Using Neural Networks to Predict High Blood Pressure from CVD Risk Factors
Author Information
Author(s): Catherine T. Falk, Laura Almasy, Christopher I. Amos, Joan E. Bailey-Wilson, Rita M. Cantor, Cashell E. Jaquish, Maria Martinez, Rosalind J. Neuman, Jane M. Olson, Lyle J. Palmer, Stephen S. Rich, M. Anne Spence, Jean W. MacCluer
Primary Institution: The New York Blood Center
Hypothesis
Can neural networks effectively predict high blood pressure using cardiovascular disease risk factors?
Conclusion
Neural network analysis did not successfully classify individuals into normal and high blood pressure groups based on CVD risk factors.
Supporting Evidence
- Neural networks were trained using two different strategies based on CVD risk factors.
- Training success rates were high, but validation success rates were low, indicating overfitting.
- Patterns found by neural networks were not correlated with high blood pressure status.
Takeaway
The study tried to use computer programs to guess if people had high blood pressure based on other health information, but it didn't work well.
Methodology
Data from the Framingham Heart Study was used to train neural networks to classify individuals based on CVD risk factors.
Potential Biases
The study faced risks of bias due to unbalanced sample sizes between high and normal blood pressure groups.
Limitations
The neural networks could not reliably predict high blood pressure, suggesting potential overfitting and issues with the input data.
Participant Demographics
Data included two cohorts from the Framingham Heart Study, but specific demographics were not detailed.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website