Clinical Scorecard: A Blood Transcriptomic Profile Utilizing Machine Learning for Digital Diagnosis and Classification of Alzheimer’s Disease
At a Glance
Category
Detail
Condition
Alzheimer’s disease (AD), a chronic progressive neurodegenerative disorder causing dementia
Key Mechanisms
Amyloid-β deposition, tau hyperphosphorylation, synaptic dysfunction, neuroinflammation, and lactylation-related epigenetic and metabolic alterations
Target Population
Elderly individuals including normal controls, amnestic mild cognitive impairment (aMCI), and AD patients
Care Setting
Clinical and research settings focusing on early detection and patient stratification
Key Highlights
Development of a Lactylation-Derived Score (LDS) based on seven key genes for non-invasive AD diagnosis using blood transcriptomics and machine learning
LDS correlates with disease severity (Braak stage, MMSE) and improves detection of AD and amyloid-tau positive individuals when combined with plasma p-tau181 and p-tau217
Lactylation dysregulation links epigenetic chromatin remodeling and mitochondrial dysfunction, revealing metabolic-immune heterogeneity in AD subtypes
Guideline-Based Recommendations
Diagnosis
Utilize blood-based biomarkers including LDS and plasma p-tau181/217 for early and non-invasive AD detection
Consider transcriptomic profiling to identify lactylation-related gene expression changes for patient stratification
Combine LDS with established plasma biomarkers to enhance diagnostic accuracy for AD and amyloid-tau positivity
Management
Incorporate molecular subtype information (lactylation-based clusters) to tailor therapeutic strategies targeting metabolic and immune pathways
Monitor cognitive decline using MMSE alongside biomarker profiles to guide clinical decision-making
Monitoring & Follow-up
Track LDS scores longitudinally to assess disease progression and response to interventions
Use plasma p-tau181 and p-tau217 levels as complementary markers for monitoring AD pathology
Risks
Recognize limitations of invasive CSF biomarkers and high costs restricting large-scale screening
Account for heterogeneity in lactylation patterns which may affect biomarker interpretation
Patient & Prescribing Data
Elderly individuals including normal controls, aMCI, and AD patients from multi-cohort transcriptomic and plasma cohorts
LDS provides a mechanistically informed biomarker for early detection and stratification, potentially guiding personalized therapeutic approaches targeting epigenetic and metabolic dysfunction
Clinical Best Practices
Apply machine-learning derived LDS in conjunction with plasma p-tau biomarkers for improved AD diagnosis
Use non-invasive blood transcriptomic profiling to overcome limitations of CSF biomarker testing
Consider lactylation-based molecular subtyping to understand patient heterogeneity and optimize management
Integrate cognitive assessments with biomarker data for comprehensive evaluation
