Among the most extensive global databases of scientific literature, Scopus, we looked for relevant articles. Publications such as peer-reviewed journals, conference proceedings, commercial magazines, and book series are among the over 5,000 titles currently available on Scopus. Journal articles, conference proceedings, patents, and other publications covering any and all fields make up its approximately 75 million records. To find relevant documents, we used the following keywords: “artificial intelligence” OR “machine intelligence” OR “robot*” OR “robot technology” OR “assistant robot” OR “robot-assisted” OR “computational intelligence” OR “computer reasoning” OR “deep learning” OR “computer vision system” OR “sentiment analysis” OR “machine learning” OR “neural network*” OR “data learning” OR “expert* system*” OR “natural language processing” OR “support vector machine*” OR “decision tree*” OR “data mining” OR “deep learning” OR “neural network*” OR “bayesian network*” OR “intelligent learning” OR “feature* learning” OR “feature* extraction” OR “feature* mining” OR “feature* selection” OR “unsupervised clustering” OR “image* segmentation” OR “supervised learning” OR “semantic segmentation” OR “deep network*” OR “neural learning” OR “neural nets model” OR “graph mining” OR “data clustering” OR “big data” OR “knowledge graph”)) AND ALL ((solitude OR aloneness OR lonesomeness)) AND ALL ((“senior∗” OR “aged∗” OR “elder∗” OR “geri∗” OR “age∗” OR “older adult∗” OR “geriatric∗” OR “aging” OR “very elderly” OR “frail elderly”. After a thorough examination of every title of the retrieved articles, two experts who possess a minimum of five years of experience in conducting systematic reviews evaluated and gave their approval to our search strategy. They made their selection from among all of the publications that were pertinent, and any disagreements were settled by consulting with a third author. Finally, all of the retrieved articles were saved.

Only articles that utilized machine learning as a result and focused on generative learning were incorporated into the bibliometric analysis. No language restrictions were imposed. Letters, editorials, conference abstracts, and book chapters were excluded from consideration. In contrast, research papers and reviews published in peer-reviewed journals were included in the selection process. Due to the focus on the ten-year trend, publications issued before 2014 were excluded from consideration. Two experts in bibliometric analysis, possessing extensive experience, assessed the acquired articles in accordance with established standards and guidelines. ¹³

To produce consistent and measurable data pertinent to policy management, bibliometric analysis ¹⁴ serves as a quantitative assessment of scientific research, evaluating and examining prevailing research trends within a specific domain. ¹⁵ Bibliometric analysis offers a comprehensive overview of a research subject and identifies potential areas for further investigation, along with the methodologies that authors have employed to achieve their objectives. ¹⁶ Displaying the complete field will facilitate readers’ comprehension of GL research patterns and trends.

The publication patterns of the leading ten publications, institutes, authors, and countries were analyzed. The categories, percentages, and frequencies for each journal were presented. Ratios, frequencies, and percentages of related authors were computed for each country, considering both single and multiple countries. The frequency and citation rates for each author and institution were provided as well. Countries, journals, institutions, and authors were ranked based on their publication volume and research impact. Data were constructed and visualized utilizing VOSviewer (Leiden University) and Bibliometrix (an R package).

A total of 267 items were identified through the electronic database search. After screening, 112 individuals were excluded. The comprehensive bibliometric analysis comprised 155 publications. Figure 1 illustrates the growth in the number of GL articles over the past decade.

Table 1 presents the geographical distribution of the ten leading countries in GL research publications. The United States produced the highest number of publications (18), followed by China (14) and Canada (9). China exhibited the highest multiple country publication (MCP) ratio at 0.5, followed by Canada at 0.444 and the United Kingdom at 0.429. Figure 2 illustrates the countries that are linked through at least one research collaboration.

Table 2 displays the ten leading institutes in GL research productivity. The University of California and the University of Pennsylvania each published seven articles, leading all universities, while the University of Copenhagen and the University of Zurich each published six articles.

In order to conduct a co-authorship study and offer a thorough perspective of the entirety of the nation network in GL research, the VOSviewer tool was employed. The relationship between two nations is determined by the total number of publications that they have jointly authored. The study of authorship by nation produced a single cluster, which consisted of fifteen authors who are related to one another in that they have each published at least one paper. These categories are shown in Figure 3.

When it comes to the subject of GL research, 267 publications were published by 640 different authors. The most prominent ten GL research authors are enumerated in Table 3. Three papers were published by Rigaud AS, and three papers were published by Pino M, who were the leading authors.

The 267 articles consisted of five distinct keyword clusters. Table 4 presents the ten most frequently used keywords in the retrieved articles. Human, ageing, male, female, and loneliness were significant categories identified. The co-occurrence analysis of the top 100 phrases revealed the research hotspots in GL ( Figure 4).

We conducted an in-depth examination of the knowledge base within the GL research domain. Table 5 presents the ten most highly cited articles. As of June 1, 2024 Arbabshirani, Plis, ¹⁷ had the highest citation count, totalling 602. Published in NeuroImage in 2017. Figure 5 illustrates the three-field plot.