Artificial intelligence, complex system analysis and simulation modeling in recognition of N1–2 lymph node metastases of non-small cell lung cancer

Abstract

Listen

Translate article

17123 Background: Precise recognition of N1–2 lymph node metastases (RLN) of non-small cell lung cancer (LC) means great importance in prediction LC patients (LCP) survival after surgery. We examined the immunologic factors associated with LCP with N0 and N1–2. Methods: In trial (1987–2005) the data of consecutive 289 LCP after complete pneumonectomies/lobectomies and mediastinal lymph node dissection (age = 58.1 ± 0.5 years; tumor size = 4.4 ± 0.1 cm; m = 260, f = 29) with pathologic stage I-III (T1–4N0–2M0) (squamous = 169, adenocarcinoma = 102, large cell = 18; G1 = 67, G2 = 110, G3 = 112; T1 = 100, T2 = 114, T3 = 54, T4 = 21; N0 = 147, N1 = 70, N2 = 72; pneumonectomies = 135, bi/lobectomies = 154) was reviewed. Variables selected for ED study were input levels of 64 immunity blood parameters, sex, age. Representativeness of samplings was reached by means of randomisation based on unrepeated random selection. Logistic regression, clustering, discriminant analysis, neural networks computing, structural equation modeling, Monte Carlo and bootstrap simulation were used to determine any significant regularity. Results: Logistic regression modeling displayed that RLN of LC significantly depended on: CDw26, CD16, phagocytic number, ratio of lymphocytes, T-lymphocytes, CD4+2H to LC cells (P = 0.003–0.043). Neural networks computing, genetic algorithm selection and bootstrap simulation revealed relationships between N1–2 and blood IgM (rank = 1), eosinophils (2), T-lymphocytes (3), CD8 (4), phagocytic number (5), CD8+VV (6), CDw26 (7), B-lymphocytes (8), IS2 (9), CD4 (10), CD4+2H (11), NST-A2 (12), monocytes (13), index of thymus function (14), IgA (15). Conclusions: Correct RLN of LC was 79.2% by logistic regression (odds ratio = 15.65), 82.0% by discriminant analysis and 99.6% by neural networks computing (area under ROC curve = 0.99; error = 0.059). No significant financial relationships to disclose.