“AI is like salt in food. […] that being there or not there, can make a huge difference.”

– Arnav Varghese, 2^nd year MBBS, CMC Vellore

A debate was held on the first day of the conference which I was very excited to be a part of. We discussed the question of liability especially in the case of failed healthcare operations, or to what extent AI should be allowed to intervene or completely take over operations in healthcare. The scenario was such that an emergency case, where a musician met with an accident, suffered an epidural haematoma and had to be rushed to hospital, was required to be operated on by Neurobot, a neurosurgical AI system, which was to assume full control over the operation. Surgery was carried out without consent. The patient had survived the operation, however with paralysis of the right arm. So the debate was whether the AI company should have been held liable, or whether even without consent, saving the life of a patient was enough reason to address the need of the hour and proceed with surgery.

My personal stance is that AI is like salt in food. You add salt to make your food taste better, but it is not essential. At the same time, that being there or not there can make a huge difference. AI should be used, but it should be used diligently, carefully, with supervision. It’s still developing and it does have a lot of scope, but it requires a lot more datasets considering that medical science and health is not very objective 9/10 times – there will be an exception to a rule and that exception can mean life or death for someone. So improving the datasets, at the same time, maintaining data privacy, is crucial to building an AI tool especially in relation to medical sciences.

A take-away from this conference is that it is so important to be up-to date with the kind of advancements that are happening, because the world is changing, we’re evolving, and change is the only constant. I think conferences like these give people a great platform to come and actually look at what’s happening in the outside world, learn what we can and make the best out of it. Do I see myself being involved with AI in my future? Definitely. More than a choice, it’ll become an obligation in the years to come with how the health systems are evolving. I have the privilege of working on an AI-related study with Dr Amith Balachandran from Medicine, where we are trying to understand how good AI photography apps are in differentiating poisonous and look-alike non-poisonous plants.

AI and health data really interests me: seeing how technology is incorporated, and try and improve the way AI works. It has to be a global effort; it has to be all health systems trying to improve the current datasets available to try and make AI a better tool for healthcare; but I’m sure we’ll get there eventually and it’ll be an integral part of all healthcare systems.

“[…] understanding how doctors think, how clinicians will expect engineers to collaborate […] – these are all important to discuss…”

– Asjad Nabeel, PhD scholar, NIT Surathkal

Our research team headed by Dr Jeny Rajan, NITK, has collaborated with Dr Hannah Thomas, one of the organisers of this conference. I have a background in engineering, so I’m familiar with the tools and technologies related to AI, however this conference was related to AI in health data; I have only been working in the medical field for the last two years.

It was an enriching learning curve for me – understanding how doctors think, how clinicians will expect engineers to collaborate, and how communities can benefit – these are all important to discuss and helpful for future work. Overall, I enjoyed my experience at this symposium. CMC Vellore has a great campus, and I got to meet lots of people from different parts of India and abroad.

“AI is like an infant, or a new-born.”

– Sivaranjani R., Madras University

The future of AI will be based on how we mould its environment and decide on the inputs to be given. Technology is a double-edged sword, only time can say whether it will be a good tool or bad tool. The conference was very well-organised, with eminent speakers.  The knowledge we have gained over three days is invaluable. Looking forward to more.

“The most crucial and interesting discussion was the need to establish regulatory frameworks in the use of AI.”

– Serena, intern-volunteer, Office of Communications (Directorate), CMC Vellore

AI is an integral part of communication today, and increasingly relevant to content writing. Exposure to the cutting edge of AI discussions as an intern-volunteer at the symposium was an insightful experience. A deeper understanding of AI, new developments, its massive potential as well as gaps and potential risks, is important to know in a world where AI is becoming an inevitable tool.

The most crucial and interesting discussion was the need to establish regulatory frameworks in the use of AI. To what extent AI should be allowed to intervene in work that can be done by humans – like creating content for media – was food for thought. Working under publicity and communications in healthcare, familiarising myself with AI as a tool for content as well as its uses and developments in healthcare is essential.

“It interests me how AI and digital communication can be used to provide patients with health advice, schedule reminders, recommend next steps, etc.”

– Allen Theodore E, intern-volunteer, Office of Communications (Directorate), CMC Vellore

I particularly enjoyed the presentations on healthcare-related AI apps; it interests me how AI and digital communication can be used to provide patients with health advice, schedule reminders, recommend next steps, etc. The potential of AI to assist with health diagnostics also enhances the pace and accuracy of patient visits, resulting in faster and more individualized care.

As I work primarily with healthcare content using multimedia tools that also incorporate AI, it was useful attending the symposium. I’ve gained a better understanding of AI and knowledge of AI developments and issues in health data that can also be applicable in non-medical fields.

Will AI make clinicians obsolete?

Dr John Cherian Oommen, CMC alumnus and Vice-Chairman of the CMC Vellore Association

Are good clinicians disappearing anyway, even without AI?

Speaking from a perspective of having lived and worked for over 30 years in a rural hospital in Odisha. In a world focused on data points and dollar points, AI can be a tool for speed, efficiency & accuracy, he said, but only clinicians can act as agents of compassion & share in the pain of their patients. Dr. Oommen reiterated that as long as patients are human, AI cannot replace the clinician nor can AI give the clinician meaning and purpose in his calling. AI is a great tool, but a poor master. Patient care is a science and an art, of head, heart and hands. Implications of a fully computerised hospital? Digital inequalities will power digital colonialism. AI and impact on jobs in healthcare. A take from Christian healthcare: healing is far more than cure, healing is restoration to the purpose for which God created us; wherever there is pain, God is present, sharing the pain; when we touch people in pain, we may touch the divine; our calling is to share the pain of people (com-passion); our work is our worship offering.

Importance of Health Data Standards – ICD/SNOMED/DICOM/IOT/Interoperability

Dr Karthik Adapa, Regional Adviser, Digital Health at WHO-SEARO

Digital Health: all-encompassing systematic applications of ICT. Health journey in three contexts: non-digital, digital but fragmented, digital and interoperable. Pilotitis – digital but fragmented. Necessity for collaborative effort among healthcare professionals. Interoperability between systems and software. How to prevent pilotitis? Is data exchangeable with already existing systems? DPG Alliance – major registry of all the open source tools in the world – all major problems under SDG3 (health) have tools existing to an extent already – take existing solutions, built to standards – and customise to specific needs. If making new, register it to DPG Alliance: 1. Buy standards-based and interoperable software 2. See if software exists already. Use digital building blocks. Issue with standards is not availability but implementation and outdation. Applications require maintenance. Is AI moving towards how the human brain works, or is it not? There is some mimic, but language is just the surface of what humans do. This is an exciting time, as AI moves to deeper levels.

Medical Digital Twins

Dr Nitin Williams, Dept of Neuroscience and Biomedical Engineering, Aalto University, Finland

Vision of personalised medicine. Virtually identifying ablasions and scarring of tissues after a heart attack. Medical digital twins simulating what-if scenarios: surgical planning (eg. heart surgery), selecting clinical interventions (cancer treatment), chronic disease management (diabetes). Workflow using AI to improve brain digital twins. Digital twin has successfully predicted surgical outcomes, good enough for FDA approval – Dassault and Siemens, two companies that sell heart digital twins approved by FDA. Brain digital twins can identify Epileptic Zone in drug-resistant epilepsy, but have not yet been shown to improve surgical planning – better digital twins predicted in the long-run. Good physiological systems like the heart in place; need to then work on interaction between systems, eg. between heart, lungs and brain – closer to the idea of a digital patient, allowing increasingly accurate patient-specific predictions because everything modeled and can in turn, better inform therapy. Using continuous real-time data to update the digital twin of a patient. Leading to the vision of preventive medicine: making predictions about what is going to happen, what may go wrong, what interventions should be done at what time and at what dose to prevent something going wrong. EDITH project started in Europe.

AI and Mental Health Apps

Ms Jagrika Bajaj, Clinical Development Officer at WYSA

WYSA, a clinically validated app built in 2016 which uses AI to support emotional health and wellbeing. Ethical use of AI in mental health user experience. Mission to ensure that mental health support is accessible. Use of natural language processing to detect emotions and flag at-risk words like SOS, self-harm or abuse. Routed through safe empathetic flows written by clinicians. No self-learning; feedback from real user incorporated to make scripts better. Rule-based scripts pros and cons: guiding users through structured CBT exercises but users remark that it feels repetitive, lacks context and sometimes misses the mark emotionally. GenAI integrated into WYSA to meet user needs and demonstrate risks; GenAI follows predetermined protocols approved by clinicians. Unfairness and biases mitigated through comprehensive testing. Information security, privacy and clinical safety ensured. Cross-functional teams help bring together technical, clinical and humanistic angles so users are treated such that they feel empathised with. American Psychological Association recently developed Companion Checklist to assist practitioners in evaluating AI-enabled clinical or administrative tasks including data privacy.

Implementing AI in healthcare – Regulations and the lack of clear guidelines

Mr Rahul Mathan, Trilegal, Bengaluru

AI and clinical decision support; drug discovery and GenAI risks. Regulation from two perspectives: risks to be managed, and opportunities to be unlocked. A good regulatory framework will balance the two. Trade-off is implicit in balance.

Bias: AI built on western datasets, implications for the Indian context?

How do we unlock Indian data for the good of everyone while still ensuring data protection? Tremendous opportunities for AI, but mistakes/inaccuracy are an inevitable part of its probability-based nature. Ability to fail responsibly needs to be worked upon. Consumer protection approach to liability. We have to put our heads together to think about how to protect patients and still use what AI has to offer. Intercessory medicine in Law. How do we unlock data for the good of everyone, while still ensuring data protection? Do we need new laws for AI? Already the case in the UK, Europe, California. What about India? How we can think about this: AI Ethics Committee, Red Teaming, AI Safety Institutes, Regulations to promote AI, Regulations on Drug Discovery, Copyright law, Data Protection law, Non-personal data.

AI – the New Alchemist?

Dr Vikram Gota, Clinical Pharmacologist, ACTREC-TMC

Almost every step in drug discovery is amenable to AI intervention. Can possibly reduce development time by half – average time with conventional approach is 10-15 years, which with AI tools could be reduced to 7-9 years.

Surgery 2.0. – Enhance human skill with AI and Robotics

Mr Jayakrishnan, Founder and CEO of Asimov Robotics, Kochi

Levels of surgical robots, each level increasing in automation and feature

Level 3: Robots with AR and haptic feedback

Level 4: Automatic suturing

Level 5: Fully independent

Digital Pathology and AI in laboratory diagnostics

Dr Swapnil Rane, Associate Professor, ACTREC-TMC (Advanced Centre for Cancer Treatment, Research and Education in Cancer, Tata Memorial Centre)

Applications of AI in pathology – ‘virtual staining’, 3D Histology, whole-block imaging and colour normalisation. Digital pathology suffers due to high expenses and high storage space required.