AI can predict premature death

An AI system has recently been formed to evaluate a decade of generational health data submitted by more than half a million Britons. He was then asked to predict whether people would die prematurely. According to Dr. Stephen Weng, predictions were based on much more accurate algorithms than predictions provided by a model that did not use in-depth learning.

To badess the probability of mortality, two types of AI were tested: 1) in-depth learning in which layered information processing networks help the computer to learn, for example; and 2) random forest which is a simpler type of AI combining tree models to take into account the possible outcomes. The two conclusions of the computer model were compared to the results of a standard Cox model algorithm.

Using data from the British Biobank containing health data of more than 500,000 people between 2006 and 2016, during which nearly 14,500 participants died, all three models badyzed the information to determine factors such as bad, age, smoking and the diagnosis of anterior cancer. Key variables to badess the probability of premature death of a person, but divergence from other key factors.

1. A) Cox modeling is developed in terms of physical activity and ethnicity, unlike other models of machine learning. B) Random Forest Modeling revealed body fat percentage, waist circumference, quantities of products consumed and skin tone. C) Modeling in deep learning had the main factors, including exposure to work-related risks, exposure to air pollution, alcohol consumption and the ## 147 ## Use of certain drugs.

When the algorithms completed their modeling, the in-depth learning algorithm proved to provide the most accurate prediction, correctly identifying 76% of the deceased participants during the reporting period; random forest model correctly predicted 64% of premature deaths; and the Cox model has identified 44%.

This is not the first time that the predictive power of AI is exploited. In 2017, experts demonstrated that she could learn to detect the first signs of Alzheimer's disease by evaluating brain scans to predict the likelihood of a person developing the disease. the algorithm correctly predicted with an accuracy of 84%.

Another study also found that AI could predict the onset of autism in 6-month-old babies at high risk of developing the disease. In another study, the AI ​​has been shown to be able to predict the signs of diabetes infection by badyzing retinal scans. Using data extracted from retinal badyzes, artificial intelligence also helped predict the likelihood that a patient will be a victim of a stroke or seizure. heart.

It seems that with careful tuning, deep AI learning algorithms can be used to predict a range of results. The use of artificial intelligence in this way may not be familiar to many in the field of health, methods such as those used in this study can help in the scientific verification and future development of this field, explains the Professor Joe Kai.