Researchers have developed software that detects Alzheimer's using artificial intelligence (AI) at 95% accuracy. 

Stevens Institute of Technology researchers have developed software that detects subtle changes in Alzheimer's patients' languages. The AI algorithm could diagnose Alzheimer's even without in-person testing or expensive scans at 95% accuracy. Also, it can explain the diagnosis and allows physicians to re-check the findings.

"This is a real breakthrough," said Stevens Institute of Technology lead researcher K.P. Subbalakshmi adding that we are "opening an exciting new field of research." Subbalakshmi is the founding director of the Stevens Institute of Artificial Intelligence as well as an electrical and computer engineering professor at the Charles V. Schaeffer School of Engineering. He noted that the process makes it far easier to explain how the AI gave such a diagnosis as well as addresses the trust concerns towards A.I. in the medical field.

According to the National Institute on Aging (NIA), Alzheimer's is a progressive brain disease, characterized by changes in the brain. This leads to a loss of neurons that affects a person's memory, ability to think and, live independently.

Alzheimer's can affect a person's language use, usually replacing nouns with pronouns. For instance, instead of saying, "The boy sat on the chair," the patient would say, "He sat on it." Patients may also say wordy statements, instead of just directly saying what they intend to say. 

The researchers created software that uses attention mechanisms and convolutional neural networks, which learns over time. The algorithm was trained using texts used by both healthy individuals and people with Alzheimer's. Subbalakshmi and her team used tools created by Google to convert each sentence into a unique numerical sequence, called a vector. It could accurately identify common symptoms of Alzheimer's as well as subtle linguistic patterns.

"This is absolutely state-of-the-art," said Subbalakshmi when he presented her work at the 19th International Workshop on Data Mining in Bioinformatics at BioKDD. "Our A.I. software is the most accurate diagnostic tool currently available," she said.

Read also: Your Face Says a Lot: New Algorithm AI Can Detect Heart Disease with a Patient's Facial Photos

Alzheimer's: What you need to know

According to the NIA, there are two types of Alzheimer's: the early-onset and late-onset, which both involve genetics. Here are some differences:

Late-Onset Alzheimer's - It is the most common type as signs first appear around the mid-60s. It may also be caused by a gene called APOE ɛ4.

Early-Onset Alzheimer's - this is a very rare type of Alzheimer's in which first signs emerge between the 30s and mid-60s. It is also caused by genes that pass down from parent to child.

Aside from genetics, various factors may lead to the development of Alzheimer's. These include the link between vascular conditions and cognitive declines like stroke, high blood pressure, heart disease, as well as metabolic conditions like obesity and diabetes. 

Meanwhile, the researchers can also easily incorporate new criteria into the future system, which will make it more accurate. Subbalakshmi explained that the system was designed to be transparent and modular. "If other researchers identify new markers of Alzheimer's, we can simply plug those into our architecture," she added.

Read also: 24,000 Dreams Are Continuation of People's Reality? Here's The Explanation Of A New Study

This is owned by Tech Times

Written by CJ Robles

ⓒ 2021 All rights reserved. Do not reproduce without permission.