artificial intelligence
Cred­it: Tara Win­stead from Pex­els

Researchers in Den­mark are har­ness­ing arti­fi­cial intel­li­gence and data from mil­lions of peo­ple to help antic­i­pate the stages of an indi­vid­u­al’s life all the way to the end, hop­ing to raise aware­ness of the tech­nol­o­gy’s pow­er, and its per­ils.

Far from any mor­bid fas­ci­na­tions, the cre­ators of life2vec want to explore pat­terns and rela­tion­ships that so-called deep-learn­ing pro­grams can uncov­er to pre­dict a wide range of health or social “life-events”.

“It’s a very gen­er­al frame­work for mak­ing pre­dic­tions about human lives. It can pre­dict any­thing where you have train­ing data,” Sune Lehmann, a pro­fes­sor at the Tech­ni­cal Uni­ver­si­ty of Den­mark (DTU) and one of the authors of a study recent­ly pub­lished in the jour­nal Nature Com­pu­ta­tion­al Sci­ence, told AFP.

For Lehmann, the pos­si­bil­i­ties are end­less.

“It could pre­dict health out­comes. So it could pre­dict fer­til­i­ty or obe­si­ty, or you could maybe pre­dict who will get can­cer or who does­n’t get can­cer. But it could also pre­dict if you’re going to make a lot of mon­ey,” he said.

The algo­rithm uses a sim­i­lar process as that of Chat­G­PT, but instead it ana­lyzes vari­ables impact­ing life such as birth, edu­ca­tion, social ben­e­fits or even work sched­ules.

The team is try­ing to adapt the inno­va­tions that enabled lan­guage-pro­cess­ing algo­rithms to “exam­ine the evo­lu­tion and pre­dictabil­i­ty of human lives based on detailed event sequences”.

“From one per­spec­tive, lives are sim­ply sequences of events: Peo­ple are born, vis­it the pedi­a­tri­cian, start school, move to a new loca­tion, get mar­ried, and so on,” Lehmann said.

Yet the dis­clo­sure of the pro­gram quick­ly spawned claims of a new “death cal­cu­la­tor”, with some fraud­u­lent sites dup­ing peo­ple with offers to use the AI pro­gram for a life expectan­cy prediction—often in exchange for sub­mit­ting per­son­al data.

The researchers insist the soft­ware is pri­vate and unavail­able on the inter­net or to the wider research com­mu­ni­ty for now.

Data from six million

The basis for the life2vec mod­el is the anonymized data of around six mil­lion Danes, col­lect­ed by the offi­cial Sta­tis­tics Den­mark agency.

By ana­lyz­ing sequences of events it is pos­si­ble pre­dict life out­comes right up until the last breath.

When it comes to pre­dict­ing death, the algo­rithm is right in 78 per­cent of cas­es; when it comes to pre­dict­ing if a per­son will move to anoth­er city or coun­try, it is cor­rect in 73 per­cent of cas­es.

“We look at ear­ly mor­tal­i­ty. So we take a very young cohort between 35 and 65. Then we try to pre­dict, based on an eight-year peri­od from 2008 to 2016, if a per­son dies in the sub­se­quent four years,” Lehmann said.

“The mod­el can do that real­ly well, bet­ter than any oth­er algo­rithm that we could find,” he said.

Accord­ing to the researchers, focus­ing on this age bracket—where deaths are usu­al­ly few and far between—allows them to ver­i­fy the algo­rith­m’s reli­a­bil­i­ty.

How­ev­er, the tool is not yet ready for use out­side a research set­ting.

“For now, it’s a research project where we’re explor­ing what’s pos­si­ble and what’s not pos­si­ble,” Lehmann said.

He and his col­leagues also want to explore long-term out­comes, as well as the impact of social con­nec­tions have on life and health.

‘Public counterpoint’

For the researchers, the project presents a sci­en­tif­ic coun­ter­weight to the heavy invest­ments into AI algo­rithms by large tech­nol­o­gy com­pa­nies.

“They can also build mod­els like this, but they’re not mak­ing them pub­lic. They’re not talk­ing about them,” Lehmann said.

“They’re just build­ing them to, hope­ful­ly for now, sell you more adver­tise­ments, or sell more adver­tise­ments and sell you more prod­ucts.”

He said it was “impor­tant to have an open and pub­lic coun­ter­point to begin to under­stand what can even hap­pen with data like this”.

Pernille Tran­berg, a Dan­ish data ethics expert, told AFP that this was espe­cial­ly true because sim­i­lar algo­rithms were already being used by busi­ness­es such as insur­ance com­pa­nies.

“They prob­a­bly put you into groups and say, ‘Okay, you have a chron­ic dis­ease, the risk is this and this’,” Tran­berg said.

“It can be used against us to dis­crim­i­nate us so that you will have to pay a high­er insur­ance pre­mi­um, or you can’t get a loan from the bank, or you can’t get pub­lic health care because you’re going to die any­way,” she said.

When it comes to pre­dict­ing our own demise, some devel­op­ers have already tried to make such algo­rithms com­mer­cial.

“On the web, we’re already see­ing pre­dic­tion clocks, which show how old we’re going to get,” Tran­berg said. “Some of them aren’t at all reli­able.”


More information:Germans Savcisens et al, Using sequences of life-events to predict human lives, Nature Computational Science (2023). DOI: 10.1038/s43588-023-00573-5© 2024 AFPCitation:How long you got? Danish AI algorithm aims to predict life, and death (2024, March 21)retrieved 3 April 2024from document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, nopart may be reproduced without the written permission. The content is provided for information purp

oses only.