Machine studying software can predict malignancy in sufferers with a number of pulmonary nodules


A machine learning-based tool was able to predict the risk of malignancy in patients with multiple pulmonary nodules and above-average human experts, previously validated mathematical models and a previously established artificial intelligence tool. This is evident from results published in the journal Clinical Cancer Research of the American Association for Cancer Research.

Tools currently available can predict malignancy in patients with single nodules. Predictive tools for patients with multiple nodes are limited. "With the widespread use of chest computed tomography (CT) for lung cancer screening, the detection of multiple lung nodules has become increasingly common," said study author Kezhong Chen, MD. Vice President of the Thoracic Surgery Department at Beijing University People's Hospital in China. Among the patients who presented with a lung nodule on CT scan in a previous lung cancer screening study, about 50 percent had multiple lumps, Chen said. "Current guidelines recommend the use of clinical models containing nodule and socio-demographic characteristics to estimate the likelihood of cancer prior to surgery. While there are several tools for patients with a single lump, there is currently no such tool for patients with multiple nodules that are an urgent medical need, "added Chen.

To address this unmet need, researchers developed a machine learning-based model to predict the likelihood of lung malignancy in patients with multiple lung nodules. Initially, the study's authors used data from a training cohort of 520 patients (comprised of 1,739 nodes) treated at Peking University People's Hospital between January 2007 and December 2018. Using both radiographic nodal features and sociodemographic variables, the authors developed a model called PKU-M to predict the likelihood of cancer. The model's performance was rated by calculating the area under the curve (AUC), where a rating of 1 corresponds to a perfect prediction. In the training cohort, the model achieved an AUC of 0.91. The most important predictive features of the model included node size, number of nodes, distribution of nodes, and the age of the patient.

The model was then validated using data from a cohort of 220 patients (consisting of 583 nodes) who underwent surgery in six independent hospitals in China and Korea between January 2016 and December 2018. The performance of the PKU-M model in this cohort was similar to its performance in the training cohort with an AUC of 0.89. The researchers also compared the performance of their model with four previous logistic regression models developed to predict lung cancer. The PKU-M model outperformed all four previous models, whose AUC values ​​were between 0.68 and 0.81.

Finally, the researchers conducted a prospective comparison between the PKU-M model, three chest surgeons, a radiologist, and a previously established artificial intelligence tool for diagnosing lung cancer called RX. This comparison was performed on a separate cohort of 78 patients (consisting of 200 nodes) who had surgery between January 2019 and March 2019 in four independent hospitals in China. Similar to the training and validation cohorts, the PKU-M model's performance was achieved with an AUC of 0.87, which was higher than that of the surgeon (with AUCs between 0.73 and 0.79), the radiologist (AUC of 0.75) and the RX model (AUC of 0.76).

"The increasing rate of detection of multiple pulmonary nodules has created an emerging problem in diagnosing lung cancer," said study author Young Tae Kim, MD, Ph.D., professor in the Department of Thoracic and Cardiovascular Surgery at Seoul National University Hospital and the Seoul National University College of Medicine in the Republic of Korea. "Since many nodules have been found to be benign either after long-term observation or after surgery, it is important to carefully evaluate these nodules prior to invasive procedures. Our predictive model, designed solely for patients with multiple nodules, cannot help only mitigate unnecessary surgery, but also make it easier to diagnose and treat lung cancer. "

"Models are designed to aid clinical diagnosis, which means they should be practical," said study author Jun Wang, MD, a professor in the Thoracic Surgery Department at Beijing University People's Hospital. “We have therefore developed a web-based version of the PKU-M model, in which doctors can enter various clinical and radiological characteristics and the software automatically calculates the risk of malignancy in a particular patient. This tool can quickly make an objective diagnosis and can help with the clinical decision-making. "

Because only data from Asian patients were used in this study, it may not be generalizable to a Western population or other population groups, which is a limitation of this study.

The integrated classifier identifies benign lung nodules

More information:
Clinical Cancer Research (2021). DOI: 10.1158 / 1078-0432.CCR-20-4007

Provided by
American Association for Cancer Research

The machine learning tool can predict malignancy in patients with multiple lung nodules (2021, February 24).
accessed on February 24, 2021

This document is subject to copyright. Apart from fair treatment for the purpose of private study or research, no
Part may be reproduced without written permission. The content is provided for informational purposes only.


Please enter your comment!
Please enter your name here