There’s not currently a great sure-fire way to predict or even diagnose Alzheimer’s. Spinal taps can show Alzheimer’s biomarkers, but those are invasive and expensive, so most of the time, we just use a series of tests and an analysis of a person’s behavior to determine whether they’re likely to have the disease. But every new advancement we make in understanding Alzheimer’s brings us a step closer to having a better way to predict when it’s coming and diagnose it.
Now new research shows that a person’s writing style may help predict whether they’re likely to develop Alzheimer’s disease in the next several years.
Researchers at IBM used an artificial intelligence program to sift through the written work of 80 men and women, all of whom were in their 80s. Using special algorithms, researchers were able to determine which of their subjects’ writing featured similar styles.
“Certain words that we use every day in speech, like: ‘a,’ ‘and’ – short little words began to fade away, drop out, as speech became more simple over time in writing,” says Dr. Peter S. Pressman, a neurologist at UCHealth.
These simple spelling and grammatical errors were identified in almost all of the subjects who later ended up developing Alzheimer’s disease. Repetition was also common among this group.
The AI program predicted Alzheimer’s with about a 75 percent accuracy rate. Dr. Pressman says that’s good for a computer program but isn’t considered good enough to integrate into routine Alzheimer’s diagnosis processes. There is more work to be done, but he hopes this is a step in the right direction.
“It adds to a lot of research and work that has already been done in this space,” says Dr. Pressman. “I think this is really important because we do really need to diagnosis Alzheimer’s as early as possible.”
the researchers hope that their work may eventually help detect not just Alzheimer’s but other neurological disorders as well.
Watch the video below to learn more about this intriguing research.