There is no mysticism: scientists have launched a neural network that predicts a person’s life and date of death
Imagine a computer program that can quite accurately guess the character of any person (for example, whether they are an introvert or an extrovert) and even calculate the approximate year of their death – solely based on the facts of their biography.
This is exactly what the life2vec website, launched last month in Denmark, is doing. The process of development of the yogo algorithm avtori reported in the breast issue of the journal Nature.
The developers emphasize that there is no mysticism or fantasy here. The accuracy of the calculations is ensured by the enormous volume of the computer-analyzed database, which contains all known statistics about each of the country’s six million inhabitants.
Including purely personal, medical ones. In other words, everything that is available to the National Statistical Office.
“It’s not so much about the correctness or incorrectness of the predictions we make, but rather about gaining new knowledge,” says one of the leading developers, Professor Sjune Lehmann of the Technical University of Denmark. “This knowledge can be shared and turned into action.
“Divination is not threshing.”
As Prof. Lehman explains, the life story of any person can be represented as a chain of events that happen to them sequentially.
Observing millions of similar processes, artificial intelligence begins to notice connections between recurring life events, some patterns, and identifies the most important factors of change, including three main ones: education level, income level, and chronic diseases.
The identified connections are used to predict the probability of a particular development of events. In particular, the probability of death from any cause at a particular age is also estimated.
Many people whose risk of dying was determined by the algorithm as high have already died. In the low probability zone, there are causes, such as car accidents, that are almost impossible to predict.
The researchers urge not to make far-reaching personal conclusions from the program’s forecasts, but to consider them as an indicator of the state of Danish society as a whole.
They also remind us that they conducted their research in a country with a unique culture, its own laws and social norms.
This means that it makes little sense to transfer the model to other countries.
