AI Breakthrough Targets Alzheimer’s Underdiagnosis in Underserved Communities
Table of Contents
A new artificial intelligence tool developed by researchers at UCLA is poised to dramatically improve the detection of Alzheimer’s disease, particularly among populations historically facing significant barriers to diagnosis. The findings, published in the journal npj Digital Medicine, offer a promising solution to a critical gap in care and address long-standing health disparities.
The persistent underdiagnosis of Alzheimer’s and related dementias has been a major concern for years. According to recent data, African Americans are nearly twice as likely to develop the neurodegenerative disease compared to non-Hispanic whites, yet receive a diagnosis at only 1.34 times the rate. Similarly, Hispanic and Latino individuals are 1.5 times more likely to have the disease but only 1.18 times as likely to be diagnosed.
“Alzheimer’s disease is the sixth leading cause of death in the United States and affects 1 in 9 Americans aged 65 and older,” said Dr. Timothy Chang, the study’s corresponding author from UCLA Health Department of Neurology. “The gap between who actually has the disease and who gets diagnosed is substantial, and it’s more significant in underrepresented communities.”
Addressing Bias in AI-Driven Diagnosis
Previous attempts to leverage machine learning for Alzheimer’s prediction using electronic health records have often been hampered by inherent biases within traditional diagnostic frameworks. The UCLA team adopted a novel approach called semi-supervised positive unlabeled learning, specifically engineered to promote fairness while maintaining a high degree of accuracy. This method allows the model to learn from both confirmed cases and patients with unknown Alzheimer’s status, a significant departure from conventional techniques.
The model was trained on the health records of over 97,000 patients at UCLA Health. The results demonstrate a substantial improvement in diagnostic accuracy across diverse populations. Sensitivity rates ranged from 77 to 81% for non-Hispanic white, non-Hispanic African American, Hispanic/Latino, and East Asian groups – a marked increase compared to the 39 to 53% sensitivity achieved by conventional supervised models.
Beyond Neurological Symptoms: Identifying Subtle Indicators
The UCLA tool doesn’t solely rely on typical neurological indicators like memory loss. Researchers discovered that patterns in health records, including unexpected conditions such as decubitus ulcers and heart palpitations, could signal undiagnosed Alzheimer’s cases. This broadened scope of analysis allows for a more comprehensive and potentially earlier detection of the disease.
The model’s effectiveness was further validated through genetic data analysis. Patients flagged as potentially having undiagnosed Alzheimer’s exhibited significantly higher polygenic risk scores and genetic markers associated with the disease, specifically elevated counts of the APOE ε4 allele.
Implications for Early Intervention and Equitable Care
Dr. Chang emphasized the potential of this tool to help clinicians identify high-risk patients who could benefit from further evaluation and screening. “Early identification is crucial as new Alzheimer’s treatments become available and lifestyle interventions can slow disease progression,” he stated.
The research team is now planning prospective validation studies in partnering health systems to assess the model’s generalizability and clinical utility before widespread implementation. “By ensuring equitable predictions across populations, our model can help remedy significant underdiagnosis in underrepresented populations,” Dr. Chang concluded. “It has the potential to address disparities in Alzheimer’s diagnosis.”