Journal Publications in Medicine: Ranking vs. Interdisciplinarity

TL;DR

This study models interdisciplinarity in medical research as MeSH-based correlation networks and compares highly impactful versus less impactful journals across three time points. Using cosine-normalised co-occurrence networks, signed difference networks, and multiple topology metrics, it finds that high-impact journals tend to be less interdisciplinary on average, while cancer-related research acts as a major driver of interdisciplinarity. Differences between journal groups exhibit weak co-location of strong differences and form topic-clustered patterns that evolve over time, contrasting with a stable absolute core of knowledge in medicine. The work highlights policy and evaluation implications, advocating for frameworks that better recognize and reward interdisciplinarity to maximize patient benefits.

Abstract

Interdisciplinary research is critical for innovation and addressing complex societal issues. We characterise the interdisciplinary knowledge structure of PubMed research articles in medicine as correlation networks of medical concepts and compare the interdisciplinarity of articles between high-ranking (impactful) and less high-ranking (less impactful) medical journals. We found that impactful medical journals tend to publish research that are less interdisciplinary than less impactful journals. Observing that they bridge distant knowledge clusters in the networks, we find that cancer-related research can be seen as one of the main drivers of interdisciplinarity in medical science. Using signed difference networks, we also investigate the clustering of deviations between high and low impact journal correlation networks. We generally find a mild tendency for strong link differences to be adjacent. Furthermore, we find topic clusters of deviations that shift over time. In contrast, topic clusters in the original networks are static over time and can be seen as the core knowledge structure in medicine. Overall, journals and policymakers should encourage initiatives to accommodate interdisciplinarity within the existing infrastructures to maximise the potential patient benefits from IDR.

Figures (11)

• Figure 1: Distribution of impacts (SJR) of the Scimago journal ranking for the year 2022 in medicine. Red vertical line (SJR=1.46) represents the 10% cutoff that captures 42% of total impact of 7187 journals.
• Figure 2: Illustration of the Impactful network in 1999. Links with strength less than $0.080$ were filtered out for better visualisation. Communities were detected based on the Louvain algorithm blondelFastUnfoldingCommunities2008 and are labelled by MeSH codes. Label size represents the size of node strength. We find a very similar community breakdown for the non-impactful network in 1999 (not shown).
• Figure 3: Link Strength Distribution for (a) the I, NI, and NI-June networks, and (b) the I and NI-June networks. Logarithmic binning was applied on the x-axis with $30$ bins. For comparison, a power law distribution with exponent $\alpha=1.9$ and minimum value of link strength $x_{min}=10^{-1.5}$ is plotted as the dotted line against the tails of the distributions.
• Figure 4: Node strength distribution of the I and NI networks. An exponential distribution with decay rate $\lambda=0.6$ with node strength $s\in[1,6]$ is plotted as the dotted line against the distributions.
• Figure 5: (A) Dependence of the size of the largest connected component (LCC) on the cutoff threshold for the I and NI June for all three points in time. The blue line represents the empirical network and the red line represents the mean of 100 randomised networks (through link shuffling) with 95% confidence interval (red ribbon). Note that the observed cut-off of the red ribbon around $0.8$ in each network is due to the fact that the threshold reaches the largest link weight, beyond which there is no variation of the LCC size. (B) Inner core component membership (see text for a definition) for the I and NI June networks. There are cases where LCC of size 10 does not exist. In those cases LCC of size that are immediately smaller than 10 are shown, namely, size 9 for 1999I, 2010I, 2010NI-June, size 8 for 2022I, 2022NI-June, and size 10 for 1999 NI-June.