Artificial intelligence can effectively predict premature death

Artificial intelligence can effectively predict premature death

Computer algorithms can learn to predict premature death, according to a study by British researchers. Such applications of artificial intelligence can significantly improve preventive health care.

Researchers at the University of Nottingham who analyze healthcare data have developed special algorithms based on machine learning technology. All to predict the risk of premature death from chronic diseasesob in a large population.

In tests, the researchers observed that the artificial intelligence system they created was highly accurate in its predictions and performed better than the current standard approach developed by expertow.

The study was published in the journal „PLOS ONE”.

Syndromeoł researchersoIn the study, he used health data from over poł million waspsob between the ages of 40 and 69. The data was collected in 2006–2010 and continued through 2016.

– Preventive medicine is a growing priority in the fight against serious diseases. We have been working for many years to improve the accuracy of computerized health risk assessment in ogoln the general population. Most applications focus on a single disease area, but predicting death from several roof different factorsow is highly complex, especially given the environmental and individual factors thatowhich can affect them – said Dr. Stephen Weng, head of theowny author of the publication.

– We have taken a big step forwardod in this area. We have developed a unique approach to predicting the risk of premature death using the systemoa in self-learning. We are using computers to create new risk prediction models thatore take into account a wide range of demographic, biometric, clinical and lifestyle factors for each person assessed. The models are so accurate that they even take into account the consumption of fruitow, vegetables and meat per day by a person – Weng added.

Teamoł scientistsow tracked mortality data, cancer registries, and even the number of hospital admissions and cfown them with results reported by an artificial intelligence system. The researchers noted that the algorithms were significantly more accurate in predicting death than standard prognostic models.

System modelsoin the self-learning used in this study are referred to as deep machine learning. The researchers also used a statistical method known as random forest. The two methods were contrasted with the traditional model for predicting the so-called “low” incidence of the disease. Cox regression model based on age and sex – considered to be the least accurate in predicting mortality – as well as with the multivariate Cox model, whichory worked better, but tends to over-predict risks.

– There is currently a lot of interest in the possibility of using artificial intelligence or machine learning to better predict the outcome ofoin health. In someor situations it may prove helpful, in others it does not. In this particular case, we have shown that with careful tuning, algorithms can significantly improve efficiency – said Joe Kai, coopublication router.

– These techniques may be new to many osob involved in health research and difficult to track. However, we believe that by clearly reporting these methods in a transparent mannerob, it may helpoc in the scientific verification and future development of this exciting field – added Kai.

University of Nottingham researchers predict that artificial intelligence will play an important role in personalized medicine. They hope to help develop effective predictive tools to develop this branch. Further research requires verification and validation of the developed algorithmsoIn artificial intelligence on other population groups and study the wayoIn implementing these systemoin to routine health care.