Questions
Can AI really predict when I'll get Alzheimer's, heart disease or cancer?
Topol says it can estimate risk and timing by combining many data types, such as genome, proteins, metabolites and microbiome. It works best with two measurements six months to a year apart, and it is more precise for events three to five years away than 18 years away.
What is an organ clock?
It is a blood-based measure of how fast each organ is aging, built from roughly 11,000 plasma proteins. Topol says it might show, for example, that a person's heart is seven years older than their actual age.
Why can't I just get all these tests today?
Topol says the tests are piecemeal: you go to one provider for your organ clock, another for epigenetic aging, another for metabolites and another for the microbiome. The AI tool is not yet joined to those layers, though he expects companies to combine them within a year or two.
Why does Topol prefer prevention to anti-aging drugs?
He says drugs such as senolytics and cell reprogramming are years away, if they arrive at all, and can raise cancer risk. Prevention using forecasts and lifestyle changes is available now and, in his view, not risky.
Who will be able to afford this kind of forecasting?
Topol says the individual tests cost under $100 or $50 each, so the package could be cheap compared with a total-body MRI. He thinks countries with national health systems, like the UK, will adopt it before the US, where insurers focus on short-term costs and access may favor the affluent.
Facts
Stated by the speakers. Each quote matches the transcript, which shows what was said, not that it is true.
Eric Topol earlier wrote The Creative Destruction of Medicine, about digital medicine, and was about to publish Super Agers, about using AI to promote healthspan.
“I wrote a book called the creative destruction of medicine which was about going digital in medicine and then most recently uh about to be published as a book called super agers”
0:01:53 · Eric Topol · Topol's books
Topol says heart disease, Alzheimer's and cancer each take about 20 years to develop and share links to the immune system and inflammation.
“they all take 20 years to incubate they all have common threads of the immune system and inflammation”
0:03:26 · Eric Topol · Age-related disease
Topol says US genetic nondiscrimination protections do not fully cover disability and life insurance, which is why he would not put genome data in electronic health records.
“we don't protect people fully for things like disability insurance and life insurance and you know various things that are not related to health insurance”
0:06:36 · Eric Topol · Genetic privacy
Topol says large language models such as GPT-4 and Llama were already well suited to medical forecasting because they had ingested the medical literature.
“the models that already were appearing like in gbt4 and and llama and others really were remarkably well suited for for this”
0:09:02 · Eric Topol · AI models
Topol says polygenic risk scores, such as those from 23andMe, give no time range for when a disease might occur.
“what if the risk is when you're 98 rather than 68? Big difference. And there was no way to be able to ascribe a time range.”
0:10:26 · Eric Topol · Polygenic risk
Topol describes the p-tau 217 blood biomarker as a major breakthrough for surveillance of neurodegeneration.
“surveillance for neurodegeneration with these new p-tower 217 biomarkers, which was a really big breakthrough.”
0:11:43 · Eric Topol · Alzheimer's biomarkers
Topol says a single blood tube can yield about 11,000 plasma proteins, which organ clocks use to estimate the biological age of individual organs.
“From a tube of blood, you can get 11 ,000 plasma proteins. It'll tell you which organ.”
0:14:54 · Eric Topol · Organ clocks
Topol cites the Horvath epigenetic clock, which reads methyl groups on DNA, as an example of a molecular aging clock.
“like the Horovath clock, where you look at methyl groups on the DNA.”
0:14:54 · Eric Topol · Epigenetic clocks
Topol says there are roughly 50 anti-aging companies, including Altos Labs, Unity and Insilico Medicine, pursuing drug approaches.
“There's 50 of these anti-aging companies, right?”
0:16:06 · Eric Topol · Anti-aging industry
Topol says Function Health was valued at $2.5 billion, with 50,000 members and a wait list.
“Function has a valuation of $2.5 billion already. It has 50 ,000 members and a wait list.”
0:25:45 · Eric Topol · Longevity startups · 2025
Topol says Neko Health had 100,000 people on a wait list.
“Neco Health has 100 ,000 people on a wait list.”
0:25:45 · Eric Topol · Longevity startups · 2025
Topol says Human Longevity, started about a decade earlier, offered total-body MRI and genome sequencing for about $25,000 and is now almost nonexistent.
“Human longevity they get total body mri and genome sequence this was before there was any ai got all these labs and whatnot that company is just basically not almost non-existent now charging 25,000”
0:24:21 · Eric Topol · Longevity startups
Topol says Function Health's Mark Hyman visited him and offered him a role with financial incentives, which he declined to avoid conflicts of interest.
“Mark Hyman. he flew here, right here sitting in my office, and he offered me to, you know, be part of the company and, you know, get all sorts of financial incentives.”
0:26:47 · Eric Topol · Conflicts of interest
Topol says he obtained the beginnings of his own immunome profile from Infinity Bio, a Johns Hopkins spinoff.
“I got uh an immunome the beginning of an immunome from a company, Infinity Bio, Johns Hopkins spinoff”
0:28:55 · Eric Topol · Immunome
Topol says two measurements taken six months to a year apart can show the pace of amyloid buildup, early cancer and artery plaque.
“if you have just two points on the curve uh you'll know what's about the brain amyloid accumulation what about microscopic cancer showing up what about atheroma forming in arteries”
0:30:25 · Eric Topol · Risk timing
Topol says Mass General Brigham uses a top 5 to 10 percent cutoff to define high genetic risk.
“That's where some medical health systems, like for example, Mass General Brigham has drawn the threshold for genetic risk.”
0:31:45 · Eric Topol · Genetic risk
Topol says that when multi-cancer early detection tests are given to everyone aged 50, about 995 of every 1,000 people test negative, and that adding risk information would raise the yield.
“You know, out of 1000 people that get the test now, 995 are negative.”
0:32:56 · Eric Topol · Cancer screening
Topol says the Wellderly study at Scripps sequenced the genomes of 1,400 people who lived past 85 without these diseases and found very little special in their genes.
“We did a so called WELDERLY study, we called it here. 1400 people lived through past age 85, never sick.”
0:34:43 · Eric Topol · Wellderly study
Topol says Demis Hassabis and Geoffrey Hinton have both said improving health will be AI's most important use, and that they endorsed his book along with Mustafa Suleyman.
“both Demis and Jeff have said that the most important ever use of AI will be to improve our health.”
0:21:56 · Eric Topol · Book endorsements
Topol says brief exposure to Yamanaka factors can make an old mouse young, but giving them to humans risks tumors, so companies are testing them in the knee and the eye.
“many companies are pursuing the yamanaka factors right and you know you brief exposure to them and you can turn a an old mouse into a young mouse”
0:41:02 · Eric Topol · Cellular reprogramming
Topol says senolytic drugs cannot yet distinguish harmful senescent cells from helpful ones, and some such approaches induce cancer in some organisms.
“they can't tell the good senescent cells from the bad senescent cells.”
0:13:41 · Eric Topol · Senolytics
Topol says he was asked to review the UK National Health Service, and expects countries with national systems to adopt preventive forecasting before the US.
“I was asked to review the national health service in the uk they're representative of most other uh rich countries”
0:45:09 · Eric Topol · Health systems
Topol says each of the additional data-layer tests costs under about $100, far less than a total-body MRI, which costs thousands.
“every one of these is, you know, it's below $100, $50, you know.”
0:45:09 · Eric Topol · Test costs
Craig Smith describes the Model Context Protocol as an Anthropic-developed standard for connecting AI systems to different data sources and tools.
“Anthropic developed standard protocol for connecting AI systems to disparate data sources or disparate tools sort of thing.”
0:17:33 · Craig S. Smith · Model Context Protocol
Topol names Metabolon as a company that measures a person's metabolome.
“there's a company, Metabolon, that does that and other companies as well.”
0:18:59 · Eric Topol · Metabolomics