A neural computational aid to the diagnosis of acute myocardial infarction.

摘要: Abstract Study Objective: Accurate identification of the presence acute myocardial infarction in adult patients who present to emergency department with anterior chest pain remains elusive. The artificial neural network is a powerful nonlinear statistical paradigm for recognition complex patterns, ability maintain accuracy when some data required function are missing. Earlier studies revealed that able accurately identify experiencing pain. However, these did not measure performance real time, significant amount may be available. They also use chemical cardiac marker data. Methods: Two thousand two hundred four presenting ED were used train an recognize infarction. Only available at time initial patient evaluation replicate conditions real-time evaluation. Forty variables from histories, physical examinations, ECG results, and determinations then test network. Results: correctly identified 121 128 (sensitivity 94.5%; 95% confidence interval 90.6% 97.9%) specificity 95.9% (95% 93.0% 98.5%), despite fact average 5% (individual range 0% 35%) input by missing on all patients. Conclusion: Network maintenance unavailable suggest potential aid diagnosis during [Baxt WG, Shofer FS, Sites FD, Hollander JE. A computational Ann Emerg Med. April 2002;39:366-373.]