Understanding the Research Landscape of Deep Learning in Biomedical Science: Scientometric Analysis.

Area of Science:

• Biomedical research
• Artificial intelligence
• Data science

Background:

• Deep learning (DL) is advancing biomedical research, creating extensive literature.
• A lack of scientometric studies hinders a comprehensive understanding of DL's progress in biomedicine.
• Existing knowledge is fragmented, necessitating a holistic overview.

Purpose of the Study:

• To quantitatively and qualitatively analyze the DL in biomedicine research landscape.
• To provide a bird's-eye view of the field using diverse bibliographic data.
• To understand the progress and structure of DL applications in life sciences.

Main Methods:

• Scientometric analysis of 978 DL studies retrieved from PubMed.
• Analysis of metadata, influential works, and cited references.
• Examination of research topics, techniques, knowledge diffusion, and collaborations.

Main Results:

• Predominant application of DL, especially convolutional neural networks, in radiology and medical imaging.
• Limited focus on protein or genome analysis.
• Radiology, medical imaging, computer science, and electrical engineering are key knowledge sources and diffusion hubs.
• Coauthorship analysis revealed collaborations between engineering and biomedicine disciplines.

Conclusions:

• Deep learning research in biomedicine is inherently interdisciplinary.
• While successful, diverse applications are needed to enhance DL's impact on biomedical challenges.
• Findings aim to guide researchers in aligning current and future work.