After seeing Parkinson's take his father, this Googler found a new mission: teaching AI to chase cures

Vivek Natarajan, a Google DeepMind researcher, is on a transformative journey to harness the power of artificial intelligence in the medical field. His motivation springs from a deeply personal experience: witnessing the impact of Parkinson's disease on his father, who dedicated 35 years to a prominent Indian newspaper before the illness altered his life significantly. As Natarajan pursued his master's degree in the U.S., his father exhibited alarming physical and cognitive changes, prompting Natarajan's mother to take on the caregiver role. Despite his struggles, his father remained committed to his work until retirement, ultimately passing away. This experience ignited a passion in Natarajan to explore how AI could revolutionize healthcare, a question that has guided his career ever since. In 2017, while at Meta, Natarajan was immersed in the world of deep learning, where he recognized the potential of AI beyond enhancing online engagement. His interest in healthcare was sparked by his upbringing in India, where access to medical care significantly influences lives. Natarajan recalls his early attempt to create a basic app, "Ask the Doctor Anytime, Anywhere," which reflected his ambition to improve healthcare access. After joining Google in 2019, he became part of a team focused on integrating AI with medicine. Guided by Greg Corrado, a Google Brain co-founder, Natarajan was encouraged to dive into the medical domain despite his lack of formal training. Within Google, he faced the complexities of the healthcare field, where building trust with physicians and patients is paramount, and scientific integrity must go hand in hand with innovation. Natarajan teamed up with Alan Karthikesalingam, a physician-scientist, to tackle deeper issues related to AI in medicine. Their work was inspired by Google’s groundbreaking advancements in AI, such as AlphaGo and AlphaFold. They were determined to create AI systems that could provide more than just statistical outputs; they wanted their technology to foster meaningful interactions in a medical context. Their collaboration led to the development of Med-PaLM, which evaluated the medical knowledge of large language models. Initial tests demonstrated rapid improvement, with models progressing from random guesses to expert-level performance. However, Natarajan and Karthikesalingam sought to push beyond simply passing exams, leading to the creation of AMIE, an AI system designed to gather patient histories and communicate effectively with empathy. The project evolved into Co-Clinician, envisioned as an AI partner within healthcare teams, facilitating interactions between patients and doctors. Their ambitions expanded further when they explored whether AI could assist in generating scientific hypotheses, culminating in the Co-Scientist initiative. This system proved capable of producing valuable research insights, even surprising seasoned researchers with its accuracy. A significant moment occurred when Co-Scientist tackled a complex challenge presented by researchers at Imperial College London. Their groundbreaking work on antimicrobial resistance was matched in quality by the AI-generated hypotheses. The researchers were astonished by the AI's ability to generate relevant and innovative ideas they had not considered. Natarajan’s ongoing projects, including Co-Scientist, aim to bridge the gap between scientific research and real-world medical treatments. He acknowledges the inherent risks of AI in healthcare, from flawed hypotheses to potential patient harm, but remains driven by the urgency to translate scientific advancements into effective therapies. As he reflects on his journey, Natarajan emphasizes the importance of understanding disease mechanisms and hopes that the insights gained from these AI systems will accelerate the quest for cures across various medical conditions.