The primary objective of this Data Analysis grant proposal is to employ and evaluate the ability of Neural networks to serve as clinical tools for predicting clinical coronary events. Several methods will be employed to adapt neural networks for survival analysis, and data from town ling-term clinical studies will be used to develop neural networks which evaluate coronary artery disease risk.
The Specific Aims for this proposal are:
Specific Aim 1 (Relation between Angiographic Progression of Coronary Artery Atherosclerosis using Quantitative Coronary Angiography and Clinical Coronary Events in CLAS): Neural networks will be developed utilizing quantitative coronary angiography (QCA) measures of coronary change.
Specific Aim 2 (Relation between Risk Factors and Serum Lipid Values and Clinical Coronary Events in the LRCCPPT): Neural networks will also be developed utilizing information of serum Lipid values available in the Lipid Research Clinics Coronary Primary Prevention Trial results (LRCCPPT Results, 1984a and Appendix B).
Specific Aim 3 (Evaluation of Methods which Adapt neural Networks to Survival Analysis): Existing neural Network designs cannot readily incorporate the censored observations which exist in long-term clinical studies. Methods developed by our research team (Lapuerta, Azen and LaBree, 1995; and Buckley and James, 1979) provide approaches which can be used to address this problem.
| Azen, S P; Lapuerta, P (1999) Neural networks in outcomes research. Hepatology 29:32S-35S |
