Neuropathological findings processed by artificial neural networks (ANNs) can perfectly distinguish Alzheimer's patients from controls in the Nun Study
2007

Using Artificial Neural Networks to Distinguish Alzheimer's Patients from Controls

Sample size: 62 publication 10 minutes Evidence: high

Author Information

Author(s): Enzo Grossi, Massimo P. Buscema, David Snowdon, Piero Antuono

Primary Institution: Bracco SpA Medical Department, Milan, Italy

Hypothesis

Can artificial neural networks effectively differentiate Alzheimer's disease patients from healthy controls based on neuropathological features?

Conclusion

The study found that artificial neural networks can perfectly distinguish Alzheimer's patients from controls based on specific brain lesions.

Supporting Evidence

  • Artificial neural networks achieved 100% accuracy in distinguishing Alzheimer's patients from controls.
  • Linear Discriminant Analysis showed a mean accuracy of 92.30%, indicating the superiority of ANNs.
  • Input relevance analysis highlighted the importance of neurofibrillary tangles in neocortex for differentiation.

Takeaway

Scientists used computers to look at brain samples and found a way to tell if someone has Alzheimer's disease just by looking at certain brain changes.

Methodology

The study analyzed brain samples from 26 Alzheimer's patients and 36 healthy controls using artificial neural networks to assess the presence of neurofibrillary tangles and neuritic plaques.

Potential Biases

Potential bias due to the selection criteria for control subjects.

Limitations

The small sample size limits the generalizability of the findings.

Participant Demographics

Participants included 26 Alzheimer's patients and 36 cognitively normal controls, with some ApoE4 positive individuals in both groups.

Statistical Information

P-Value

<0.001

Statistical Significance

p<0.001

Digital Object Identifier (DOI)

10.1186/1471-2377-7-15

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication