Artificial intelligence in lung cancer screening: assessment of the diagnostic accuracy of the algorithm analyzing low-dose computed tomography

Abstract

The diagnostic accuracy of the artificial intelligence algorithm aimed to detect lesions on low-dose computer tomograms has been independently assessed. The dataset formed as part of the lung cancer screening program in Moscow was used. The following indicators have been defined: sensitivity – 0.817%, specificity – 0.925%, accuracy – 0.860%, area under the characteristic curve – 0.930. High accuracy rates demonstrated through the independent assessment indicate a good reproducibility of the results by artificial intelligence using independent data about the population of Moscow

Diagnostic Accuracy of Artificial Intelligence for Analysis of 1.3 Million Medical Imaging Studies: The Moscow Experiment on Computer Vision Technologies

“…Philips Research Lung Cancer Screening (LDCT LC) service demonstrated comparable performance with an initial AUC of 0.930, and at the retrospective stage of the Experiment – 0.92. In the prospective study, AUC decreased to 0.78, followed by specificity (prior – 0.925, retrospectively – 0.87, prospectively – 0.68) 23 . We also previously tested the FBM (FLG) service on two datasets with a balance of normal-to-pathology 50/50 and 95/5 with resulting AUC of 0.74 and 0.64, sensitivity – 0.872 and 0.75, specificity – 0.60 and 0.535, concluding that the accuracy varied depending on the class balance 24 .…”

Patient routing and process approach implementation for regional oncology service

“…Для обеспечения системности организации медицинской помощи больным ключевое значение имеет цифровизация здравоохранения [13,14], позволяющая осуществить сбор, обработку и накопление достоверных статистических данных, оптимизацию маршрутизации пациентов, а также содержит систему поддержки врачебных решений [15][16][17]. Процессный подход в оказании медицинской помощи обычно рассматривается с позиции организации процессов на уровне МО [18,19].…”