Interpretable deep learning survival predictions in sporadic Creutzfeldt–Jakob disease
2024

Predicting Survival in Sporadic Creutzfeldt–Jakob Disease

Sample size: 655 publication Evidence: moderate

Author Information

Author(s): Tam Johnny, Centola John, Kurucu Hatice, Watson Neil, MacKenzie Janet, Green Alison, Summers David, Barria Marcelo, Seth Sohan, Smith Colin, Pal Suvankar

Primary Institution: University of Edinburgh

Hypothesis

Survival in sCJD can be predicted using data at or near the point of diagnosis.

Conclusion

Survival in sporadic Creutzfeldt–Jakob disease can be predicted using clinical data collected at diagnosis, with a model that performs similarly to traditional methods.

Supporting Evidence

  • The model had a c-index of 0.732, indicating good predictive performance.
  • 16.3% of participants survived beyond 12 months.
  • Codon 129 polymorphism and CSF 14-3-3 were identified as significant predictive features.

Takeaway

Doctors can use information collected when a patient first shows symptoms of a rare brain disease to guess how long they might live.

Methodology

The study used national CJD surveillance data and developed a survival model using an artificial neural network for survival analysis.

Potential Biases

Potential biases from different clinician raters and retrospective data collection.

Limitations

The dataset was primarily from the UK, limiting generalizability to other populations, and some bias may exist due to retrospective data collection.

Participant Demographics

The study included 655 individuals diagnosed with sporadic Creutzfeldt–Jakob disease, primarily of white British descent.

Statistical Information

Confidence Interval

95% CI

Statistical Significance

p>0.05

Digital Object Identifier (DOI)

10.1007/s00415-024-12815-1

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication