Ovarian cancer is often described as “rare, underfunded, and deadly,” a sentiment echoed by Audra Moran, the head of the Ovarian Cancer Research Alliance (Ocra), a global charity based in New York. This cancer is particularly insidious because by the time it shows symptoms, it may have already spread beyond the ovaries. Detecting ovarian cancer early can significantly improve survival rates, yet it remains a challenge, as the disease often starts in the fallopian tubes, making early detection difficult. According to Moran, “Five years prior to ever having a symptom is when you might have to detect ovarian cancer, to affect mortality.”
New advancements in artificial intelligence (AI) and blood testing are giving hope to doctors and patients alike. AI-powered blood tests are emerging as a potential game-changer, helping detect ovarian cancer and other serious conditions at their earliest stages, even before symptoms manifest.
Dr. Daniel Heller, a biomedical engineer at Memorial Sloan Kettering Cancer Center in New York, is at the forefront of this groundbreaking technology. His team is developing a blood test that uses nanotubes extremely small tubes of carbon, which are 50,000 times thinner than a human hair. These nanotubes emit fluorescent light when they interact with various molecules in the blood. Researchers have fine-tuned the nanotubes to respond to nearly anything in a blood sample, enabling scientists to detect even the subtlest traces of ovarian cancer.
However, interpreting the signals from the nanotubes was a major hurdle. The data produced was so complex and nuanced that human researchers could not make sense of it. This is where AI comes into play. By loading the data into a machine-learning algorithm, Dr. Heller’s team trained the AI to distinguish between blood samples from cancer patients and healthy individuals, even distinguishing between ovarian cancer and other diseases that might present similar symptoms.
Training the AI involved using data from hundreds of patients, which is a challenge in itself because ovarian cancer is relatively rare. Additionally, much of the available data is siloed in hospitals with limited sharing for research purposes. Despite these challenges, Dr. Heller’s AI system demonstrated better accuracy than the best existing cancer biomarkers in the first round of testing. The system is now undergoing further studies, and with more data, the algorithm is expected to improve even more, just as self-driving car algorithms get better with more real-world testing.
Dr. Heller envisions a future where AI can triage all gynecological diseases, providing doctors with a tool to quickly determine whether a patient’s condition is likely cancerous, and if so, what type of cancer it may be. He predicts that this technology could be available within three to five years.
In addition to revolutionizing cancer detection, AI is also speeding up the diagnosis of infections like pneumonia, which can be deadly for cancer patients. Pneumonia is caused by hundreds of different pathogens, making it challenging to identify quickly. Karius, a California-based company, uses AI to identify the exact pneumonia pathogen within 24 hours, allowing doctors to administer the correct antibiotics much faster. This AI-powered test has dramatically reduced testing costs and the time required for diagnosis, helping save lives and resources.
AI’s potential in medicine extends beyond just detection it is transforming the way diseases are diagnosed and treated. Dr. Slavé Petrovski, a researcher at AstraZeneca, has developed an AI platform called Milton, which uses UK biobank data to identify patterns in biomarkers for 120 diseases with over 90% accuracy. These patterns are often complex and involve multiple biomarkers working together to indicate the presence of a disease, something only AI can analyze at scale.
Despite these breakthroughs, challenges remain. Data sharing is a critical issue, as many institutions still keep their data siloed. Ocra is working to address this by funding a large-scale patient registry that will allow researchers to access electronic medical records with patient consent, enabling the training of more robust algorithms.
While AI’s role in medical research is still in its early stages, the future holds immense promise. As data sharing increases and algorithms continue to improve, AI may soon become an indispensable tool in the fight against ovarian cancer, pneumonia, and many other diseases, offering a new era of early detection and personalized treatment.
