A recent Danish study examining the potential of artificial intelligence in predicting the date of death has stirred considerable controversy on social media since its findings were disclosed in December of last year. The report, featured on the French news website France Info, aims to clarify the reality surrounding the buzz generated by this intriguing study. While the demographic study is indeed groundbreaking, it is essential to note that it does not possess the capability to accurately predict an individual's date of death.
The information circulating on social media claiming that "artificial intelligence can predict death with a success rate of 78.8%" is a result of the Danish study published on December 18, 2023. Six researchers from the Technical University of Denmark conducted the study, utilizing health data from six million Danes to develop an algorithm known as "Life2vec."
In the process, the researchers employed a sample of 50,000 individuals to "train" the algorithm and enable it to comprehend the "mechanisms of death," as outlined in the report. With access to a wealth of data provided by the Danish Statistical Institute, including medical appointment histories, diagnoses, income, education, employment, and working hours spanning from 2008 to 2020, the scientists focused on life expectancy within the demographic of 35 to 65-year-olds from 2008 to 2015.
Sonny Lyman, one of the study's authors, acknowledged the difficulty in predicting life expectancy within this age group. He pointed out that factors such as a poor lifestyle and prolonged exposure to polluted environments can significantly impact an individual's likelihood of death in the next four years. In this context, the algorithm achieved a success rate of 78.8%. However, Lyman emphasized that the algorithm cannot accurately predict the exact date of a person's death.
It is crucial to recognize that the study's findings are specific to Danes aged 35 to 65. Testing the algorithm on a different population group would introduce bias, as life expectancy varies based on variables such as year of birth and country of origin. Florian Bonnet, a researcher at the National Institute for Demographic Studies, highlighted the innovative nature of the study's methodology. Typically, life expectancy calculations consider specific variables like gender, socio-professional group, and geographical origin. However, the Danish study incorporated all these variables simultaneously, making it a distinctive approach to predicting life expectancy.