Computing: Machine learning methods can predict human age | Computational natural sciences

A research paper describes a machine learning approach that has the ability to accurately predict various aspects of a person’s life, including the likelihood of dying at an early age and personality nuances.Computational natural sciencesPublished in This model can lead to a quantitative understanding of human behavior.
Sociologists have long debated the question of whether human life span can be predicted. Although the sociodemographic factors that play important roles throughout human life are well understood, they have not been able to accurately predict life outcomes.
Now, Sonny Lehmann and his colleagues are using data on the education, health, income, occupation and other life events of nearly 6 million people registered in the Danish National Register. To use a machine learning approach in our attempt to build a pathway, Lehmann and others have applied language processing technology to express the human life cycle in a model using the method Similar to language. This approach allowed them to generate vocabulary for life events in a way similar to how language models capture complex relationships between words. The proposed model, called “life2vec,” establishes complex relationships between diagnoses and health-related concepts from place of residence to income level, creating an embedded vector that serves as the basis for predicting life outcomes, and encodes an individual’s life through expression. Lehman and his colleagues show that their model predicts premature mortality rates (specifically, the likelihood that 35- to 65-year-olds in this group will be alive for four years starting on January 1, 2016) and personality nuances. We have proven that the data capture capacity is more than 11% higher than the latest model and the base model.
Our findings demonstrate that representing complex relationships between social and health outcomes can accurately predict life outcomes. However, Lehman and colleagues stress that this study is only a possibility, and should only be applied in the real world under regulations that protect individual rights.

doi:10.1038/s43588-023-00573-5

“Highlights from Nature Magazines” is a translation of a statement prepared for the press by Nature’s public relations department. If you require more accurate and detailed information, please be sure to refer to the original paper.

Return to the “Highlights” articles list.