From treating cancer to the early detection of ALS, medicine as we know is is being revolutionized by artificial intelligence — at a rate unthinkable just a handful of years ago.
This past August, Moderna and Merck announced a “landmark moment” in the testing of a vaccine for melanoma, the deadliest form of skin cancer.
The trial took place on 1,100 patients and was only possible thanks to AI, which analyzed a piece of each patient’s removed tumor to create a highly personalized version of the vaccine.
“We’re potentially at the dawn of medicine’s golden age,” Zev Williams, director of the super computer-driven Columbia University Fertility Center, told The Post.
“We’re able to have insights and abilities that had never been possible.”
The new breakthrough could help prevent melanoma from spreading further or returning. It is still in the testing stage, but scientists say the results are very positive and the approach and research behind it will likely be adaptable to other cancers in the future.
“None of this is possible without digital technology,” Wade Davis, Moderna’s senior vice president for digital, told The Post. “[The technology] is critical to navigating the inherent complexities that come with creating one therapy for one patient and helping us to move with speed and precision.”
AI’s advantage resides in its ability to analyze vast sums of data — decades of opinions, case studies, research and testing results — more than any one doctor could remember, many times over.
“It allows us to consider the possibilities across all drugs and all diseases,” explained Dr. David Fajgenbaum, who co-founded an AI driven non-profit called Every Cure as well as Castleman Disease Collaborative Network. “We utilize the world’s knowledge, rather than relying only on what is in our brains.”
Kaila Mabus has AI and Castleman Disease Collaborative Network to thank for her health. In her teens, she was diagnosed with Castleman disease, a rare disorder of the lymphatic system. It stymied her immunity, led to what she called “fluid overload” and resulted in renal failure.
She once became so debilitated while shopping that, she told The Post, “I had to lay on the floor of a grocery store because I was too tired to move.”
After trying some dozen treatments, including several rounds of chemotherapy, she feared that the disease was untreatable. So did her doctors. When the teenager was out of earshot, she now knows, “they talked about me not being able to make it.”
Then it was suggested that Mabus try a medication called Ruxolitinib. The drug was designed to treat blood cancer — not Castleman disease — and the FDA had not approved it for such application (nor, to this day, has it).
But Fajgenbaum pioneered an innovative application of artificial intelligence that uses the technology as a kind of matchmaker for diseases and FDA-approved drugs.
“We don’t pick a disease upfront,” explained Fajgenbaum who, in 2010, while attending medical school, got hit with a strain of Castleman Disease different to Mabus’s and found an off-label treatment for his version.
At that point, he did it all by hand, spending hours and hours pouring over research. Now, he’s thrilled to have the technology to be more efficient in helping match sick people with medicines that can help them.
“We tell the AI to score every drug with every disease,” said Fajgenbaum, who authored “Chasing My Cure.” “There are 4,000 drugs and 18,000 diseases. That means there are 75 million matches.”
The technology allows him to speedily pair loads of unexpected combinations. When asked for up-and-coming examples, he mentioned “Botox for depression and Lidocaine [an anesthetic normally used intravenously to treat irregular heartbeats] for breast cancer.”
For Mabus, the Ruxolitinib proved to be a lifesaver.
“Within a couple weeks it was giving me a new life,” said the young woman, who is now attending Marquette University where, inspired by the medical people who saw her through to her recovery, she is studying to be a nurse. “Until then, everything was going not in my favor. I would have needed a kidney transplant for sure. But this kind of woke up my kidneys. I feel fine now. I don’t think I would be here if this treatment was not found for me.”
AI doesn’t just have applications for disease. At Columbia University, researchers are using AI and powerful microscopes to uncover and isolate sperm in the semen of men who feared they were hopelessly impotent because their semen appeared to have no sperm.
“Ultimately,” said Columbia’s Williams, “the challenge is to find one sperm in a sea of billions of cell fragments and debris. Then you need to recover the sperm so gently that you do not damage it. The AI is what allows us to see the sperm.”
Finding those single life-producing cells had previously been the work of humans for whom, Williams said, “It was mind numbing, but AI excels at it.”
The process used by Williams and his colleagues is called the Sperm Tracking And Recovery (STAR) system.
Williams noted one patient where surgery failed to find any sperm but STAR later did.
“We took a sample and recovered one sperm” — which would have been unfindable without the help of AI, the doctor said. “Now they have a baby.”
Meanwhile, researchers at the University of Hong Kong have developed CardiOmicScore, an AI-fueled blood test that predicts heart disease 15 years before its onset.
At UCLA, artificial intelligence, 3D bioprinting and advanced imaging are being combined to pull lifesaving information from hundreds of lab-grown replicas of tumors and create customized, highly effective drug protocols for cancer patients.
“Historically, treatments for patients with cancer have been more or less, a one-size-fits-all approach of surgery and anti-cancer interventions such as chemotherapy,” Dr. Michael Teitell, director of the UCLA Health Jonsson Comprehensive Cancer Center, told The Post. “We now know that within a single patient’s tumor, there can be significant differences in molecular expression patterns and drug responses. The standard of care has been evolving to be more individualized to the biology of each patient’s unique disease.”
Outside of his own work, Fajgenbaum is particularly intrigued by an application that uses a combination of artificial intelligence and audio processing to diagnose early onset ALS and then monitor the progression of the disease.
He doesn’t believe it’s clinically available yet, but he’s excited to see it and other life-saving technology come into use.
“Over these next few years, we’re going to make so much progress,” said Fajgenbaum. “It’s going to be incredible.
