Scientists have made a groundbreaking achievement in the field of artificial intelligence by successfully teaching an AI model to predict human death with an impressive accuracy rate of 78%. The research, published in Nature Computational Science, marks the first successful attempt to combine AI with sociological data. Previous attempts by other teams to predict future health, family relationships, and education using AI models failed to produce accurate results. However, this latest study utilized a large language model (LLM) similar to the one used in ChatGPT, which enabled the AI to analyze vast amounts of data and identify important patterns. The researchers named this model life2vec.
To conduct the experiment, the team obtained permission to use Danish national registers from 2008 to 2016, which provided data on the work and health status of approximately 6 million individuals. The AI algorithm processed this data, including information on salary, social benefits, job titles, hospital visits, and diagnoses, to create a timeline of events. Researchers then tasked the model with predicting whether any individuals in the dataset might die prematurely before 2020.
Impressively, the AI’s predictions were accurate in 78% of cases. It identified several factors associated with an increased risk of premature death, such as low income, mental health diagnoses, and male gender. However, the model struggled with predicting accidents or heart attacks, as these events are inherently difficult to anticipate.
While the results of this study are intriguing, the researchers acknowledge the need for further investigation to determine whether the same algorithm can achieve similar accuracy when working with data from other countries. Sune Lehmann, a researcher at the Technical University of Denmark and one of the study’s authors, believes that this technology could prove beneficial by alerting individuals in advance to take better care of their health and mitigate the risk of premature death.
In summary, scientists have made a significant breakthrough in the field of artificial intelligence by successfully developing an AI model that can predict human death with an impressive 78% accuracy rate. The model, named life2vec, analyzed sociological and health data from Danish national registers to identify patterns and factors associated with premature death. While further research is needed to determine the model’s effectiveness with data from other countries, it holds promise as a tool for identifying individuals at risk and promoting proactive health strategies.
