The AI ​​model that 'predicts people's health, income and death” was officially released after training on data from 6 million people

According to news on December 26, scientists have created a “fortune-telling” AI model that can predict your health, income and death by gaining insight into the life data of millions of people. And the accuracy rate is very high.

If this approach can be shown to work across different social classes, it could provide a huge benefit to society, says Matthew Salganik, a sociologist at Princeton University. Scientists provide a new tool for analyzing the impact on personal destiny under the complex interaction of personal personality and events

This "fortune-telling" AI is called life2vec, which screens multiple data from Danish countries. The database covers employment, health and other data of 6 million people, and combines elements such as income, social welfare, job position and medical history, and then converts it into life experience through synthetic language.

For example: "Agnes earned 30,000 DKK while working as a midwife in a hospital in Copenhagen in August 2010."

The team used data from 2008 to 2016 to train the life2vec model and compared it with data from the Danish National Bureau of Statistics. By taking "will you die before 2020" as the entry point, we found that the accuracy of the prediction results reached 78%

Through the successful confirmation of the life2vec model, we discovered several factors that increase the risk of premature death. Includes low income, mental health diagnosis, etc. Occasional inaccuracies are usually due to unpredictable events, such as accidents or heart attacks. G., Eliassi-Rad, T., Hansen,

L.K

. et al. Using sequences of life-events to predict human lives. Nat Comput Sci (2023).

https://www.php.cn/link/529ec1f7ea830a51f00dafb5b06cf091

The above is the detailed content of The AI ​​model that 'predicts people's health, income and death” was officially released after training on data from 6 million people. For more information, please follow other related articles on the PHP Chinese website!