Tuesday, April 23, 2019
Home > Global News > Premature death can be predicted-New AI Research

Premature death can be predicted-New AI Research


Medical researchers just found out Artificial Intelligence’s new use to predict anyone’s early death.

In recent time, scientist made an AI system to calculate decade of personal health information which was submitted by the half a million people in UK.After that the AI system is tasked to predict if the people were at risk to dye early. In other words, earlier than the Average life span- from common or regular diseases, they said in the new study.

Leads study Writer Dr. Stephen Weng, an assistant professor of data science and epidemiology at Nottingham University (UN) in the U.K told in a statement that the early death predictions that were made by AI algorithms were more precise and true by the predictions made by machine learning.

To calculate the probability of topic’ Premature Mortality, the scientists tested 2 types of AI:”Deep learning,” in which stored data processing networks help computer to learn from different examples. And second “Random Forest,” A regular AI which combines multiple, models (tree-like) to find out possible outcome.

After that Cox Model is used to find out result by comparing all AI models’comclusions.

By using above three models, the researchers evaluated data in UK Bio bank – an open, free access database of health, genetic and physical data given by more than 500,000 people between 2006 and 2016. During that period, about 14,500 participants died, mainly because of heart disease, respiratory diseases and cancer.

The another study of AI can also detect the future chances encroaching diabetes. It can also predict the onset of autism in infants that were high risk of that disease. While the use of AI this way may unknown to many scientists, showing the methods used in research, “could help in scientific verification and further growth to this interesting field,” coauthor Joe Kai (UN professor of primary care), said in recent presentation.



Leave a Reply

Your email address will not be published. Required fields are marked *