Train a large model that predicts the time of human death on a dataset of 6 million people

Mondo Technology Updated on 2024-01-30

According to a report by NewScientist, the research team of the Technical University of Denmark recently released a new large-scale model productThe odds of death for humans can be severed

According to team leader Sune Lehmann J Rgensen, the model is calledlife2vec, trained on a uniquely rich dataset that covers the education, medical and medical visits, diagnoses, income, and occupations of 6 million people from 2008 to 2020.

In the experiment, the research team said they collected data from a group of people between the ages of 35 and 65, half of whom died between 2016 and 2020. The researchers asked Life2vec to determine who was alive and who was dead. The results showed that it was 11% more accurate than the existing AI models or actuarial life tables used to price life insurance policies in the financial industry.

In addition, life2vec provides more accurate** personality test results for a subset of the population than specially trained AI models.

Sune Lehmann J Rgensen argues that the model has been trained with enough data so it should reveal general health and social themes. This means that it can be used to detect health problems early, or it can be used to reduce inequalities – which could open up a whole new paradigm for the future of public health science.

Related Pages