AI Revolutionizing Early Detection and Speed of Blood Tests: A Leap Forward in Cancer and Infection Diagnosis
This is the third installment in a six-part series exploring how artificial intelligence (AI) is transforming medical research and treatment approaches.
The Challenge of Ovarian Cancer Detection
Ovarian cancer is often referred to as “rare, underfunded, and deadly,” a statement by Audra Moran, the head of the Ovarian Cancer Research Alliance (Ocra), a prominent global charity based in New York. This cancer, which is notorious for its late-stage diagnoses, presents significant challenges for both researchers and patients alike.
As with many cancers, the earlier ovarian cancer is detected, the more treatable it becomes. However, ovarian cancer’s unique nature presents difficulties in early detection. Most cases of ovarian cancer start in the fallopian tubes, meaning by the time it reaches the ovaries, it could have already spread to other areas of the body.
“The earlier you detect ovarian cancer, the better,” says Ms. Moran. “Five years prior to ever having a symptom is when you might have to detect ovarian cancer to affect mortality.”
AI’s Role in Detecting Ovarian Cancer
The good news is that new technologies are emerging, powered by AI, that can potentially detect ovarian cancer in its earliest stages. These advancements are making it possible to spot the cancer before it manifests any noticeable symptoms.
Dr. Daniel Heller, a biomedical engineer at Memorial Sloan Kettering Cancer Center in New York, is leading the development of a revolutionary blood test that employs nanotubes – tiny carbon tubes, 50,000 times smaller than a human hair. The concept behind these nanotubes is that they emit fluorescent light, and researchers have figured out how to tweak their properties to respond to almost any substance in the blood.
These nanotubes can be introduced into blood samples, and depending on what binds to them, they emit light at different wavelengths. While this technology has immense potential, one challenge is decoding the signals that these nanotubes emit. Dr. Heller compares this task to finding a fingerprint match – but in this case, the fingerprint is a pattern of molecules that bind to sensors at varying sensitivities and binding strengths.
The patterns, however, are too subtle for humans to identify. “We can look at the data and we will not make sense of it at all,” says Dr. Heller. “We can only see the patterns that are different with AI.”
AI and Machine Learning: Key to Decoding Data
To make sense of the data, Dr. Heller’s team loaded the data from blood samples into a machine-learning algorithm. The algorithm was trained to differentiate between samples from patients with ovarian cancer and those from individuals without the disease, as well as those from patients with other cancers or gynecological conditions that might resemble ovarian cancer.
One significant challenge in developing AI-based blood tests for ovarian cancer is the scarcity of data due to the rarity of the disease. As a result, much of the available data is isolated in individual hospitals, with minimal sharing between institutions. Dr. Heller refers to training the algorithm on data from just a few hundred patients as a “Hail Mary pass.”
Despite these limitations, the AI system outperformed current cancer biomarkers and achieved impressive accuracy in its initial trials. “The AI was able to get better accuracy than the best cancer biomarkers that are available today – and that was just the first try,” says Dr. Heller.
While further studies are underway to improve accuracy, larger sets of samples will be needed to refine the algorithm. More data, much like in self-driving car technology, will help improve the system over time.
Triaging Gynecological Diseases: A Future Vision
Dr. Heller remains optimistic about the future of AI in cancer detection. “What we’d like to do is triage all gynecological diseases – so when someone comes in with a complaint, can we give doctors a tool that quickly tells them it’s more likely to be cancer or not, or which type of cancer it is?” he envisions. According to Dr. Heller, this could be a reality within three to five years.
Speeding Up Pneumonia Diagnosis with AI
AI’s impact isn’t limited to cancer detection. It’s also making strides in speeding up diagnoses for potentially deadly infections such as pneumonia. For cancer patients, catching pneumonia early is crucial, as the infection can be particularly fatal. Since over 600 different organisms can cause pneumonia, doctors typically need to perform numerous tests to identify the infection.
California-based Karius is leveraging AI to streamline pneumonia diagnosis. Their AI-powered system can pinpoint the exact pathogen within 24 hours, enabling doctors to select the appropriate antibiotic much quicker than before. Alec Ford, the CEO of Karius, explains, “Before our test, a patient with pneumonia would have 15 to 20 different tests to identify their infection in just their first week in the hospital – that’s about $20,000 in testing.”
Karius achieves this by comparing patient test samples to a vast microbial DNA database containing tens of billions of data points. AI makes it possible to rapidly compare the sample to this extensive database, identifying the pathogen and speeding up treatment decisions.
AI in Disease Diagnosis: The Power of Pattern Recognition
AI is proving invaluable in disease diagnostics beyond ovarian cancer and pneumonia. Dr. Slavé Petrovski, a researcher at pharmaceutical giant AstraZeneca, has developed an AI platform called Milton. This platform analyzes biomarkers in the UK Biobank’s vast dataset, successfully identifying over 120 diseases with an accuracy rate exceeding 90%. Dr. Petrovski notes that AI excels at detecting complex patterns in data, where multiple biomarkers must be considered simultaneously to identify a condition.
Dr. Heller applies a similar pattern-matching approach to ovarian cancer, recognizing that while the sensors respond to proteins and small molecules in the blood, pinpointing which specific biomarkers are linked to cancer remains a challenge.
Data Sharing Challenges and the Future of AI
One of the ongoing hurdles in AI development for medical diagnostics is the lack of data sharing. “People aren’t sharing their data, or there’s not a mechanism to do it,” says Ms. Moran. To address this, Ocra is funding a large-scale patient registry that will collect electronic medical records from patients willing to have their data used for algorithm training.
“It’s early days – we’re still in the wild west of AI now,” she acknowledges. But with ongoing efforts and the growing availability of data, AI’s potential to revolutionize disease detection and treatment is becoming clearer every day.
As AI continues to evolve, its ability to diagnose diseases at an early stage and improve the accuracy of medical tests is transforming the landscape of healthcare. The ongoing research and developments in AI offer hope for more effective, faster, and affordable diagnostic tools that can ultimately save lives.
0 Comments