News

Lifetime: Can an AI predict when patients will die?


Computers can recognize diseases on pictures of organs
Since the beginning of mankind, we have been asking ourselves whether there is a way to predict our life expectancy. Researchers developed an artificial intelligence that is able to evaluate images of organs and thereby predict the approximate time of death.

The University of Adelaide researchers found that newly developed artificial intelligence is able to predict our remaining life expectancy. The experts published a press release on the results of their study.

Prediction accuracy of 69 percent
Computer-based analysis was able to predict with 69 percent accuracy which patients would die within the next five years compared to predictions by experts, the University of Adelaide said.

Prediction can affect the early detection of serious illnesses
The computer-based evaluation of scans of the organs made it possible to calculate the remaining lifespan of the patients relatively accurately, the authors say. This is a further step in the direction of artificial intelligence that can predict the exact time of death of people. Such a prediction could have an impact on the early detection of serious illnesses and the subsequent treatment, the doctors explain. However, the question remains whether it is really desirable to know the time of one's own death.

Biological age and longevity are usually difficult to assess
Predicting a patient's future is useful because it enables physicians to optimize customized treatment for individuals, the researchers explain. The experts add that the precise assessment of the biological age and the prediction of a patient's longevity remain limited, since there are only limited possibilities to look into the human body and measure the health of the organs.

Computers learn to recognize the complex imaging phenomena of diseases
In the current study, a technique was used by which an artificial intelligence can learn how exactly images have to be analyzed and evaluated, explains author Dr. Luke Oakden-Rayner from the University of Adelaide’s School of Public Health. Although only a small sample of patients was examined for this study, the computer has learned to recognize the complex imaging phenomena of diseases, the researchers say. This actually requires extensive training from human experts.

Artificial intelligence doesn't just focus on diagnosing diseases
What exactly the computer system in the images have to analyze for its predictions is still unclear, the researchers explain. Most predictions were made for patients with severe chronic diseases such as emphysema and congestive heart failure. Instead of focusing on diagnosing diseases, the automated systems can also predict medical outcomes in a way that normal doctors are not trained for, the experts explain. To do this, the computers include very large amounts of data and capture subtle patterns, says author Dr. Oakden-Rayner continues.

Breakthrough in the detection of serious diseases?
Our research opens new avenues for the application of technology with artificial intelligence in medical image analysis, says the expert. The scientists hope that this could lead to a breakthrough in the detection of serious illnesses and enable treatment through specific medical interventions. The same technique may also be used in the future to predict other important medical events such as heart attacks. (as)

Author and source information



Video: Five Reasons Computers Predict When Youll Die The Dr. Data Show (October 2021).