An artificial intelligence-based platform by Professor Feigenbaum of the University of Pennsylvania crosses thousands of existing drugs with thousands of diseases, enabling the discovery of breakthrough treatments for rare and untreatable diseases.
Joseph Coates, 38, has spent much of the past year hovering between life and death. He is battling a rare disease called POEMS syndrome, which has severely damaged his kidneys. He has spent most of his time in a hospital bed, unconscious, with doctors draining gallons of blood from his stomach every week. His condition was so bad that he would not survive one of the standard treatments for the disease: a stem cell transplant. Doctors were ready to give up.
Then salvation came, from an unexpected direction. Someone came to the rescue. Someone truly intelligent and extraordinary: Coates' girlfriend.
Oh, were you expecting artificial intelligence? Wait a minute.
Tara Theobald, Coates' partner, refused to give up. She remembered a doctor she had met a year earlier who had developed an extraordinary platform: artificial intelligence that came up with ideas for innovative treatments for existing diseases. She contacted that doctor – Dr. Feigenbaum – by email and begged for help.
The next morning, Dr. Feigenbaum had already sent her a response, recommending a particularly strange treatment: a combination of chemotherapy, immunotherapy, and steroids. That combination had never been tested or approved as a treatment for Coates' disease.
“[The combination of drugs] seems a little crazy,” Coates’s doctor admitted. In fact, he was concerned that the combination might kill Coates faster. But there were no alternatives, and as the doctor said In an interview with the New York Times, "Someone has to be the first to try."
And so they tried. Coates' body responded quickly to the treatment, and his condition stabilized within a week. After a few months, he was able to cope with the more conventional treatment of a stem cell transplant. Today, a year later, he is in good health.
And it's all thanks to Coates's company, and a little artificial intelligence.
Dr. David Feigenbaum's Artificial Intelligence
When Dr. Feigenbaum was 25, he was diagnosed with a rare disease called Castleman's disease. Conventional treatments were not helping him. Feigenbaum went through rounds of emergency room visits and realized he had to help himself. Luckily, he was studying medicine at the time. He pored over the medical literature, trying to find different treatments for his disease. Eventually, he found a drug that seemed completely unrelated: the kind given to kidney transplant recipients, and is supposed to prevent rejection of the new organ.
And behold, it was a miracle – the drug managed to stop Feigenbaum's disease for more than a decade.
Feigenbaum realized that the world is full of drugs that have already been invented and approved by the authorities, and can be used for more than one purpose. Doctors have reported on treatments with those drugs in one of the millions of articles published in the scientific literature, but those articles tend to sink and disappear into the depths of the Internet very quickly. In many cases, they look like a grocery list: the doctor describes how he treated a particular disease in several trials with different drugs, and how the different substances affected the patient's body. He ultimately focuses on the drug that worked. Who would even notice that one of the substances he included in the same experiment worked in a way that could also help a completely different disease?
Like all good Pokémon, Dr. Feigenbaum evolved to become Professor Feigenbaum at the University of Pennsylvania. In his lab, he focused on finding new uses for existing drugs. He identified several successful treatments, but each such study required his team to focus on one disease and one potential cure at a time. He realized he needed to speed up the process, and founded a nonprofit called Every Cure, which focuses on the use of artificial intelligence.
Now, instead of focusing on one drug and one disease, Feigenbaum’s lab is cross-referencing thousands of drugs and thousands of diseases simultaneously. That’s no exaggeration. The machines Feigenbaum runs test about 4,000 existing drugs against 18,500 diseases known from the medical literature. When the AI identifies a high likelihood that a particular drug could tackle a disease, the researchers carefully examine the logic behind the recommendations. If they’re convinced, they contact doctors who are willing to try the drug.
And sometimes – as in Coates’ case – it works.
More than a decade ago, Feigenbaum found a cure for the rare disease he suffered from. Now, he’s helping others find treatments for that disease, and many others. One of the most recent treatments recommended by AI was for a human patient who was already on hospice at the time. That patient’s doctor was surprised to hear about the recommended treatment: adalimumab, a drug used primarily to treat arthritis.
"I didn't think it would work, because it's a pretty weak type of medicine," the doctor said in an interview with The New York Times. But he and the patient didn't think they had anything to lose. They tried the drug. Within weeks, The healthy patient.
Feigenbaum understands that his AI is not a magician or a miracle worker. He knows full well that it can also be wrong – which is why it is important for human researchers to go through its ideas and carefully monitor them. Human doctors also have a critical part in the process, since they are the ones who can judge whether a patient is even capable of handling some of the treatments that the AI offers them. Still, Feigenbaum’s machine has the potential to create “something out of nothing”: to take all the medical trials conducted over the past century, at a cost of trillions of dollars, and derive new insights from them. Insights that could save many lives.
Given this potential, it is no wonder that Feigenbaum's organization - "Any Cure" - recently won more than $100 million in funding from the United States government andFrom the TED "Aspirational Project" associationThey understood what could be achieved through artificial intelligence: to produce knowledge at a level that rivals that of humans, in order to save lives.
The potential for the future
The definition of a “rare disease” is one that affects fewer than 500 people in the United States. The small number of patients discourages most pharmaceutical companies from trying to find cures for those diseases. After all, why invest billions of dollars in developing a new drug if it will only be sold to a few thousand people at best? As a result, only 7,000 of the XNUMX rare diseases today have a cure. There are recognized and approved treatments..
Artificial intelligence can help find treatments for these rare diseases. Not only that, but it can find treatments in drugs that have already gone through the long, tedious, and expensive approval processes of the American authorities. And it can do this at zero cost compared to pharmaceutical companies, which have to invest millions and billions of dollars in finding each new drug.
We will therefore continue to follow Prof. Feigenbaum's laboratory and others who are advancing the field. And if you yourself are suffering from a rare and untreatable disease, perhaps you too can send Feigenbaum an email and receive a proposal for an innovative treatment the next morning.
Just don't forget to consult your doctor before trying it.
More of the topic in Hayadan:
