Only the most dedicated statistics geeks would get into actuarial science, although the field of study is secretly pretty cool: actuaries are charged with enormous responsibility — their job is essentially to predict the future, although unlike futurologists, there are real and severe consequences to getting it wrong. The world’s “least famous scientists” will be in Lyon, France later this month for the ISFA-Columbia University Actuarial Science Workshop to talk about how smart machines will bring better predictions to an increasingly unpredictable world.

Rather than predict what future technologies will corrupt the brains of future children and grandchildren, actuaries concern themselves with money and risk.

How will climate change affect home insurance claims from hurricanes ten years from now?

If 20-somethings are promised a pension plan on certain terms today, will the world of 40 years from now be able to pay up?

How much will it cost to treat the unknown diseases of the future?

There are literally trillions of dollars on the line in these questions, and the world’s advanced economies could not function if not for actuaries, who make it possible to calculate and manage risk.

The bad news is that the world is getting less predictable. The good news is that prediction tools are getting much, much smarter. A Baltimore company called Insilico Medicine is building deep neural networks — computers with a sort of artificial intelligence that allows for machine learning — to better predict and track health in individuals. Recently, scientists published findings from one such network, designed to track health and aging from a simple blood sample.

Zhavoronkov: a self-aware scientist.

“Actuaries are perhaps among the least famous scientists and do not frequently make headlines, but these scientists are making the world’s economy tick,” says Alex Zhavoronkov, CEO of Insilico. “Actuaries are in charge of building mortality tables, where one decimal point may result in trillions of dollars in liabilities and sophisticated risk models with global impact.”

The researchers trained the network with 60,000 blood samples, tied to the age and gender of the donor. By analyzing the chemical compounds in the blood, the machine was able to “learn” which substances were associated with people of a certain age, and use that to predict the age of newly introduced samples. It also learned which compounds, or biomarkers, served as the best predictors of age.

This sort of artificial intelligence could prove to be enormously useful for all kinds of health related applications. It could be, for example, a way to simply and effectively track health measures over time for a large group of people using a particular medication. If your blood predicts that you are older than your calendar age, it could be a sign that you are aging prematurely in some way that may not be otherwise obvious or easily quantifiable.

Insilico Medicine is betting that extending human health lifespan through artificial intelligence is possible with enough biological data, and sufficiently advanced analytical tools. These smart machines may one day uncover the secrets of long life, and teach us how to postpone or reverse the aging process. That’s good news for governments concerned with ballooning health-care costs and anyone who’s interested in living forever.

Photos via Charis Tsevis/Flickr