histopathology slides to flag signs of disease. In genomics, AI is decoding immune receptor data to predict autoimmune disorders. The most profound impact may lie not in treatment but in prevention — catching diseases before symptoms even emerge.

 

In the near future, you may walk into a hospital where your cough sound is analysed for TB, your blood scanned for immune patterns, and your X-ray flagged by an algorithm — all before you see a doctor. That doctor, in turn, will have a co-pilot whispering data-driven suggestions in real time.

 

If it works, the collaboration betw­een doctors and AI could be the panacea the world has been waiting for.

 

To your health 

Where AI adds most value

  Personalised treatment

 

• Tailors therapies using genetic analysis
• Enables precision medicine aligned with each patient’s profile

  Enhanced surgical precision

 

• AI-assisted robotics plus real-time image guidance
• Boosts accuracy, lowers complication rates, and supports minimally invasive surgery

  Predictive decision support

 

• Forecasts disease progression and treatment outcomes
• Helps clinicians make timely, informed choices

  Workflow optimisation

 

• Automates routine diagnostics and medication workflows
• Frees up clinicians time for high-complexity cases

  Anomaly detection

 

• AI systems surface critical issues that human eyes might miss
• Adds a safety layer across diagnostics and monitoring

  Source: Industry

 

The writer is a New Delhi-based independent journalist