BRAINAI in dementia with Lewy bodies

BRAIN Journal Interactive Companion

Artificial Intelligence in Dementia with Lewy Bodies - Current Applications, Diagnostic Challenges, and Future Perspectives

Current applications, diagnostic challenges, and future perspectives for AI across the DLB disease continuum.

Maria Ciubotaru, Laura Romilă, Alin Ciobica, Mihai Hogas, Bogdan Novac, Ancuta Miler, Ecaterina Tomaziu-Todosia Anton, Otilia Novac, and Adrian Cantemir.

Open Source
DLB AI continuum
AI-assisted diagnosis AI-supported biological characterisation AI-supported prognosis AI-enabled precision neurology Each stage requires validation and evidence of clinical utility before implementation.
Diagnostic supportThe strongest current evidence is for AI-supported diagnosis, not autonomous diagnosis.

Research Console

AI promise, clinical caution

DLB overlaps with Alzheimer’s disease, Parkinson disease dementia, vascular cognitive impairment, frontotemporal degeneration, and mixed pathology. AI is positioned as support for complex integration, not a standalone clinical replacement.

01Clinical problem

Heterogeneous syndrome

DLB involves cognitive impairment, fluctuating attention and alertness, recurrent visual hallucinations, REM sleep behaviour disorder, spontaneous parkinsonism, autonomic dysfunction, and other supportive manifestations.

02AI opportunity

Multimodal integration

Clinical features, neuropsychology, FDG-PET, MRI, EEG, α-synuclein assays, plasma biomarkers, and longitudinal trajectories may be combined to address difficult differential diagnosis and prognosis.

03Boundary condition

Evidence before deployment

Routine autonomous AI-based diagnosis is not justified. Key barriers include limited dataset size and diversity, insufficient external and prospective validation, imperfect reference standards, mixed pathology, spectrum bias, and methodological heterogeneity.

DomainCurrent signalImplementation caution
DiagnosisMachine-learning and deep-learning studies show feasibility for DLB differential diagnosis.Use as diagnostic support; routine clinical implementation remains limited by validation and generalisability.
Biomarkersα-synuclein seed amplification assays aim to detect pathological α-synuclein directly.Methodological variation remains across substrates, replicates, positivity thresholds, and biological specimens.
PrognosisRBD, MCI-LB, multimodal biomarkers, and co-pathology can inform future predictive models.Prognostic AI requires larger longitudinal datasets, external validation, and careful handling of missing data and censoring.
Clinical utilityAI may process large datasets and generate standardised probabilistic estimates.Clinicians remain central for context, atypical presentations, communication, ethics, and patient preferences.

Five-Stage Continuum

From risk signals to precision neurology

    Evidence Explorer

    Representative studies and candidate inputs

    Switch between AI-based diagnostic studies and clinical or biological studies that inform future early detection and prognostic modelling.

      Knowledge Atlas

      DLB-AI concept map

      Explore how clinical heterogeneity, biological markers, model design, validation, and clinical governance connect.

      Article Quiz

      Test your DLB-AI reading

      Answer 10 article-based questions. After each selection, the correct answer is highlighted in green and any incorrect selection in red.