The AI showed a 76% accuracy in predicting premature death risks as compared to the conventional model with 44% accuracy.
The team of researchers found out that AI can predict premature death risks quite accurately. The systems of computer-based in machine learning algorithms have been successful in predicting premature deaths due to chronic diseases. The scientists studied the health data of around 500,000 people aged from 40 to 69 who were recruited to the UK Biobank between 2006 and 2010 and followed up until 2016. The team of researchers used two types of AIs to evaluate possible risk death of participants.
One type used was for the ‘random forest’ algorithm that included a combination of various branched models designed to test various possible outcomes. The other type was ‘deep learning’, in which layered information helps the computer to learn from different examples. During these tests, factors such as gender, age, smoking history, previous cancer diagnosis were taken into consideration. The results obtained from AI computers were then compared with the results produced by a standard algorithm referred to as the Cox proportional hazards model.
The deep learning algorithm primarily focused on job-related hazards and air pollution followed by alcohol consumption and the use of particular medications. The random forest model focused on body fat percentage, skin tone, fruit and vegetable intake, waist circumference while the Cox model concentrated on physical activity and ethnicity. The researchers found out that, deep learning algorithm machine predicts death risk more accurately than conventional models. It identified death risk with 76% of accuracy whereas the random forest and Cox model predicted with 64% and 44% accuracy respectively. The scientists think that AI can be used to predict death risks in the future; however, it will require more fine-tuning and improvement.