Meena lives in a village 80 kilometres from the nearest eye specialist. She has been diabetic for six years but has never had a retinal scan — the nearest ophthalmologist is in a city she cannot easily travel to.
Last year, a community health worker visited with a basic smartphone and an AI diagnostic tool. Meena looked into the phone camera for thirty seconds. The AI analysed the retinal image and flagged early-stage diabetic retinopathy — damage to the blood vessels at the back of her eye that, if untreated, leads to blindness.
Her local doctor might have missed it. The AI did not.
The Healthcare Gap That Still Exists
On paper, India's overall doctor-to-population ratio has improved significantly. As of July 2024, the Ministry of Health confirmed a ratio of approximately 1:836 — better than the WHO standard of 1:1,000. (Source: Ministry of State for Health, written reply to Lok Sabha, July 2024 — reported by Deccan Herald)
But the aggregate national ratio conceals a severe distribution problem. Doctors are concentrated in cities. Rural and semi-urban areas — where the majority of India's population lives — have dramatically fewer practitioners per person. For specialists like cardiologists and ophthalmologists, and for diagnostic equipment, the gap between urban hospitals and rural health centres is enormous.
This gap has existed for decades. What has changed in the last two years is that AI is beginning to close it — not by teleporting doctors, but by bringing diagnostic capability directly to the patient.
AI as Diagnostic Partner
AI diagnostic tools in ophthalmology, radiology, and dermatology are achieving accuracy that matches specialist-level performance on specific, well-defined tasks. In a 2018 study published in Nature Medicine, a DeepMind-developed AI assessed over 50 retinal conditions from OCT scans with 94% accuracy, matching or exceeding the referral recommendations of world-leading retinal specialists. The collaboration was between DeepMind, Moorfields Eye Hospital, and UCL Institute of Ophthalmology. (Source: De Fauw et al., Nature Medicine, August 2018)
The critical phrase is specific, well-defined tasks. These AI systems are extraordinarily good at the narrow diagnostic task they were trained on. They are not general doctors. They cannot replace the clinical judgment, patient relationship, and contextual reasoning that an experienced physician brings.
But for high-volume screening of specific conditions in populations that currently receive no screening at all, they are genuinely transformative. The typical workflow: a community health worker captures the relevant data with a basic device. The AI processes it and flags anything requiring specialist attention. The specialist reviews only the flagged cases, remotely. Patients who would never reach a specialist get specialist-level screening.
AlphaFold and the Drug Discovery Revolution
Every protein in your body is a string of amino acids that folds into a specific three-dimensional shape — and that shape determines what the protein does. Predicting what shape any given sequence would fold into was, for half a century, beyond human ability.
In 2020, DeepMind's AlphaFold solved this problem. It entered the CASP14 competition — the international protein-folding prediction challenge — and produced results that, in the words of the scientific community, made every other submission look like it was from a different era. DeepMind then released the results publicly: a database of 200 million protein structure predictions, covering essentially every known organism on earth. Free. Accessible to any researcher in the world. (Source: DeepMind AlphaFold database announcement, July 2022)
For drug discovery, this is fundamental. Developing a drug is a problem of finding a molecule that fits a specific protein — like a key for a lock. For fifty years, drug developers were trying to make keys without clearly seeing the locks. AlphaFold changed that.
"AI will not replace doctors. But it will reach the hundreds of millions of patients that doctors cannot."
Mental Health: The Quiet Crisis AI Is Helping Address
Mental health is one of the most underserved areas in Indian healthcare. India has approximately 0.75 psychiatrists per 100,000 people — far below the WHO's recommended minimum of 3 per 100,000. A 2023 Parliamentary Standing Committee report found India had only around 9,000 practising psychiatrists for a population of 1.4 billion. (Source: Parliamentary Standing Committee on Health and Family Welfare, 148th Report, 2023 — reported by The Print, August 2023; Business Standard, October 2025)
AI mental health tools — from text-based therapy companions to symptom-tracking apps — are not replacements for human therapy. They are clear about that. What they provide is a first point of contact: a non-judgmental, always-available space where someone can articulate what they are feeling, receive evidence-based coping techniques, and get help identifying when they need professional intervention.
The nuance matters here. AI cannot diagnose complex mental health conditions, cannot prescribe, and cannot provide the deep relational healing that good therapy offers. But for someone who would otherwise receive nothing, even a good first step is genuinely valuable.
Hospital Workflow Automation
Inside hospitals, AI is reducing the administrative burden that consumes a disproportionate share of clinical time. In the US, a large-scale study by Sinsky et al. published in the Journal of General Internal Medicine (2023) found that physicians across all specialties spend an average of 57.8% of their clinical time in the electronic health record — on documentation, chart review, orders, and inbox management. (Source: Sinsky et al., Journal of General Internal Medicine, 2023 — reported by the AMA)
Note: this is US data and Indian clinical environments vary. But the pattern of administrative burden consuming clinical time is documented across healthcare systems globally.
AI transcription and documentation tools can capture a clinical conversation and generate a structured note automatically, returning hours of time to clinical care. AI scheduling systems are reducing wait times by predicting no-show rates. AI triage tools in emergency departments are helping prioritise patients more accurately than legacy scoring systems.
Can We Trust AI With Life Decisions?
The honest answer: it depends on how AI is used. AI making autonomous life-or-death decisions without human oversight is not something the field is ready for, and responsible developers are not trying to do it. AI assisting trained clinicians by providing better information, flagging things that might be missed, and reducing cognitive load — that is both ready and beneficial.
The evidence from deployment suggests that AI-assisted clinicians outperform both AI alone and clinicians alone on specific diagnostic accuracy tasks. The combination is what works. The AI catches what the human eye might miss on volume. The human provides judgment, context, and the clinical relationship that makes medicine more than pattern recognition.
What This Means For India
India is both one of the largest potential beneficiaries of AI in healthcare and one of the most important test cases for whether it can work at scale. The AYUSHMAN Bharat digital health mission is building the infrastructure that AI health tools need — digital health IDs, standardised records, interoperable systems.
In tier-3 towns and villages, AI diagnostic tools running on smartphones and low-cost devices could provide a level of healthcare access that has never existed before. The technology is ready. The challenge now is deployment — training community health workers, integrating tools into existing workflows, building trust with patients, and ensuring that the AI serving these populations was trained on data that reflects their demographics.
The gap is real. The tool to close it is here. The work now is making sure it gets to the people who need it most.
