It seems that Artificial Intelligence (AI) really has no limits. The proposal this time is to try to guess the date nobody wants to know: the day of departure; that, of our death. A study was conducted by the Laboratory of Big Data and Predictive Analysis in Health (Labdaps), part of the School of Public Health of the University of Sao Paulo, with the purpose of developing an algorithm capable of predicting deaths. The results were presented by Estadao, and show that the machine was able to hit 70% of its predictions.
The idea with the study is that curiosity about death is answered, especially for doctors, who can make more appropriate decisions and extend the life of their patients by predicting disease and indicating more appropriate treatments in certain situations. “Such a tool can be used by doctors and hospitals to initiate preventive treatments, set hospitalization priorities and perform clinical interventions," says Alexandre Chiavegatto Filho, director of Labdaps and responsible for the study. “Artificial Intelligence provides information that humans sometimes lack for more assertive decision making.”
To assess a probability of premature mortality, scientists at the University of Nottingham used the genetic, physical and health data of 500,000 English aged 40 to 69 years submitted between 2006 and 2016 to the British Biobank. During this time, 14,500 of the participants had died, mainly from cancer and heart and respiratory diseases.
These data together, the researchers first tested the "deep learning", a type of AI in which superimposed information processing networks help a computer to draw examples. Then, they tried the Decision Tree Forest, a simpler form of AI combining several tree models to account for as many results as possible. They then compared these findings with those of a standard algorithm of the Cox model name.
Today, with advanced AI machine learning technologies, you can now visualize the Convolutional Neural Network. With CXR-risk Neural Network, you can now analyze visual information.
Michael Lu, from Massachusetts General Hospital, is one of the scientists. Lu and his associates built up a convolutional neural system, which is an AI instrument for investigating visual data, called CXR-hazard. He said that the device was another approach to extricate prognostic data from ordinary analytic tests. The instrument utilizes the data that is now there.
Technically, AI allows you to build models to quickly analyze data and deliver results while leveraging historical and real-time data. Through machine learning, healthcare providers can make better decisions about patient diagnosis and treatment options, leading to an overall improvement in medical services.
Previously, it was challenging for healthcare professionals to collect and analyze large amounts of data for effective prediction and treatment because no technology or tools were available. Now, with the development of machine learning, big data technologies such as Hadoop are mature enough to accommodate large-scale applications, and these jobs are relatively easy. In fact, 54% of companies in the market today who are only medically available are using Hadoop as a big data processing tool to get important data analysis results about healthcare. 94% of Hadoop users have received satisfactory results after analyzing huge amounts of data that were previously thought impossible.
This research began as a scientific tutorial on how to develop algorithms in healthcare. The first base was only one thousand people, but now we work with about 500 thousand people, accompanied by more than ten years. We collect from socioeconomic and behavioral data to biochemical information such as blood and urine test results. Putting it all together, the algorithm learns the interaction of factors that leads to death. We have always known that people die not only from one factor but from their interaction. These are socioeconomic, behavioral, genetic, disease factors that interact imperceptibly with the human brain, but the algorithms can capture them.
There is now a growing interest in the potential of using AI to predict health problems, in some situations it may help, in others not, and in this particular case, we have shown that with many precautions, algorithms can improve the predictions in an efficient way. These techniques can be new for many professionals and difficult to follow. But we are convinced that explaining these methods in a transparent way, this could help the scientific verifications and the future development of this exciting field in the world of health ", conclude the authors of the study. But this is obviously not the first of its kind. Scientists have been using artificial intelligence for many years to try to determine the risk of population mortality.
AI leads the new revolution in medicine, opening up countless possibilities. Greater precision in diagnosis and individualized interventions are some examples that should shape the care of the future, which, according to Filho, is near. “No one goes to death just because they eat poorly or just because they are sedentary. It is a complex interaction that leads to serious problems. Therefore, I believe that by the end of the year, there will be a real explosion of AI in health,” he says.
© 2019 THE TECHNOLOGY HEADLINES. ALL RIGHTS RESERVED.