Story at a glance
Increased numbers of patients with Alzheimer’s disease, in addition to a workforce shortage, may hinder care for this population in the future.
To help ease the diagnostic burden on providers, researchers developed an artificial intelligence to classify patients with certain forms of impaired cognition.
When compared with neurologists, the algorithms showed similar diagnostic accuracy.
Diagnosing Alzheimer’s disease can be tricky and as the global population continues to age, the burden of new cases will put added stress on providers.
Using artificial intelligence (AI), researchers at Boston University School of Medicine designed multiple computer models that used patient data to identify disease-specific signatures.
From these signatures, the AI was able to discern which patients had normal cognition, mild cognitive impairment, Alzheimer’s disease, and non-Alzheimer’s disease dementia.
Findings were published in the journal Nature Communications.
“Even in circumstances where a specialized neurologist or neuro-radiologist is busy to directly provide a diagnosis, it is foreseeable that some degree of automation could step in to help, thereby enabling doctors and their patients to plan treatment accordingly,” said co-author Vijaya B. Kolachalama in a statement.
Previous research has demonstrated AI is capable of discerning between absence and presence of a disease.
But the models developed were able to identify certain signals based on dementia related changes in MRI scans. The signals were then found to be associated with brain regions with microscopic evidence of degenerative tissue changes.
Authors noted this unique capability much more closely mirrors real-world scenarios as the computer honed in on the source of the patient’s illness despite multiple possibilities.
Dementia, or chronic alterations in one’s mental status, can be a hallmark of Parkinson’s disease, geriatric depression, or nutritional deficiency as opposed to just Alzheimer’s disease, Kolachalama explained.
“Our study is novel because, unlike work before it, we demonstrate a computational strategy for providing an accurate diagnosis during this diverse landscape of neurologic disease,” he said.
Patient data fed into the algorithms included results of functional testing, demographics, medical history, and MRI scans, all of which can be collected during routine doctors’ visits.
When compared with diagnoses made by neurologists and neuroradiologists, researchers’ models met those of the experts.
The investigators plan to conduct further research including a prospective observational study in memory clinics to better compare the algorithm’s performance with that of clinicians.
“If confirmed in such a head-to-head comparison, our approach has the potential to expand the scope of machine learning for [Alzheimer’s disease] detection and management, and ultimately serve as an assistive screening tool for healthcare practitioners,” they wrote.