Can Artificial Intelligence Be Used to Develop a Decision Support Tool for Radiotherapy in NSCLC?

By: Sarah Campen, PharmD
Posted: Wednesday, October 12, 2022

The application of artificial intelligence may be valuable for developing a tool to guide treatment decisions for patients with non–small-cell lung cancer (NSCLC) treated with stereotactic body radiotherapy, according to a study published in JCO Clinical Cancer Informatics. Artificial neural networks constructed in this study estimated that nearly half (48%) of patients with Tis–4N0M0 NSCLC who were treated with stereotactic body radiotherapy were at low risk for cancer progression. “Our results are anticipated to open new avenues for neural network predictions and provide decision-making guidance for patients and physicians,” stated Atsuya Takeda, MD, PhD, of Ofuna Chuo Hospital, Kamakura, Japan, and colleagues.

The investigators used two neural networks to stratify 792 patients with Tis–4N0M0 NSCLC into high-, intermediate-, and low-risk cohorts for overall survival and disease progression in the first 5 years after stereotactic body radiation therapy. Multiple clinical, patient, and treatment factors were incorporated in the development of the neural networks, which were tested using internal and external test data sets.

In the largest training data set (n = 576), the 5-year overall survival rates for the high-risk, intermediate-risk, and low-risk groups were 17.8%, 55.1%, and 77.2%, respectively; the corresponding 5-year cancer progression rates were 56.6%, 22.3%, and 5.6%, respectively.

The neural networks resulted in concordance indexes for overall survival of 0.76, 0.68, and 0.69 and AUC for cancer progression of 0.80, 0.72, and 0.70 in the training, internal test, and external test data sets, respectively.

“The performance of the neural networks was consistently superior to that of the traditional statistical models and the other machine learning models,” concluded the study authors.

Disclosure: For full disclosures of the study authors, visit ascopubs.org.

JCO Clinical Cancer Informatics

---