Evolution of AI in Medicine

AI is already changing the questions patients bring into a consulting room — people arrive having used ChatGPT, Claude or Perplexity to summarise a previous visit or research a diagnosis, so the consultation starts further along.

It is not a substitute for training. AI can hallucinate, and the medical textbooks a doctor's judgement is built on still have to be read.

AI has assisted mammogram detection since 2014, well before the current wave of attention — mainly to flag areas worth a radiologist's second look.

Next-generation sequencing is opening the door to more personalised treatment, but it stays expensive and needs more standardised data before it scales.

Breast cancer is now the most common cancer among Indian women, and early lumps are frequently painless — which is exactly why they get neglected. Risk factors: genetics, obesity, smoking and delayed childbearing past 30, with pollution and stress the hardest to quantify.

On lifestyle he is blunt: there is no safe limit for alcohol or smoking, fatty liver is becoming common, and Indian women may lack certain enzymes that process alcohol as efficiently as is often assumed.

Advice for medical students: use AI for personalised study plans and virtual logbooks that summarise clinical cases — and add a formal AI course to the MBBS curriculum rather than leaving it to be picked up informally.

Frontline safety is an administrative failure, not a clinical one — the structural problem is a shortage of resources and hospital beds. The India–West technology gap has narrowed to roughly two to three years; affordability, not access, is what remains.

The open opportunity: no platform in Indian healthcare closes the loop from booking an appointment through surgery into follow-up care. He backs the thesis with his own money as an angel investor in medical AI startups including Curium and HexaHealth.