This wasn't supposed to happen.
In a Silicon Valley conference room I'm being told by a team of cutting-edge computer profilers that I have a 28 percent chance of having a heart attack by 2017, with my chances getting worse each year between now and then. By 2027, my risk jumps to 70 percent.
This compares with my internist's prognosis that my heart attack risk is a mere 4 percent in 10 years, a number he got during a routine checkup by matching up my cholesterol levels, age, and other factors to a scale considered state-of-the-art by most physicians.
When the profilers at Entelos, a computer bio-modeling company, agreed to test me on their advanced algorithm for predicting heart attack, they assumed my results would match up with my internist's findings.
They didn't, which suggests that the traditional assessment missed critical factors caught by the Entelos model-a still-experimental test that the company is developing to sell as a consumer product as early as next year for patients who might be at risk for heart attack.
To create my profile, Entelos ran a series of detailed blood tests assessing everything from lipids and triglycerides to the particle size of my good and bad cholesterol (large, medium, small). I had a CT scan taken of my heart, and an ultrasound of my carotid artery in my neck, which can show a plaque buildup similar to what might be accumulating in a person's heart.
In addition, the modelers plugged in previous results from DNA tests that in some cases reveal a higher-than-normal genetic risk factor for heart attack. (See column: "I'm Doomed. Or Not.")
In the conference room, Entelos' chief innovation officer, Thomas Paterson, is looking a bit sheepish having to deliver my results. Paterson is a computer scientist who co-founded Entelos after working for defense contractors to design, among other things, nuclear-attack scenarios for President Ronald Reagan's aborted Star Wars defense system.
My data shows up as a bright yellow line on a black chart plotting my heart's potential demise. For the next five years, I'm okay, reports the chart, with a nearly zero percent risk, then bam! The line swings up-and up.
There is good news too. The yellow line assumes that I will have a normal weight gain of one pound a year for a man over 40. If I keep a stable weight, my risk factor will be closer to my internist's score, at about 4 percent, a predictive line on the chart that is colored green.
If I really want to reduce my risk, the chart also shows me that taking cholesterol-lowering statins would push my risk score to zero.
The message here: If I gain weight, I have a higher-than-average risk of one day clutching my chest and writhing on the floor, and possibly dying.
The Entelos model draws its predictive power by comparing my results with a population of thousands of simulated patients based on real patients from previous heart studies. This virtual-patient pool also includes data from actual human subjects being tested in clinical trials run by pharmaceutical companies testing new heart drugs.
The modelers look for a cluster of patients in their database that have profiles most closely resembling mine. They then create a "Virtual David" as they call it-a prediction of my heart attack future based on the outcomes of this cluster of real and imagined patients from their model.
One explanation for my unanticipated outcome is a specific gene marker that increases my odds for a heart attack. Another is a mechanism discovered by the model suggesting that my cholesterol level has an unusual linkage with my weight. Too many cheeseburgers and chips will drive my cholesterol up higher and faster than average, putting me on the dangerous "yellow line" future rather than the safer "green line."
Founded in 1996, Entelos is developing its personal heart attack augur (and others for diabetes, Rheumatoid Arthritis, and additional maladies) as an outgrowth of its major business to date: working with pharma companies to simulate how drug candidates might work in humans.
Called "in silico" (in computer) testing, these profiles from Entelos and other companies are becoming sophisticated enough that clients, including Pfizer, Johnson & Johnson, and Merck paid $21.8 million to Entelos last year to model new drugs. One day modeling could reduce patient exposure to possible side effects, and save money and time on a process that takes up to 15 years and costs over $1 billion per successful drug.
The biotech investment and analysis firm Burrill & Co. has predicted that the majority of drug discovery will be done in silico by 2020.
Co-founder and chief technology officer Alex Bangs says that Entelos spent about $50,000 developing my "virtual heart attack" model-the first of its kind. I didn't pay anything; it offered to run the tests on me as a demonstration of its effort.
Once the program is finalized and scaled up, Entelos hopes to price its product between a few hundred dollars for minimal tests to perhaps $2,000 for a deluxe version. This may seem expensive for a still largely untested augur of one's future cardiac health, although it may not be too long before these predictive models become a more integral part of medicine and individual health.Visit Portfolio.com for the latest business news and opinion, executive profiles and careers. Portfolio.com© 2007 Condé Nast Inc. All rights reserved.