Machine learning model predicts time of death from life events with 78% accuracy
-
A new study used machine learning to create detailed life sequences for 6 million people in Denmark to predict time of death with 78% accuracy.
-
The model, called "life2vec," mapped out life events like health records, salary, residences to make predictions.
-
Life2vec outperformed existing methods by 11% and was most accurate for younger people and women.
-
The model can also predict personality traits like self-esteem and sociability without questionnaires.
-
While not ready for real-world use yet, the technology could impact insurance companies and related fields.