Neural networks as a prognostic tool of surgical risk in lung resections

Evidence-based medicine, medical decision analysis, and pathology

“…8 Quantitative methods are being used to integrate complex multivariate information in precise, comprehensive, and cost-effective multidisciplinary guidelines for diagnosis, treatment, estimation of prognosis, and response to therapy. [29][30][31][32][33][34][35][36][37] An epistemological approach to the current practice of pathology would involve the formulation of basic questions regarding existing methods and their intended purposes, a review of the limitations of current practices, a consideration of what kind of new information is necessary in the construction of diagnostic models, and a discussion of how these issues might be addressed more effectively in the future.…”

“…This approach has seldom been applied in the existing literature in surgical pathology and cytopathology. [29][30][31][32][33]42 What Kind of Information Is Provided in Pathology Reports: Diagnosis, Prognosis, Predictions of Response to Therapy, or Combinations Thereof?…”

“…That issue has been considered in a preliminary fashion in some experimental studies on prognostic models by using logistic regression, linear discriminant analysis, artificial neural networks, and other tools for addressing statistical uncertainty. [29][30][31][32][33]42 EVIDENCE-BASED MEDICINE EBM has been defined by Sackett and associates as "the integration of best research evidence with clinical expertise and patient values". [8][9][10]38,44,[58][59][60][61][62][63][64][65][66] It is an evolving discipline that applies the analytical and quantitative methods of statistics, epidemiology, and information science ("informatics").…”

“…[26][27][28] Several cyberapproaches have attempted to simulate the abilities of the human brain with mathematical tools such as algorithms, multivariate statistics, artificial neural networks, fuzzy logic, and Bayesian belief networks. 30,31 The ubiquity of personal computers has also broadened the use of statistical tools, as made available in spreadsheet and other software. These advances have led to the development of new applications in business, military, engineering, medicine, and other fields and to the growth of decision analysis as a distinct discipline.…”

Surgical lung cancer

“…The study populations are frequently heterogeneous due to the different selection criteria used [7,12], the variable proportions of pneumonectomies and of patients aged 70 and over [13], or even due to the inclusion of benign diseases [3,6,[14][15][16]. Additionally, the definition of morbitity and mortality varies among studies [2,5,6,[17][18][19][20][21]. Some differences exist due to the definition of those variables considered for entry into multivariable analysis, such as functional preoperative variables [15,17], or clinical preoperative variables as dyspnoea [3,8].…”