Artificial Intelligence in Indian Healthcare
Researchers at IIT-Madras and the Translational Health Science and Technology Institute are developing an AI model named Garbhini-GA2 to predict the age of a foetus using ultrasonography. This model, trained on data from 3,500 pregnant women, showed significantly higher accuracy compared to traditional methods like Hadlock’s formula, which often miscalculated due to population differences.
Application in High-Risk Pregnancies
- Almost 50% of Indian pregnancies are high-risk, often leading to complications such as severe anaemia and high blood pressure.
- NGO ARMMAN, in collaboration with UNICEF, trains ANMs through digital platforms to manage high-risk pregnancies. AI chatbots have been introduced to support this initiative.
- The chatbot, used by 100 ANMs, has received 94% positive feedback for its accuracy but faces challenges with regional language variations.
Virtual Autopsies
- Virtual autopsies (virtopsies) use CT and MRI scans to create 3D body images for postmortem analysis, offering a non-invasive alternative to traditional autopsies.
- These can be completed in 30 minutes but may miss small soft tissue injuries, which a combination with verbal autopsies can help mitigate.
Challenges with AI Implementation
AI tools in healthcare face data privacy concerns and biases:
- Data Privacy: Governed by the Information Technology Act 2000 and the Digital Personal Data Protection Act 2023, which do not specifically address AI.
- Automation Bias: The tendency to overly trust AI, potentially impacting clinical judgment, as evidenced by a study involving radiologists.
Conclusion
While AI has the potential to revolutionize healthcare by making it more efficient and accessible, it inherits human biases and requires robust governance and clinician training to ensure ethical practices. The development and implementation of AI in healthcare need continuous assessment and adaptation to ensure its reliability and effectiveness.
