The day I found out when I’m going to die
The day that I found out when I’m going to die started normally. I filled out my DeathClock health profile (height, weight, allergies, family medical history, daily sugar and alcohol intake, and so on) and let the app know when my oldest grandparent died (in her 90s.) I detailed how I rarely sleep more than seven hours per night (I have a toddler) but how I do cardio every day (again, I have a toddler.) In other words, I gave all of my information to a faceless AI app being run from Silicon Valley. Then I signed up for the blood test.
The date of death was decided before the bloodwork — 83 years old, with a “worst case scenario” of 50 and a stretch goal of 104.5 — and announced with an image of a little cartoon Grim Reaper reclining on a donut-shaped inner tube, grinning out the screen as he flashed a victory sign with his skeleton fingers. But this was simply an estimate based on a multiple choice test, and that wasn’t good for me. Besides: 83, really? When my World War Two-era grandparents kept going ‘til their mid-90s? I wasn’t going to accept such slander.
Like a lot of medical facilities in New York City, the phlebotomist’s office was part-professional, part-surreal horror movie. I selected one of the blood testing offices available through DeathClock in the evening and arrived at a nondescript building in midtown the next morning, took the rickety elevator up to the tenth floor, then followed some graying corridors around the building until I arrived at a claustrophobic little office under halogen lights. The waiting room was deserted; there were no sounds of nurses or doctors, though I supposed they were probably somewhere in the back. Behind the desk was a medical receptionist, who looked about as happy to be there as I was.
“Hi,” I said, “I’m here for a blood test.” The woman rolled her eyes and pointed at a piece of paper on a clipboard on the desk in front of her. I dutifully filled out my name (first name only — a detail that struck me as unsettling, though I couldn’t fully articulate why), my appointment time, and what kind of bloodwork I was there for. On a Post-It note, I was told to add my home address. Then I was told to wait.
It was an unorthodox waiting room. The tiny space was peppered with signs that had been handwritten in permanent marker on A4 paper and haphazardly pasted around the place: “DO NOT SLAM TOILET DOOR!”, “DO NOT TOUCH THIS WALL!”, “OPEN THIS ONE QUIETLY!” and so on. As I sat, alone, with my hands in my lap, I watched the receptionist get up from her chair, disappear briefly into a small room beside the desk, and re-emerge, now dressed in scrubs. She then stood came round the desk, stood directly in front of me, and, making a show of looking down at the paper and scanning the completely empty room, shouted: “Holly?” When I got up, she introduced herself, as though we hadn’t met minutes before.
At this point, I was obviously so-so about whether to allow the woman with two personalities to take my blood, but this iteration of her — the phlebotomist identity — was smiley and professional, and I allowed myself to be buoyed along by the idea that I was statistically unlikely to have my organs harvested and my best friend knew where I was. The actual bloodwork was quick and almost painless, and I was out the door within ten minutes. As I exited the rickety elevator at the bottom of building and headed straight to a bakery — per the rules, I’d been fasting since the evening before — I felt a renewed vigor. I was alive, baby! For at least 45 more years!
One thing I will certainly say for DeathClock: they’re fast. I got a text the very next morning that the tests had been completed, and when I logged in to the website, I felt a genuine jolt of victory. My date of death had changed! Holly is going to die Wednesday, February 4, 2028, proclaimed the reclining Grim Reaper, happily, at age 92. Almost an entire decade added in the space of 24 hours, due to some really reassuring cholesterol markers and a good-quality metabolic panel. Bit of a bummer to die right after my birthday, but I supposed it made sense that something would take me out in the winter at that age. And the bloods, oh my the bloods! My lipids were doing well; my C-reactive proteins were all in range. My “bad” cholesterol was low and my “good” cholesterol — the type that cleans up veins and lowers your risk of heart attack and stroke — was unusually high. My “biological age” had been revised down from 36 to 29. My Vitamin D levels were insufficient, but that’s solvable with a supplement. Hell, if I got those in order, then even with my chronic stress and my child-inflicted insomnia, maybe I’d hit my new potential of 104.8.
The reason DeathClock measures these things in particular, founder Brent Franson told me a couple days later, is because they’re cheap, they’re informative, and they’re the most reliable predictor of “phenotypic age” according to medical researchers. The C-reactive proteins look at inflammation in your body that could be indicative of autoimmune issues or disease. Lipids, glucose and other metabolic factors are good at predicting whether you’re at risk for conditions like diabetes. The cholesterol, of course, speaks to your cardiac health.
I tell him I felt a little disappointed when I saw my pre-bloodwork biological age and it was the same as my actual age, and he says that’s a problem in the industry: that there are a lot of health apps out there now that are “artificially deflating”. Like vanity clothes sizes, they slap a number on you and make you feel better. You’re also more likely to share a flattering number around social media, which, of course, is important for a brand in the attention economy. But Franson likes to be a lot more traditional and science-led with his approach: “I mean,” he adds, “this is healthcare.”
Whether or not something is healthcare or not obviously hasn’t stopped Silicon Valley founders from messing around before. I ask Franson if he’s had to deal with investors who were burned by the notoriety of the scamming startup Theranos and its subsequent implosion.
“What we’re doing is the opposite of Theranos,” he says. “We’re doing really basic labs.” He points out that DeathClock is just running tests your doctor could order at your annual physical, from the same major labs that are used by mainstream healthcare professionals. Technically, those labs do have to be signed off by doctors, but I noticed an MD technically signed off my own bloodwork during a quick thirty-second loading screen after I requested their free panel.
That MD might be happy to do it, but are doctors in general happy with the idea of something like DeathClock? On the app, I have a menu of bloods I can order — an Alzheimer’s risk test that looks for genes associated with dementia for $99; a $29 “stress” test that measures levels of cortisol; a $139 hormone test; a $159 micronutrient panel that measures a number of biomarkers, from folate and Vitamin B12 to copper and magnesium. It’s a hypochondriac’s dream. If I upgrade to the paid subscription, DeathClock will give me a list of “Holly’s most likely ways to die”. I can also get a personalized “longevity plan” from their Clinical Board, which includes certified oncologists, endocrinologists and a leading longevity researcher.
“Basically, the healthcare system is really bad at preventative health for a whole bunch of structural and economic reasons,” says Franson, “...and we think the future is basically a well-trained AI, plus a platform that's purpose-built for preventative health that allows you to do a bunch of things without having your doctor as intermediary.” You can then “make better use of the time you do have with your doctor.”
A lot of doctors won’t accept bloodwork results that they didn’t order themselves, but that’s something Franson is confident they can work out. He points out, too, that a lot of people across the U.S. don’t take the time to travel to their annual physical or do bloodwork; this simplifies things for people on the go, connecting up phlebotomy labs with your address and cutting out unnecessary middlemen, and that surely can’t be a bad thing. Although DeathClock has a “provocative brand,” he says — the dancing skulls and the grinning Grim Reapers hanging out in swimming pools, laughing about your date of death, are just the beginning: their launch ad featured Brent himself dragging a body bag around — the actual function is conservative. You can’t order a fully-body scan, for instance, which might lead to a bunch of false positives and an endless anxiety spiral. In other words, he says, “We’re not doing all the crazy kitchen sink stuff.”
Franson knows himself how such testing can get out of control. When deciding which tests to include in DeathClock, he ordered himself a CT scan for a calcium score, which shows levels of calcified plaque in the arteries, on the app’s advice. It wasn’t a test his doctor had recommended, but the AI suggested it, so he did it. Unfortunately for Franson, the scan — which involves injecting a dye into the veins, so the scan can investigate movement of blood around the body — showed his calcium score was fine, but he has an anomalous artery. His right coronary artery was on the left side, putting him potentially at high risk of suddenly dying during exercise.
Franson is an avid gym-goer and a marathon runner, so this news was devastating to him. Finding out whether the anomalous artery was the dangerous, kill-you-dead-in-seconds type or the ordinary, low-risk type would require either drilling into his breastbone and performing open heart surgery or undergoing another procedure that sticks a wire through your carotid artery and tests its composition. The wire procedure carried its own, not insignificant risk of death — a one in a hundred chance — but Franson now felt like he’d have to undergo it or the knowledge would plague him for life. He ended up finding out that he has the non-dangerous type of anomalous artery, but “imagine a world in which I die learning that the condition they’re trying to understand is benign,” he says, laughing. “Like: Oh hey, you know, the good news is this heart anomaly wasn’t going to kill him. The bad news is: we killed him learning that.”
That experience has made Franson acutely aware of which tests to include in DeathClock. He also refined the product so it gives more information about the risks of the tests involved. He still thinks it was a good thing that the AI flagged up to him the CT scan — he’s glad he found out that he has the anomalous artery, and that it isn’t the dangerous kind — but he acknowledges that it would’ve been the same outcome if he’d never known. It’s important, he adds, to keep this top of mind, as it also plays out in medical research: Screening for lung cancer too regularly in high-risk groups can actually heighten your risk of death from lung disease, for instance, because harmless patches that might otherwise have gone away on their own get flagged up and biopsied. Those biopsies do damage to the lungs that can lead to a cascade of events, or can lead to collapsed lungs during the procedure, and so on. In healthcare — especially healthcare where people are making money — everything’s a very delicate balance.
As for how this all fits in among the tech bros of Silicon Valley, Franson laughs as he describes how he recently went on the popular TBPN podcast — a daily show recently acquired by OpenAI — and introduced his product with the statement, “We’re all going to die.”
The show immediately threw up a big red X onscreen to invalidate his statement. That’s typical of a lot of transhumanist-leaning AI founders these days, and even their followers: “You've got the self-experimentation movement. ‘Hey, I wanna be low dosing GLP-1s. I wanna basically be using medications off-label because I heard somebody talk about it on Joe Rogan or on Twitter or whatever it is,’” says Franson. “And you have to make a decision in longevity, like, are you gonna be aligned with the Joe Rogans, the bleeding edge? Or are you gonna be with the researchers, and are you gonna be with the docs? And for us, we're just very firmly with the researcher, with the doctor.” That means, inevitably, that he is “the nerdy, old-school person in this conversation.”
Franson doesn’t have much time for the founders he meets who outright state that they are not going to die, and that they’re building or selling products that can make that come true. “I think that's totally reckless,” he says. “I think if you've got a product to sell and you're intimating immortality, that's one of the oldest frauds in the book. And so what we're saying is: Hey, look, we're all gonna die. That's just a fact of life. And we don't think that we should be in denial about that. Life is precious, and there are things you can do to live longer.”
The aim with DeathClock, he adds, is to empower people to “get an extra 10 or 15 good years. And that's time to walk your kids down the aisle, see your grandkids graduate. That’s really important stuff happening in that period. It really matters.”
Grim Reapers in inner tubes aside, it’s hard to argue with that.
