The World Economic Forum recently hailed AI in healthcare as India’s trillion dollar opportunity. This article looks to unpack this prediction further.
Taking a step back: The need for artificial intelligence (AI) arose when the volume of available data surpassed human ability to comprehend it using human cognition alone in required time. There was an increasing need to automate repetitive tasks or supplement subjective guesswork with data-backed decision making. Coincidentally, healthcare has been burdened with many repetitive admin tasks such as manual counting in diagnostics and is laden with subjectivity in care protocols. If an AI were to predict which sector needs AI the most, healthcare would be one of the top picks!
Good AI needs great datasets to learn from. With India’s 1.3 billion population and the increasing digital penetration (600M+ smartphone users), quality datasets have increasingly been available for research in India and to Indian developers. Further digitization of the healthcare systems with the advent of Ayushman Bharat Digital Mission (ABDM) – which has already created more than 200M+ Health IDs – will hopefully bring more real-world data for research and development. In addition to this, India’s technology and engineering talent is making the best of this resource to create AI solutions in India, use India as a sandbox and then export this to the world.
(1) Diagnostics: Efficient, affordable and accurate radiology screenings
Diagnostics in India is riddled with challenges – limited supply side talent, lack of standardization and quality protocols, etc. leading to severe misdiagnosis which is a leading cause of avoidable mortality and economic burden. We have seen AI bringing in efficiency in existing diagnostics processes by increasing throughput of radiologists and improving the quality of diagnosis by reducing human error.
For instance, Niramai is working towards creating predictive models for diagnosing life-threatening conditions like cancer using newer biomarkers. It uses AI to enable early detection of breast cancer so that patients can get timely medical intervention. Carpl, on the other hand, is helping doctors identify the right AI model by creating a “playstore” that hosts, compares and helps deploy multiple AI radiology solutions from the world.
(2) Mental Health: Triaging patients through AI care
While perception is slowly changing, mental health is still a taboo in India. This is compounded by the lack of access to quality mental health professionals.
Emotionally intelligent AI chatbots are enabling on-demand therapy and clinically backed interventions. When patients are feeling depressed, anxious, stressed or just overwhelmed, they can reach out immediately and feel heard. AI and ML tools are also using this data to identify patterns, bring in psychiatrists and psychologists in dire cases and improve mental health outcomes. One such company is Wysa. Through its AI powered chatbots blended with support from professional experts, it is helping users manage various mental health conditions using clinically validated interventions.
(3) Fertility: Egg, sperm and embryo grading for higher success in IVF
There are nearly 40M infertile couples in India. The process of seeking fertility treatment, especially for women is a painful one. There are multiple decision points in the process (which procedure to choose, what dosage of medicines for preparing the eggs, when and how many to retrieve, which embryos to be selected, when to do the embryo transfer etc.). IVF particularly involves the process of identifying the best quality sperm first and embryo later that should be used for a successful pregnancy. This process is manual today – thus prone to human error and limited scalability which has plateaued its global success rate to 30-40 per cent.
Multiple companies are using computer vision (to ensure machine-based screening of all available eggs/sperms/embryos) and machine learning (to predict which sperm/embryo would have higher chances of success) models to bring scientific rigor to the process and make it error-free and scalable. Indian experts are working with global innovators to create in-house solutions for developing and testing such machine learning models.
While we are seeing many disruptive uses of AI in healthcare already, we have only scratched the surface. As data quality and technology infrastructure strengthens AI solutions, we will see:
(i) unprecedented automation in multiple tasks for patients and providers which eventually simplifies healthcare (like AiDoc in radiology)
(ii) providers acting upon critical insights based on the patient’s unique genetic, psychological and health parameters which makes healthcare truly personalized (23&me being a great example)
As is rightly pointed out by the World Economic Forum, the next phase of maturity in this industry will come from responsibly foreseeing concerns like bias, ethics and privacy in the datasets we use to build out these AI models.
Trust is a key pillar in healthcare and we strongly believe that the adoption of AI in healthcare will be closely linked to our ability to create privacy respecting, secure, unbiased and regulated AI solutions. Wysa, for example, was recognized as one of the only two mental health solutions that respected data privacy by an independent analysis by the Mozilla foundation.
Authors: Salim Afshar, Namit Chugh, Tushar Sadhu