Artificial intelligence (AI) can detect patterns in blood tests, offering early warnings for diseases like cancer. This is the third feature in a six-part series exploring how AI transforms medical research and treatment.
Early Detection of Ovarian Cancer
Ovarian cancer is often “rare, underfunded, and deadly,” says Audra Moran, head of the Ovarian Cancer Research Alliance (Ocra). Detecting it early significantly improves survival chances. Most ovarian cancers begin in the fallopian tubes and may spread before symptoms arise.
“Detecting ovarian cancer five years before symptoms can reduce mortality,” explains Moran. Emerging blood tests now use AI to identify the disease’s earliest signs.
AI’s potential isn’t limited to cancer. It accelerates blood tests for other deadly infections, such as pneumonia. Dr. Daniel Heller, a biomedical engineer at Memorial Sloan Kettering Cancer Center, leads research in this area.
Heller’s team developed nanotube-based technology to analyze blood. These carbon tubes, 50,000 times smaller than a human hair, emit fluorescent light. Scientists modify their properties to detect various blood molecules. When introduced to a blood sample, nanotubes emit different light wavelengths based on what binds to them.
Interpreting these signals requires AI. “Humans can’t discern these patterns,” Heller explains. Machine-learning algorithms analyze the data, comparing samples from ovarian cancer patients with those from healthy individuals and others with related diseases.
Training AI for ovarian cancer is challenging because the disease is rare, limiting data availability. Hospitals often silo patient data, restricting access for researchers. Heller compares training algorithms on limited data to a “Hail Mary pass.”
Despite these challenges, AI achieved better accuracy than current cancer biomarkers on the first attempt. Researchers are now expanding the dataset and refining the sensors to improve results. Heller envisions a tool for diagnosing all gynecological diseases within three to five years.
Beyond Cancer: AI Speeds Up Blood Tests
AI also enhances other critical blood tests. For pneumonia, a potentially deadly condition for cancer patients, AI simplifies and accelerates diagnosis. Pneumonia stems from over 600 pathogens, requiring multiple tests to identify the cause.
California-based Karius uses AI to pinpoint pneumonia pathogens within 24 hours. This allows doctors to select the right antibiotic faster. “Before our test, diagnosing pneumonia involved 15–20 tests in the first hospital week, costing $20,000,” says Alec Ford, Karius’ CEO.
Karius’ microbial DNA database contains billions of data points. AI compares patient samples to the database to identify pathogens. Such analysis would be impossible without AI.
AI’s ability to uncover complex patterns extends to other diseases. Dr. Slavé Petrovski of AstraZeneca developed an AI platform, Milton, which identifies 120 diseases using biomarkers from the UK Biobank. Milton achieves a 90% success rate by analyzing intricate patterns in vast data.
Similarly, Heller’s work on ovarian cancer uses pattern matching. “We know the sensors respond to blood molecules, but we can’t yet pinpoint which are cancer-specific,” he says.
Data availability remains a barrier. “People aren’t sharing their data, and there’s no mechanism for it,” Moran adds. To address this, Ocra is funding a patient registry with electronic medical records to train AI.
“It’s still the wild west for AI in healthcare,” Moran concludes.
