Initial specialist validation of clinical decision support recommendations from a machine learning-enabled digital cognitive assessment - Report - MDSpire
Advertisement
Initial specialist validation of clinical decision support recommendations from a machine learning-enabled digital cognitive assessment
Clinical Report: Expert Assessment of Machine Learning in Cognitive Evaluation
Overview
This study evaluates the appropriateness of clinical decision support (CDS) suggestions derived from a machine learning-driven cognitive evaluation tool for detecting cognitive impairment in older adults. Findings indicate that while most CDS recommendations are deemed appropriate, certain pathways require refinement.
Background
Alzheimer's disease (AD) poses a significant public health challenge, with millions affected globally. Early detection of cognitive impairment is crucial, especially with the advent of disease-modifying therapies. However, cognitive impairment often goes unrecognized in primary care, highlighting the need for effective screening tools.
Data Highlights
Pathway
Median Score
Appropriateness
Borderline/Impaired-DCR
7-8
Appropriate
Cognitively Unimpaired with Green DCR
6
Below Threshold
Anti-Amyloid Treatment Referral
5
Below Threshold
Key Findings
All cognitive-impairment recommendations met the appropriateness threshold.
Seven borderline/impaired-DCR pathways were rated appropriate (median 7-8).
Two pathways fell below the appropriateness threshold: cognitively unimpaired individuals with Green DCR scores and a preliminary anti-amyloid treatment referral pathway.
Moderate agreement among experts was observed (median ICC(2,k) = 0.61).
Lower agreement for individual diagnostic-concern recommendations (median ICC(2,k) = 0.25).
Clinical Implications
The findings support the use of the CCE-derived CDS for PCPs in assessing cognitive impairment in older adults. However, refinement is needed for pathways associated with low-risk and emerging therapies to enhance clinical utility.
Conclusion
The study underscores the potential of machine learning-driven tools in supporting cognitive evaluations, while also identifying areas for improvement to optimize their application in primary care settings.
