Neurology — Weekly Evidence Brief

Recent advancements in artificial intelligence (AI) are poised to transform the diagnosis and management of neurodegenerative disorders, particularly Alzheimer's and Parkinson's diseases. An AI-driven multimodal computational framework shows promise in classifying amyloid-beta (Aβ) and tau (τ) status in Alzheimer's disease with high accuracy (AUROC of 0.79 for Aβ and 0.84 for τ), addressing barriers related to expensive PET imaging 1. Similarly, a novel AI diagnostic framework achieved 94.2% accuracy in early-stage Parkinson's disease detection, outperforming traditional clinical assessments and showing strength in identifying subtle motor fluctuations and predicting treatment response 2.

Beyond AI, new biomarkers are emerging to refine prognostic predictions in Parkinson's disease (PD) and other neurodegenerative conditions. Elevated cerebrospinal fluid (CSF) phospho-tau (p-tau) and p-tau/t-tau ratios at PD onset predict a higher risk for developing motor complications, suggesting a role for Alzheimer's co-pathology in shaping PD progression 3. Furthermore, substantia nigra hyperechogenicity (SN+) detected by transcranial sonography significantly predicts Parkinson's disease development over 8 years in patients with essential tremor and undetermined tremor, offering a non-invasive risk stratification tool 4. For older veterans with traumatic brain injury (TBI), a prognostic model utilizing electronic health record data can predict a 5-year risk of dementia or death with good accuracy, identifying factors like older age, male sex, and comorbidities as key predictors 5.

Understanding the mechanisms and modifiable factors in cognitive decline is also advancing. High socioeconomic status (SES) and lifestyle activities (LA) are associated with better baseline cognitive performance and can reduce the detrimental effects of amyloid-beta (Aβ) and white matter hyperintensity (WMH) load on language and episodic memory decline 6. Specifically, low SES may lead to faster Aβ-related gray matter atrophy, which mediates episodic memory decline 6. At a cellular level, early Alzheimer's disease is characterized by synaptic spine loss which, even without overt neurodegeneration, can boost responses of remaining synaptic inputs in hippocampal neurons, impacting excitability and plateau potential generation 7. Multimodal imaging, including DTI-ALPS and hippocampal microstructure, combined with CSF profiles, provides stage-specific insights into Alzheimer's disease progression, distinguishing symptomatic AD from preclinical stages based on amyloid-associated perivascular dysfunction and asymmetric tau-driven hippocampal degeneration 8.

In stroke, post-event outcomes and risk stratification are gaining clarity. Higher plasma levels of inflammatory cytokines, specifically CD62E and MIF, in the acute phase after stroke are associated with an increased 5-year risk of recurrent stroke or transient ischemic attack, potentially guiding patient selection for anti-inflammatory secondary prevention trials 9. Furthermore, survivors of ischemic stroke often experience fear of progression (FoP), which can be categorized into distinct profiles (low, moderate with family concerns, high with work concerns) associated with age, functional independence, and monthly income, and significantly impacting quality of life and depression 10. In schizophrenia, deep learning frameworks are unraveling shared and unique brain functional changes linked to clinical severity and cognitive phenotypes, providing insights into neural correlates for treatment targeting 11. Similarly, alterations in task-based fMRI topology reveal less centralized, integrated, and segregated functional networks in schizophrenia patients during cognitive tasks, with lower cognitive flexibility specifically linked to reduced network integration 12.