Using AI to Identify Subtypes of Dementia Progression
Author Information
Author(s): Birkenbihl Colin, de Jong Johann, Yalchyk Ilya, Fröhlich Holger
Primary Institution: Fraunhofer Institute for Algorithms and Scientific Computing (SCAI)
Hypothesis
Can deep learning methods effectively cluster dementia patients into distinct progression subtypes?
Conclusion
The study found that using AI to identify 'slow' and 'fast' progressing dementia patients can lead to more efficient and cost-effective clinical trials.
Supporting Evidence
- Two distinct subgroups of dementia patients were identified: 'slow' and 'fast' progressors.
- The classifier achieved a prediction performance of 0.70 ± 0.01 in external validation.
- Enrichment trials targeting cognitive outcomes are more than 13% cheaper than conventional trials.
Takeaway
Researchers used AI to group dementia patients into two types: those who get worse slowly and those who get worse quickly. This helps make clinical trials cheaper and more effective.
Methodology
The study used deep learning to cluster patient trajectories based on cognitive and functional scores from two independent cohorts.
Potential Biases
Potential biases may arise from the reliance on cross-sectional data for clustering.
Limitations
The classifiers were not designed for clinical decision support and may not be sufficient for individual patient predictions.
Participant Demographics
The study included early dementia patients, primarily from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and the National Alzheimer's Coordinating Center (NACC).
Statistical Information
P-Value
0.05
Confidence Interval
[95% CI: (5.32, 7.45)]
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website