- A new study says it’s possible to accurately predict a person’s time of death.
- The study used machine learning to create detailed life sequences for 6 million people in Denmark.
- Beyond morbid curiosity, this model has the potential to impact insurance companies’ operations.
Death and taxes may be the only certain things in life. And now, with advances in AI, we may be one step closer to the former prediction.
A new study has found that a subset of AI known as machine learning can be used to accurately predict a person’s death.
The study authors said the “huge scale” of the dataset was key to the study. According to the report, they collected 10 years’ worth of daily records from more than 6 million people living in Denmark.
Datasets include health records, salaries, working hours, housing, and more. According to , researchers used that data to create a deep learning model called “life2vec” to plan detailed sequences of an individual’s life events. study.
To test life2vec, the researchers ran a subset of the data to see if it could predict whether someone survived a four-year period starting in 2016. The researchers knew the answer, but the algorithm didn’t.
“To test how good you are [life2vec] So you select a group of 100,000 individuals, half of whom will survive and half of whom will die,” the study’s lead author, Sune Lehmann, a professor at the Technical University of Denmark, told CNN. They focused on using individuals between the ages of 30 and 55. We live in a time when mortality rates are difficult to predict.
According to CNN, Life2vec was correct about 78% of the time. They found that people with high incomes and those in managerial positions were more likely to survive. It also found that men, skilled workers, or individuals diagnosed with a mental health disorder were more likely to die.
The authors say their model outperformed existing “state-of-the-art methods” by about 11%. Not surprisingly, life2vec performs better for groups with a higher number of health events in their records. Researchers found that the effect was also high in younger and female cohorts.
The researchers also used the model to make predictions about more subjective human details, such as personality traits, typically collected through questionnaires. Life2vec was able to predict a variety of personality traits in humans, from their self-esteem to their sociability.
While life2vec may sound “creepy and crazy,” it is likely to impact work already being done in the real world, especially work led by insurance companies, Lehman told CNN.
Google is also working on AI technology that can predict a person’s death by analyzing health records.
Life2vec is not currently ready to perform “real-world tasks,” but in theory it “could do all kinds of predictions,” Lehmann told CNN.
Watch now: Popular videos from Insider Inc.