An innovative AI tool predicts life expectancy with high accuracy

Scientists from the Technical University of Gate have developed an artificial intelligence algorithm called “Life2vec”, capable of estimating a person's probability of death within four years with an accuracy greater than 75%.

This advanced model, detailed in a study published in Nature Computational Science, successfully predicted the deaths that occurred in 2020 with an accuracy rate of nearly 78%, according to the researchers in their report.

To create this system, the team analyzed data from over 6 million citizens, including information on age, health, education, employment, income, and other vital events. This extensive database was provided by the government of the country, which actively collaborated in the research.

What makes Life2vec unique is its innovative approach: it works similarly to a chatbot, using available information to forecast individuals' life expectancy.

Dr. Sune Lehmann, principal investigator of the study titled “Using Vital Event Sequences to Predict Human Lives,” explained the methodology in an exclusive interview. “We employed the technology that supports ChatGPT ( known as transformer models ) to analyze human lives, representing each person as a sequence of events occurring throughout their existence,” he said.

Unlike conventional AI models, Life2vec delves into the more subtle details of a person's life, covering factors such as income, profession, place of residence, and medical history.

Lehmann compared his method to the way words form sentences. Just as words connect meaningfully, events in our lives follow a determined order. They chose to predict mortality because it is a widely studied aspect of human life, allowing them to apply AI to better understand this crucial facet of existence.

To validate the effectiveness of Life2vec, scientists analyzed an extensive group of 6 million people from diverse backgrounds between 2008 and 2020. They used the model to determine which individuals were likely to survive at least four years beyond January 1, 2016. The research demonstrated that their data was versatile and could be used to create detailed timelines of each person's progress over time, helping to understand the different life trajectories.

Ethical debate on the use of AI in mortality predictions

The emergence of Life2vec not only highlights its potential to predict vital outcomes but also drives a critical analysis of the ethical and social implications associated with AI-based mortality predictions.

The accuracy and efficiency of the algorithm raise questions about the potential consequences of such forecasts, including issues of privacy, consent, and the psychological impact on individuals and their families. As society faces the growing capabilities of AI, a broader debate emerges about the responsible use of technology to understand and address the complexities of human existence.

Achieving a delicate balance between technological advancements and ethical considerations becomes crucial as we venture into uncharted territories, redefining the boundaries of the relationship between artificial intelligence and humanity.

The ongoing dialogue seeks to establish guidelines and frameworks that ensure the ethical deployment of AI, promoting a future where technological innovation aligns harmoniously with the preservation of human dignity and well-being.