The conventional drug discovery process typically spans 10 to 15 years, requiring significant financial investment to progress from initial research to an approved treatment. However, the last decade has seen a transformative impact of AI on drug discovery, drastically reducing both the time and cost of bringing new drugs to market. AI technologies, particularly machine learning models, enable researchers to analyse vast datasets quickly, identifying potential drug candidates and predicting their effectiveness and safety with remarkable speed. These technologies streamline key stages of drug development, including drug target identification, virtual screening, and toxicity prediction.
Yet, convincing stakeholders of the validity and uniqueness of AI-driven drug discovery has been a major challenge for entrepreneurs in this field, said Dr Riyaz Syed, founder and CEO of Centella AI Therapeutics, during his special address at THE WEEK Best Hospitals event held in Hyderabad.
“Even when we presented the right structure and demonstrated its value, there was scepticism. It took time to establish ourselves and prove that the data generated and validated by our system was reliable,” he shared.
Dr Syed, a seasoned “drug hunter” with 15 years of expertise in medicinal chemistry and a strong portfolio of publications and patents, highlighted the limitations of health data in India. While the country generates immense amounts of health data, the volume available for AI-based drug research remains very low. “AI models require vast datasets to identify patterns, but obtaining such data was extremely challenging in the early stages,” he explained.
Dr Syed also noted the scarcity of talent at the intersection of AI and drug discovery. “In the healthcare ecosystem, while there are experts in AI and others in drug discovery, finding individuals with expertise in both was nearly impossible,” he said. “Many diagnostic companies I’ve interacted with have data and an interest in AI applications, but they lack clarity on how to proceed. Questions of what to do, how to do it, and, most importantly, why to do it—intent and purpose—are critical. Bridging this gap required considerable effort in grooming and training teams with dual-domain expertise.”
Another hurdle lies in the computational demands of AI-driven drug discovery. High-performance systems are essential for simulations and training large-scale models, such as protein-language or chemical-language models, subsets of large-language models. “Explaining these concepts to venture capitalists has been a significant challenge. It often takes time for them to fully grasp what we’re doing and why it matters,” he added.
In preclinical settings, these challenges have slowed AI adoption. Drawing parallels to clinical settings, Dr Syed observed that patient care faces similar issues. “There is an abundance of diagnostic and disease-related data, yet it often remains underutilised,” he said.
Despite these obstacles, Dr Syed is optimistic about India’s potential in AI-driven drug discovery. “The culture of data-driven research is gradually gaining traction in India’s healthcare ecosystem. Organisations are starting to recognise the value of investing in research and extracting actionable insights from data. Transitioning from basic research to real-time applications will heavily depend on the intent and vision of management,” he said.