SheSMAT Foundation

Principal initial biomedical application

Statistical Innovation for Alzheimer's Disease Research

NeuroGraph Sparse Analytics investigates statistical approaches to complex, interconnected, and incompletely observed biomedical measurements.

The research context

Why Alzheimer's Disease?

Alzheimer's disease is a major neurological and public health challenge. Research increasingly draws on structural neuroimaging, molecular imaging, and cerebrospinal fluid biomarkers. Statistical methods can help investigators analyze these complex measurements together.

Integrating these measurements presents important statistical challenges, particularly when different data modalities are not available for every participant. This is methodology research, not a clinical service, diagnostic tool, or treatment.

Multimodal biomedical data

Different measurements. One analytical challenge.

Structural MRI

Structural magnetic resonance imaging provides information about brain anatomy and structural changes.

PET Imaging

Positron emission tomography can provide information about biological processes, depending on the tracer used.

Cerebrospinal Fluid Biomarkers

CSF measurements can provide molecular information relevant to Alzheimer's disease research.

Blockwise missing data

Not every participant has every measurement

A participant may contribute valuable MRI measurements without PET or CSF data. Complete-case analysis uses only participants with every required measurement and may discard useful observations.

NeuroGraph Sparse Analytics investigates alternatives that use available observations and relationships among biomedical variables, with rigorous evaluation of their assumptions and performance.

Illustrative measurement patterns — not participant data or research results. Check = observed; dash = unavailable.
Measurement patternMRIPETCSF
MRI only
MRI and PET
MRI and CSF
Multiple modalities
Incomplete blocks

Intended contributions

Research Goals

These are intended research contributions, not demonstrated outcomes.

  • Better statistical handling of incomplete multimodal data
  • Incorporation of relationships among biomedical variables
  • Rigorous evaluation of statistical estimation and prediction
  • Improved interpretability and reproducibility of research analyses
  • Support for researchers investigating Alzheimer's-related patterns

Research Data and Attribution

No participant data or ADNI-derived findings are published here. If Alzheimer's Disease Neuroimaging Initiative (ADNI) data are used in future approved research outputs, those outputs must include the applicable acknowledgment and data-source attribution. ADNI does not endorse SheSMAT Foundation.

Discuss Alzheimer's research