When Sofia Michopoulou looks at a medical image, she sees more than cells and tissues — she sees possibility. As a medical physics expert leading Nuclear Medicine Physics at University Hospital Southampton in the United Kingdom, Michopoulou is part of a growing movement using artificial intelligence to transform how doctors fight cancer. Her team is using AI-powered tools to design new radiopharmaceutical drugs — treatments that use tiny amounts of radioactive material to target and destroy cancer cells while sparing healthy organs.
Radiopharmaceutical therapy is already saving lives for certain cancers, but developing these drugs has traditionally taken years of expensive, painstaking laboratory work. That may be changing fast. Deep learning and generative AI models can now rapidly identify promising drug targets, predict how chemicals will interact inside the body, and engineer stable drug candidates — work that once took teams of scientists months to complete.
"AI-driven computer simulations can identify the most promising pharmaceutical candidates earlier, reduce the current volume of preclinical work, and make early-phase evaluation more focused and efficient," Michopoulou told JMIR Correspondent Benedette Cuffari in a recent feature story published in the Journal of Medical Internet Research.
Beyond speeding up drug discovery, AI is also helping doctors personalize treatment for each individual patient. Specialized neural networks — computing systems inspired by the human brain — analyze medical images to predict how drugs will spread through a patient's body. Some hospitals are even creating digital twins: virtual replicas of a patient's organs that let doctors test different treatment approaches before giving any radiation. This means doctors can fine-tune therapy to maximize damage to tumors while protecting healthy tissue.
The technology isn't perfect yet. One challenge is that AI models need vast amounts of high-quality medical data to learn effectively, and hospitals are rightfully cautious about sharing patient information. However, techniques like federated learning — where AI systems train on data from multiple hospitals without that data ever leaving its original location — could solve the privacy problem while still giving AI the diverse information it needs.
For Michopoulou and her colleagues, the promise is clear: AI won't replace doctors, but it will give them sharper tools and faster paths to treatments that could save more lives. The goal, she suggests, is smarter research that brings effective therapies to patients sooner rather than later.