Abstract:
Lung cancer remains the leading cause of cancer-related deaths worldwide, highlighting the
urgent need for improved diagnostic and prognostic tools. The identification and prognosis
of lung cancer may be improved by recent developments in machine learning. The lack
of interpretability in classic machine learning models, however, prevents them from being
widely used in clinical settings. Explainable AI (XAI) tools, which provide insights into
the decision-making processes of models, overcome this problem. This paper explores
the application of explainable machine learning approaches to improve the detection and
prediction of lung cancer. Our results demonstrate how explainable models can provide
physicians with valuable information while increasing diagnosis accuracy. A potent tool
for enhancing the identification and prognosis of lung cancer is machine learning. The
analysis of histopathological images for precise cancer classification has shown significant
promise thanks to recent developments in deep learning algorithms. Convolutional neural
networks (CNNs) have been shown in studies to have the ability to identify between
benign and cancerous lung tissue with high accuracy. For example, a research that used
CNNs to diagnose lung cancer on digital pathology images had accuracy rates of over
97%. Traditional machine learning models, however, frequently lack interpretability, which
makes it challenging for physicians to comprehend the logic behind the predictions despite
their remarkable success. This restriction has made it more difficult for AI to be widely
used in healthcare contexts. Researchers have resorted to explainable AI (XAI) methods
in order to tackle this problem. XAI techniques seek to increase the transparency of AI
models by offering insights into their decision-making processes.
