Science is very close to predicting the date of his death – 12/28/2023 – Rodrigo Tavares

The fear of death prevents us from living fully. But what if we could know when we were going to die? How would we live?

By analyzing data on the lives of 6 million people, a team of scientists from the Technical University of Denmark (DTU) managed to create a death prediction model (life2vec). The study was published last week in the journal Nature Computational Science.

Using language models, such as those that power ChatGPT or Bard, the team relied on Danish national registers that contain data on work and health for all citizens.

Hundreds of indicators were analyzed, such as salaries, social benefits, professional positions, educational training, and hospital visits and diagnoses.

The algorithm was compared with official death data up to 2020. In other words, individual histories were created with thousands of pieces of data from which it was possible to identify patterns.

The model was able to predict the death of individuals with a prediction rate of 78%. Errors (22%) were mainly caused by unexpected accidents, homicides or heart attacks, which are difficult to model. He was also able to point out the factors that influence premature death, such as having a low income, having a mental health diagnosis and being male.

Speaking to the column, Jane Greve, scientist at VIVE, a research center associated with the Danish Ministry of the Interior and Health dedicated to analyzing the Danish welfare state, states that the DTU study was only possible because Denmark has with a population database, including individual and public health data, of “high quality.”

The level of accuracy was much higher than the models already used privately by insurance companies to estimate the probability of death of policyholders and calculate the value of a policy.

Certainly the globalization challenges of this study are innumerable. There is still not enough data from other countries. Regulatory frameworks will be erected. Ethical and religious questions will emerge. Technology applied to health (gene therapy, nanotechnology, personalized medicine) will contribute to longevity and make predictions difficult. But it is a matter of time before we know, with much more precision, the effects of our biological, family and economic structure on the durability of life.

Some people may find comfort and purpose in this information. Recognizing the finiteness of life often motivates us to seek significant achievements. At the limit, we could even live in a permanent phase of mania.

Others may feel anxiety about the limitations of free will. At an economic level, issues may arise related to savings, consumption, insurance contracting, succession planning, salary remuneration and public policies.

Jane Greve believes that the Danish state should use the study results to improve public interventions for the benefit of Danes. However, she warns, it is equally important to recognize the potential limitations of data models. This is because its predictions are based on current data that could perpetuate errors.

“If, for example, the model indicates that children of highly educated parents have better chances of surviving cancer treatment, this is because they follow the treatment more rigorously or because the system treats them differently, given your social condition?”

The concept of artificial intelligence was first discussed at a conference held at Dartmouth University in 1956 by mathematician John McCarthy, attended by 20 experts in computer science and cognitive science. McCarthy said that “every aspect of learning or any other characteristic of intelligence can, in principle, be described so precisely that a machine could be created to simulate it.”

This is what has been done in the last decade. For those insurgents against the advancement of machines and defenders of the old anthropocentrism, it is important to note that technology companies already have tons of data about us that they use and sell to predict and influence our behaviors, including electoral ones, as seen in the referendum Brexit and the election of Donald Trump.

In 2018, the Google Brain team used a new type of artificial intelligence algorithm to make predictions, within 24 hours of patients being admitted to hospitals, about their likelihood of death. It was correct in 95% of cases. A study by the Paris Cardiovascular Research Center, presented this year, was able to identify people who had more than a 90% risk of dying suddenly, using artificial intelligence. In other words, the analysis of our individual data already works as a calculator of our actions in life.

At the end of each year, trends, predictions, expectations and New Year’s resolutions abound. We faithfully ask the gods, horoscopes, tarot, runes and orixás about death and life. But what will happen when the reader is able to mathematically predict the date of his death? What will be your New Year’s resolution? Minimize or maximize life?

LINK PRESENT: Did you like this text? Subscribers can access five free accesses from any link per day. Just click the blue F below.