The long search to conquer cancer has long been one of humanity's most significant scientific challenges that demands continuous innovative approaches and research. One such approach which has become common in day-to-day life is the convergence of artificial intelligence (AI) that is reshaping and finding new solutions in the landscape of drug discovery and cancer research in oncology.

An intriguing aspect of generative AI

AI's potential lies in its synergy with natural product drug discovery, that is a prolific source of anticancer agents, with compounds derived from plants, fungi, and marine organisms forming the foundation of numerous therapies. Generative AI accelerates this process by modelling interactions between natural molecules and cancer-related proteins, refining molecular structures for enhanced potency, and even simulating potential resistance mechanisms. This has paved way in multiple areas, including-

  • Insilico Medicine utilized AI to design ISM3091, a cancer treatment for pulmonary fibrosis that progressed from concept to Phase I clinical trials in just a few months.
  • AlphaFold, that is developed by DeepMind, has revolutionized protein structure prediction that enables researchers to assess drug ability and design targeted therapies for complex diseases like cancer.
  • Exscientia uses AI to design drug molecules, for example- for obsessive-compulsive disorder (OCD) that entered clinical trials.
  • BenevolentAI applied its platform to find new drugs, for example for COVID-19.
  • Atomwise utilizes deep learning to predict binding affinity between drug candidates and target proteins, for diseases like leukaemia and Ebola.
  • IBM's AI platform has helped analyze vast amount of scientific literature to identify potential drug targets and connections that researchers might miss.

Challenges, Regulatory and Ethical Considerations

Despite AI providing number of advantages, the use of AI in drug discovery also raises significant ethical concerns that need careful consideration. For example-

  • AI systems rely on multiple data. If that data reflects societal or scientific biases, it could lead to discriminatory outcomes. For instance, AI might prioritize diseases affecting wealthier populations due to available data, neglecting diseases prevalent in underrepresented regions.
  • AI models, especially deep learning systems, might not act appropriate and non- transparent where their decision-making processes aren't easily understood. For example- AI-designed drugs might interact with biological systems in unexpected ways, raising concerns about unforeseen side effects or risks.
  • The datasets used to train AI systems often come from diverse sources, including clinical trials. Ensuring that these datasets are taking care of sensitive data and are ethically obtained is also essential.
  • The adoption of AI in drug discovery also raises critical regulatory and ethical questions. Regulatory bodies, including the FDA, are adapting to this paradigm shift, as evidenced by the approval of AI-designed drugs. However, challenges remain, such as ensuring transparency in AI-driven research, addressing intellectual property concerns, and safeguarding data privacy. The case of Stephen Thaler's AI system, DABUS, underscores the complexities surrounding inventorship and patents in the era of AI.

Our thoughts-

Undoubtedly, Generative AI represents a paradigm shift in the fight against cancer. However, knowing that AI models mainly depend on clean, accurate, and well-structured data current datasets may usually lack representation from diverse populations, resulting in biased or limited AI predictions. Cancer is extremely complex, evolving daily, involving intricate biological pathways and usually unpredictable mutations. Therefore, AI may struggle with understanding the nuances of these dynamic systems, which are usually beyond the reach of even current human expertise. There is no doubt that once trained, AI would definitely help in working at fast paces but, preprocessing and trainings, to AI for such results is something that demands enormous time, dedication and investments. Had it been some standardized area, the predictability and the reliability, would have been much easier to accept and move ahead with.

Authors:

Nisha Wadhwa, Principal Associate, Dentons Link Legal

Disclaimer: The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.