Making AI That Cares for Indian Patients, How AIIMS Is Turning Frugal Technology into High-Precision Diagnostic Power
As Digital Transformations Enter the Consultation Room, India’s Premier Medical Institute Is Forging a Path for Safe, Ethical, and Scalable AI Tailored to the Country’s Unique Needs
Last month, doctors and researchers from the All India Institute of Medical Sciences in Delhi were privileged to participate in conversations around artificial intelligence at the AI Summit in Delhi. It was apparent that we are in a period of profound transition. As expansive digital transformations enter the physician’s consultation room, I find myself introspecting. Algorithms can calculate, but they cannot care. Healing remains a sacred bond built on trust, ethics, and human judgement.
This reflection captures the central tension at the heart of medical AI. The technology promises unprecedented diagnostic precision, operational efficiency, and the ability to reach patients who have never seen a doctor. But it also risks reducing the doctor-patient relationship to a transaction, replacing human intuition with algorithmic calculation, and importing biases from datasets that do not represent Indian patients. The challenge is not to resist technology but to shape it—to ensure that AI serves Indian patients, not the other way around.
AIIMS’s Mandate
AIIMS Delhi has a mandate to anchor the digital transformation of Indian medicine. As the designated Centre of Excellence for AI in Health under the Ministry of Health and Family Welfare, and the medical partner for the national AI-CoE supported by the Ministry of Education, it bears immense responsibility in defining standards for safe, ethical, and scalable AI adoption.
The Centre of Excellence focuses on initiatives such as cancer screening, chronic disease management, visual impairment, maternal and child health, and infectious diseases. These are not random choices. They represent the most pressing health challenges facing India—the diseases that kill the most people, disable the most families, and strain the most resources.
AIIMS has forged a powerhouse ecosystem with the Indian Institute of Technology Delhi, other IITs, and the Indian Institute of Science Bengaluru. Supported by the Indian Council of Medical Research and the Anusandhan National Research Foundation, these partnerships move innovations beyond the lab into public delivery. The goal is not to publish papers or file patents but to ensure AI solves Indian priorities.
The Data Bias Challenge
A significant challenge is the “data bias” of AI tools, which are developed using patient profiles different from those of Indians. Most medical AI models are trained on data from North America, Europe, and East Asia. These populations have different genetics, different disease patterns, different environmental exposures, different lifestyles, and different skin tones. An algorithm trained on European skin may miss melanoma in an Indian patient. A model developed for Western diets may misdiagnose malnutrition in an Indian context.
The solution is not to import AI developed elsewhere but to build AI tailored to India. AIIMS’s strategy is to turn frugal technology into high-precision diagnostic power. While global models for diabetic retinopathy often depend on expensive high-resolution cameras, AIIMS has helped develop MadhuNetRAI, which works with low-cost, handheld cameras practical for district hospitals. This is not a compromise; it is a deliberate design choice that makes the technology accessible where it is needed most.
By prioritizing portability and affordability, AIIMS is ensuring these innovations reach and serve relatively inaccessible areas. The Arogya Aarohan app, a collaboration between AIIMS Delhi and IISc Bangalore, enables frontline healthcare workers to assess oral cancer risks using only smartphone photographs of a patient’s oral cavity. Similarly, AI-powered cough analysis for tuberculosis brings high-level “doorstep diagnostics” to resource-constrained rural clinics, enabling early, life-saving interventions where they were previously impossible.
Indigenous Datasets for Indian Conditions
AIIMS is applying the same logic across multiple domains. In dermatology, it is building indigenous datasets for Indian skin tones. Skin diseases present differently on darker skin, and algorithms trained on light-skinned populations can miss critical signs. By creating datasets that reflect India’s diversity, AIIMS is ensuring that AI-based dermatology tools work for all Indians.
In breast cancer screening, AIIMS is adapting AI to the susceptibilities of Indian women. Breast cancer in India tends to present at younger ages and with different biological characteristics than in Western populations. Screening algorithms designed for Western populations may miss cancers that are common in India. By training AI on Indian data, AIIMS is building tools that catch cancers when they are still treatable.
These efforts are not about creating separate, inferior technologies. They are about building technologies that are superior for the Indian context. A tool that works with a smartphone camera is not a lesser tool; it is a tool that can be deployed where a high-resolution camera is not available. A dataset that includes Indian skin tones is not a limited dataset; it is a dataset that reflects the population it serves.
The Digital Co-Pilot
At the radiology department, an approved AI system acts as a “digital co-pilot,” flagging abnormalities in chest x-rays. In a high-volume public hospital, this helps specialists prioritize critical cases and reduce diagnostic delays. A radiologist in a busy hospital may see hundreds of x-rays a day. Fatigue is inevitable. A missed finding can be a matter of life and death.
The AI does not replace the radiologist; it augments them. It scans every image, flags potential abnormalities, and prioritizes the most urgent cases. The radiologist focuses their attention where it is most needed. The combination of human expertise and machine precision is greater than either alone.
This model—human-in-the-loop, AI as augmentation rather than replacement—is the guiding philosophy at AIIMS. The goal is not to automate doctors out of existence but to give them tools that make them more effective.
Training the Next Generation
As the country’s population structure shifts and chronic disease burdens grow, India must strengthen its demographic resilience—the capacity to anticipate and manage profound population changes. Human resource expansion alone cannot bridge these widening gaps. India does not have enough doctors, nurses, or specialists to meet the needs of its growing and aging population. The gap will only widen. India must amplify its capacity through technological augmentation.
Training is the cornerstone of this adaptation. AIIMS is integrating AI literacy into the medical curriculum, ensuring clinicians master algorithmic logic, bias recognition, and clinical accountability. The doctors of tomorrow must understand not only how to use AI tools but also how to evaluate them, how to recognize their limitations, and how to maintain clinical accountability when algorithms are involved.
This is not a small shift. Medical education has traditionally focused on knowledge acquisition—learning facts, recognizing patterns, applying protocols. AI literacy requires a different set of skills: critical evaluation of algorithmic outputs, understanding of statistical biases, and the ability to integrate machine recommendations with human judgment. AIIMS is leading the effort to define what this curriculum should look like.
Building the National Ecosystem
AIIMS supports the national ecosystem by releasing anonymized datasets. By sharing structured data, it can incubate an industry of innovators. Medical AI requires large, high-quality datasets for training and validation. These datasets are expensive to create and difficult to share due to privacy concerns.
AIIMS is solving this problem by creating anonymized, structured datasets that can be used by researchers and developers across the country. A startup in Bengaluru can access the same data as a research group in Delhi. A researcher in Chennai can build algorithms that are validated on data from across India. By creating a shared resource, AIIMS is lowering the barrier to entry for medical AI innovation.
Global Partnerships
India’s influence in the health-tech domain transcends borders. The Indo-French Centre for AI in Health, a partnership between AIIMS, Sorbonne University, the Paris Brain Institute, and IIT Delhi, aims to build trusted AI and foster global interdisciplinary research. This is not a one-way transfer of technology from the Global North to the Global South. It is a partnership of equals, with India contributing its expertise in frugal innovation, population-scale deployment, and clinical diversity.
Prime Minister Narendra Modi’s vision for India@2047 is one where India leads through self-reliance and indigenous innovation. By building medical AI tailored to India’s unique needs and proving its efficacy in the world’s most demanding public health conditions, India is doing more than solving domestic challenges. It is crafting a roadmap for the future that other countries, facing similar challenges, can follow.
Conclusion: Algorithms That Care
The director’s reflection captures the essence of AIIMS’s approach: “Algorithms can calculate, but they cannot care.” The challenge is not to choose between human care and machine precision. It is to build machines that support human care, to use technology to amplify the capacity of caregivers, to ensure that the doctor-patient relationship remains the centre of medicine even as the tools around it change.
India has the opportunity to define what medical AI looks like in the 21st century. Not by importing models developed elsewhere but by building models tailored to Indian conditions. Not by replacing doctors with algorithms but by giving doctors tools that make them more effective. Not by sacrificing care for efficiency but by using efficiency to deliver more care.
The vision is ambitious, but the foundations are being laid. Indigenous datasets, frugal technologies, integrated curricula, shared resources, global partnerships—each is a piece of the puzzle. When assembled, they will form a medical AI ecosystem that is not just advanced but appropriate, not just efficient but equitable, not just global but Indian.
Q&A: Unpacking AIIMS’s Medical AI Strategy
Q1: What is AIIMS’s role in India’s medical AI ecosystem?
A: AIIMS Delhi is the designated Centre of Excellence for AI in Health under the Ministry of Health and Family Welfare and the medical partner for the national AI-CoE supported by the Ministry of Education. It has forged partnerships with IIT Delhi, other IITs, IISc Bengaluru, ICMR, and the Anusandhan National Research Foundation to move innovations beyond the lab into public delivery. Its goal is to ensure AI solves Indian priorities in areas like cancer screening, chronic disease management, visual impairment, maternal and child health, and infectious diseases.
Q2: What is the “data bias” challenge in medical AI, and how is AIIMS addressing it?
A: Most medical AI models are trained on patient data from North America, Europe, and East Asia, which have different genetics, disease patterns, and skin tones from Indian populations. An algorithm trained on European skin may miss conditions in Indian patients. AIIMS is addressing this by building indigenous datasets for Indian conditions—for dermatology (Indian skin tones), breast cancer (susceptibilities of Indian women), and other areas. It is also developing frugal technologies like MadhuNetRAI for diabetic retinopathy that work with low-cost, handheld cameras practical for district hospitals.
Q3: What are some examples of AI tools developed through AIIMS’s partnerships?
A: MadhuNetRAI enables diabetic retinopathy screening using low-cost, handheld cameras. The Arogya Aarohan app, a collaboration with IISc Bangalore, enables frontline workers to assess oral cancer risks using smartphone photographs. AI-powered cough analysis for tuberculosis brings “doorstep diagnostics” to rural clinics. An approved AI system in radiology acts as a “digital co-pilot,” flagging abnormalities in chest x-rays to help specialists prioritize critical cases.
Q4: How is AIIMS preparing the next generation of doctors for AI integration?
A: AIIMS is integrating AI literacy into the medical curriculum, ensuring clinicians master algorithmic logic, bias recognition, and clinical accountability. Medical education is shifting from traditional knowledge acquisition to critical evaluation of algorithmic outputs, understanding statistical biases, and integrating machine recommendations with human judgment. AIIMS is also releasing anonymized datasets to support the national ecosystem of innovators.
Q5: What is the Indo-French Centre for AI in Health, and why does it matter?
A: The Indo-French Centre for AI in Health is a partnership between AIIMS, Sorbonne University, the Paris Brain Institute, and IIT Delhi aimed at building trusted AI and fostering global interdisciplinary research. It represents a model of equal partnership, with India contributing its expertise in frugal innovation, population-scale deployment, and clinical diversity. This positions India as a leader in medical AI, not just a consumer of technologies developed elsewhere.
