image: @your_photo | iStock
Artificial intelligence (AI) model, life2vec, is capable of predicting life events, including estimating the time of death, by analysing health data
In a groundbreaking research collaboration between DTU, the University of Copenhagen, ITU, and Northeastern University in the US, scientists have developed an artificial intelligence (AI) model that uses large amounts of personal data to predict events in people’s lives, including estimating the time of death.
The research project, outlined in the recent Nature Computational Science article titled ‘Using Sequences of Life-events to Predict Human Lives,’ introduces a cutting-edge model named life2vec.
AI prediction: What is Life2vec?
The life2vec model, based on transformer architecture similar to OpenAI’s ChatGPT, was trained on extensive health and labour market data from 6 million people.
After an initial learning phase, the model demonstrated superior predictive capabilities, outperforming other advanced neural networks. It successfully forecasted outcomes such as personality traits and even accurately estimated the time of death.
Professor Sune Lehmann, the first author of the article and a researcher at DTU, explained, “We used the model to address the fundamental question: to what extent can we predict events in your future based on conditions and events in your past? Scientifically, what is exciting for us is not so much the prediction itself, but the aspects of data that enable the model to provide such precise answers.”
Predicting life events
The predictions generated by life2vec include intriguing insights, such as the likelihood of death within a specified timeframe. Analysing the model’s responses revealed consistent results with existing social science findings, highlighting factors like leadership roles and higher income correlating with increased chances of survival.
It was found being male, skilled, or having a mental health diagnosis was associated with a higher risk of mortality.
Life2vec organises data into a complex system of vectors, structuring information related to the time of birth, education, salary, housing, and health.
Professor Lehmann emphasised the innovative approach: “What’s exciting is to consider human life as a long sequence of events, similar to how a sentence in a language consists of a series of words. This is usually the type of task for which transformer models in AI are used, but in our experiments, we use them to analyse what we call life sequences, i.e., events that have happened in human life.”
Unpredictable life events
However, the researchers acknowledge ethical concerns surrounding the life2vec model, including protecting sensitive data, privacy issues, and potential biases in the data.
These challenges must be thoroughly addressed before the model can be applied, for instance, in assessing an individual’s risk of contracting a disease or experiencing preventable life events.
Professor Lehmann stressed the importance of incorporating these ethical considerations into public discourse, saying, “Similar technologies for predicting life events and human behaviour are already used today inside tech companies that, for example, track our behaviour on social networks, profile us extremely accurately, and use these profiles to predict our behaviour and influence us.”
As the next step, researchers plan to integrate additional types of information, such as text, images, and data about social connections, to enhance the model’s predictive capabilities.
This groundbreaking use of data opens up new avenues for collaboration between social and health sciences, potentially revolutionising our understanding of human life events.
Based on labour market data and health records from 2008 to 2020, the study provides a glimpse into the future of AI applications in predicting and understanding the complexities of individual lives.
