It’s no secret that artificial intelligence and machine learning models have already been implemented in the medical world, using their quantitative and analytical powers to help detect cancers. But new research suggests that now machine learning algorithms could help researchers to predict the risk of early death due to chronic disease in a largely middle-aged population.
Data was collected between 2006 and 2010, and then followed up until 2016, and recruited over half a million people aged between 40 and 69. Artificial intelligence was then used to analyse a wide range of demographic, biometric, clinical and lifestyle factors from the participants. Furthermore, the team in Nottingham conducting the research took daily consumption of fruit, vegetables and meat per day, and using all of the above information they went about predicting the mortality of these individuals.
Using the ONS death records, the UK cancer registry and hospital statistics, machines were then able to map resulting predictions to mortality data from the group of people involved. What was discovered was that machine-learned algorithms were significantly more accurate at predicting death than the standard prediction models which have been developed by human experts.
So, what does this mean?
Essentially, this could be seen as a big win for preventative medicine. Researchers of the study are excited about the results – it means that there could come a time where medical professionals are able to identify potential health threats in patients with frightening accuracy, and then take steps to prescribe the right steps of prevention.
Ultimately, by clearly reporting these methods, it could help with scientific verification and the future development of health care. The research has the potential to help build the foundation for important medical tools, which could be capable of delivering personalised medicine and tailored risk management to individual patients. The research from Nottingham was based on a previous study in which AI was able to predict cardiovascular disease.
Preventative healthcare is a growing priority in the fight against serious diseases, so improving the accuracy of computerised health risk assessment in the general population is a huge development.
The AI machine learning models used in the new study are known as ‘random forest’ and ‘deep learning’ – and these were pitched against the traditionally-used ‘Cos regression’ prediction model which really only looked at age and gender, which was found to be the least accurate at predicting mortality.
There’s currently a lot of interest in the potential for using AI to predict health outcomes, and while it may help in some situations, it’s important to realise that this won’t always be the case. But in this instance, these algorithms, with careful tuning, can usefully improve prediction. The researchers in Nottingham who carried out this study expect that AI will play a vital role in the development of future tools which will be capable of delivering personalised medicine. But, we should note that further research requires verifying and validating these AI algorithms in other population groups and exploring ways to implement these systems into routine healthcare.