Machine Learning Applications For Lung Cancer - An Overview
Machine learning applications for lung cancer have served as supplements for cancer intervention response evaluation and prediction, demonstrating gains in optimizing therapeutic decisions that increase the likelihood of effective recovery. Prognosis and survival prediction in clinical oncology is difficult but necessary work for clinicians since knowing the survival time may affect treatment decisions and assist patients in cost management.
Predictions for most medical history rely mainly on the physician's expertise and experience based on earlier patient accounts and medical records. However, studies have shown that physicians perform poorly in forecasting prognosis and survival expectancy, frequently over-predicting survival time. In comparison, machine learning has demonstrated its ability to predict a patient's prognosis and survival in genomic, transcriptomic, proteomic, radiomic, and other data sets.
Machine learning is a subfield of artificial intelligence (AI) and computer science that focuses on using data and algorithms to mimic how people learn, progressively improving accuracy. Machine learning algorithms have been utilised to integrate complex biological data (such as imaging or sequencing data) for many elements of lung cancer, and there is also potential for future applications in lung cancer clinical research and practice.
Lung cancer is among the most common malignancies and the leading cause of cancer deaths globally; around 2.20 million new patients are diagnosed with lung cancer each year, with 75% dying within five years of diagnosis. Current research onomics analysis, such as genomes, transcriptomics, proteomics, and metabolomics, has extended our research tools and capacities. Cancer research is shifting more towards integrating diverse data sources and massive quantities. However, using diversified and high-dimensional data types for clinical tasks requires significant time and expertise, even with dimension reduction methods such as matrix and tensor factorizations. Analyzing the exponentially increasing cancer-associated databases poses a considerable challenge to researchers.
As a result, utilizing machine learning models to automatically understand the inherent properties of various data sources to aid physicians' decision-making has grown in importance. Machine learning is an artificial intelligence subfield that focuses on creating predictions by recognizing patterns in data using mathematical algorithms. For decades, it has been used to aid cancer phenotyping and therapy. It is widely used in advanced methodologies for early detection, cancer type classification, signature extraction, tumor microenvironment deconvolution, prognosis prediction, and drug response evaluation.
A lot of published research also appears to be lacking in adequate validation or testing. Machine learning approaches may significantly (15-25%) enhance the accuracy of predicting cancer susceptibility, recurrence, and death is better planned and verified research.
Early detection is a critical step in lowering lung cancer-related fatalities. The primary monitoring method for those at high risk of lung cancer is chest screening using low-dose computed tomography. The computer-aided diagnosis system was created to aid clinicians in interpreting medical imaging data to improve diagnostic efficiency. In addition to added tomography images, convolutional neural network-based models are commonly utilized in histology imaging to aid in detecting lung cancer. As opposed to computed tomography imaging, histological imaging can reveal more biological information regarding cancer at the cellular level.
Machine learning applications for lung cancer have expanded in early detection, diagnostic decision making, prognosis prediction, medication response evaluation, and immunotherapy practice. Despite the extensive application of machine learning studies in lung cancer clinical practice and research, problems remain. The following are essential difficulties and future research directions.
- In contrast to prior techniques, characteristics from a convolutional neural networks model are not developed by humans and accurately reflect the fundamental qualities of the nodule. The Vision Transformer (ViT) has recently emerged as the current state of the art in computer vision. ViT beat convolutional neural networks by nearly four regarding computing efficiency and accuracy, and it was more resilient when trained on fewer datasets.
- Sophisticated Machine learning models have served as supplements for cancer intervention response evaluation and prediction. They have shown advancements in optimising therapeutic decisions that increase the likelihood of full recovery [95, 96]. Numerous metrics for evaluating cancer therapeutic response are available, including the response evaluation criteria in solid tumours.
- Immunotherapy has grown in popularity in recent years. It helps a patient's immune system fight cancer by activating T cells in most situations. Various innovative immunotherapy therapies are now being explored for lung cancer, and a few have become regular immunotherapy components. Immune checkpoint inhibitors, particularly programmed cell death protein 1 (PD1)/programmed cell death protein ligand 1 (PDL1) blocking therapy , are effective in the treatment of non-small cell lung cancer patients.
As a result, predicting whether a patient will respond to immunotherapy is critical in cancer treatment. Recently, artificial intelligence-based systems have been created to predict immunotherapy responses based on immune genetic characteristics and medical imaging signatures.