Mode
Text Size
Log in / Sign up

Machine learning models predict falls in older adults but need more testing before doctors use them widely

Share
Machine learning models predict falls in older adults but need more testing before doctors use them …
Photo by Growtika / Unsplash

Researchers looked at many computer programs that try to predict if older people will fall. These programs use special math to look at health data and guess the future. The average accuracy for these tools was about seventy-nine percent. This means they get the answer right roughly four out of five times.

However, the results were very different between studies. Some programs guessed correctly often, while others were not very good at all. The time they looked into the future also changed, from three months up to seven years. Because of this big difference, it is hard to know exactly how well these tools will work in your specific area.

Most of the studies had serious problems with their design. Many only tested the program on the same people who helped build it. This is like testing a new car only on the road where it was made. Only one program was tested on a completely different group of people. Also, the studies did not report enough details about how well the tools matched real life.

Even though these tools show promise, doctors should be careful. The tools might seem better than they really are in the real world. It is important to test them carefully in your own community before using them to help patients stay safe.

What this means for you:
Computer tools can help predict falls, but they need more testing to be trusted by doctors and patients.
Share
More on Falls