Large scale genomics and AI in Indian healthcareOctober 8, 2019
The Indian healthcare system does not traditionally have much of a computational aspect apart from regular data entry. This, however, is changing with the landscape of healthcare being transformed with the emergence of artificial intelligence (AI). With several state governments and other stakeholders, such as FICCI, playing a major part in promoting AI, it is time for the clinical fraternity to be more proactive and become leading innovators in using AI. One major announcement recently from the Department of Biotechnology was about plans for Indian genome sequencing to be carried out in the next five years. This will generate large amounts of data and can give a better genetic understanding about the Indian scenario. If scientific and clinical communities can collaborate in using these datasets to develop machine learning models and develop AI platforms, it can enable disease risk prediction and better molecular diagnostics.
Even though such projects may give large amounts of data, the large skill gap in the Indian market makes AI more of a necessity as it can enable decision making faster. Doctors need to intervene only when confidence in decision making is low. Such an approach can improve the overall quality of healthcare, as a doctor’s time is better spent on important cases. India will also become a good place to test many AI-based technologies, especially in rural areas, in order to reduce the gap in healthcare accessibility.There is a need for all stakeholders to be actively involved in the development, adoption and implementation of AI for overall success. The five stakeholders are practitioners, developers, research and industry bodies, government, and funders. Of these, the practitioners are very important as
they play a role in the entire pipeline, from product conception to implementation. There are 14 or more hospitals which have incorporated technology in their systems primarily in the diagnosis and patient monitoring.
One major aspect of the introduction is that clinicians and scientists can use and evolve best practices in clinical interpretation of genetic data, which is a complex process involving genetic variants, therapies, diagnostics and phenotypes. Genome-wide association studies with Indian genomic data will give a better map of the regions of DNA associated with disease risk and diagnosis. However, this will be the first time in history that our ability to collect data about our own biology will outpace our ability to interpret it. Therefore, the implementation of proper machine learning approaches and AI will bring the best out of such studies. Apart from diagnosis, whole-genome sequencing may also give us better drug targets and a keener understanding of drugs. Traditionally, drugs are developed for one target and have an effect similar to the lock and key model. However, there is a growing body of research showing off-target interactions and unanticipated side effects. The implementation of AI in clinical outcomes can truly bring much-needed changes for clinical practices.
The author is medical scientist and former director of SGRF, Bangalore