Understand how machine learning impact lung cancer research from 2010 to 2021: A bibliometric analysis

Abstract

Advances in lung cancer research applying machine learning (ML) technology have generated many relevant literature. However, there is absence of bibliometric analysis review that aids a comprehensive understanding of this field and its progress. Present article for the first time performed a bibliometric analysis to clarify research status and focus from 2010 to 2021. In the analysis, a total of 2,312 relevant literature were searched and retrieved from the Web of Science Core Collection database. We conducted a bibliometric analysis and further visualization. During that time, exponentially growing annual publication and our model have shown a flourishing research prospect. Annual citation reached the peak in 2017. Researchers from United States and China have produced most of the relevant literature and strongest partnership between them. Medical image analysis and Nature appeared to bring more attention to the public. The computer-aided diagnosis, precision medicine, and survival prediction were the focus of research, reflecting the development trend at that period. ML did make a big difference in lung cancer research in the past decade.

Keywords: bibliometric analysis; burstiness; collaboration; global trend; lung cancer; machine learning.

Figure 1

Flowchart of data collection and analysis.

Figure 2

Overview of relevant literature from 2010 to 2021. (a) Annual numbers of publication and citation times of literature. (b) Price’s curve was calculated to predict the trend, with satisfying coefficient of determination of 0.9903, indicating that the annual publication number will continue to grow. (c) Annual numbers of publication of HCP and top paper. (d) Type of total publication literature and HCP.

Figure 3

Output and collaboration of relevant literature at country and institution level. (a) Global geographical distribution of publications and collaboration between countries. Co-authorship analysis between institutions: (b) in network map based on publication number and (c) in density map based on their collaboration frequency.

Figure 4

Publication status and co-occurrence analysis of journal source. Publication of (a) total literature and (b) HCPs in journal source. (c) Co-occurrence analysis between journal source in network map.

Figure 5

Network visualization of category co-occurrence in total literature.

Figure 6

Network visualization of keyword co-occurrence in total literature and HCPs. Research field-based keyword cluster map of (a) total literature and (c) HCPs. Chronological keyword overview of (b) total literature and (d) HCPs.

Figure 7

Descriptive statistical analysis of ML-assisted survival prediction topic. The cooperative network map and visualization of (a) country, (b) institution, and (c) source. (d) Keyword co-occurrence analysis of relevant literature in this field.