Researchers have made an advancement in early Alzheimer's disease detection in a study that assessed the speech pattern of 206 participants recruited from a research program at Emory University in Atlanta.
The study, published in the peer-reviewed journal Alzheimer's Association, was led by Dr. Ihab Hajjar, professor of neurology at UT Southwestern Peter O'Donnell Jr. Brain Institute in Dallas.
Alzheimer's disease involves progressive neuropathological changes, some of which may begin decades before other symptoms begin to manifest. Subtle features of connected speech become affected very early on. These changes can be so subtle that they are not detectable by one's family or primary care physician. These patterns may affect sentence structure, word count, and grammatical features. Subtle changes can be detected through natural language processing (NLP). Advancements in machine learning provide an opportunity to explore these changes in a non-obtrusive way.
AI helps cut testing to under 10 minutes
Researchers were able to achieve voice acquisition in 6 to 8 minutes and complete testing in under 10 minutes. The new test is also more reliable than traditional tests being able to identify the changes with greater accuracy, as many of the changes are undetectable to the human ear. Additionally, it is far less labor-intensive than traditional methods. Due to the speed, accuracy, and cost-saving nature of these tests, testing could be implemented by primary care providers with relative ease.
By implementing machine learning tests in primary care, providers can enable both the patients and their families more time to plan and prepare for the future. The changes would also benefit care providers and clinicians, providing them more flexibility in recommending lifestyle changes that would be beneficial to the patient.
However, further studies are still needed in order to confirm the accuracy and effectiveness of machine learning and NLP. Researchers also cautioned that care must be taken with the data collected saying "the ease of collecting voice recording even without the acknowledgment of the individuals should also be considered and safeguards for privacy and prevention of bias and discrimination should be implemented as part of this area development."