Annual publication counts are summed over a three-year rolling window to smooth out single-year indexing noise and reveal sustained growth trends. Spinal Cord Injuries is at or near its all-time peak (3-year sum ≈ 5,350) and continues rising, while Motor Neurons and Spinal Cord have both passed their peaks and are declining in recent years. Most strikingly, Electrophysiology has nearly vanished from the record (3-year sum < 10), and Genetic Therapy remains near its peak — together signaling a broad shift away from classical electrophysiological characterization and toward repair- and disease-focused research.

LAG() on publication counts within each term partition enables year-over-year percentage change without a self-join. Each panel is an independent subfield, covering 2012–2025. Green-shaded bars mark net-growth years; red-shaded bars mark declines; the dashed line marks 0 %. Optogenetics and Stem Cells show predominantly positive-growth years over the window, while Renshaw Cells and Central Pattern Generators show more mixed or declining trajectories. A brief negative spike visible across multiple panels in 2020–2021 reflects COVID-19-era publication disruptions.

A self-join on the authorships table (a1.author_id < a2.author_id) produces every unique co-author pair exactly once. Grouping by decade shows that distinct collaborating pairs grew roughly 51× — from ~70 k in the 1970s to ~3.6 M in the 2020s — reflecting both the expanding field and the rise of multi-site consortia. At the same time, the average number of joint papers per pair declined monotonically from 1.28 to 1.03. This likely reflects the changing nature of modern publications: today's papers tend to be larger, multi-year studies requiring broader author teams, whereas earlier work consisted of smaller single-experiment papers that could be turned around quickly — so the same set of collaborators produced more publications per unit of time than modern consortia do.

Relative Citation Ratio (RCR) is field- and year-normalized: a value of 1.0 equals the average NIH-funded paper in the same field and year cohort. RANK() OVER (PARTITION BY subfield ORDER BY rcr DESC) assigns ranks within each subfield independently, allowing fair cross-subfield comparison. Bubble area encodes raw citation count. The Demyelinating Diseases subfield contains by far the most influential paper in the dataset — the 2017 McDonald diagnostic criteria revision (The Lancet Neurology), with an RCR ≈ 299 and over 5700 citations — illustrating how a single clinical consensus paper can dominate a subfield. Inflammation papers also cluster at very high RCR (50–145), while Central Pattern Generators top out around RCR ≈ 22. Hover a point to see the title, journal, and exact RCR.

Trend direction is determined by comparing each term's normalized frequency (term count ÷ total publications that year) across a recent window (2021–2025) versus a baseline window (2010–2020). Normalizing by annual publication volume removes the confound of literature growth and isolates genuine shifts in research focus. The dominant emerging terms — Humans (+8.3 pp), Multiple Sclerosis (+5.4 pp), Brain (+5.2 pp), White Matter, and MRI — all point toward a pivot to clinical and neuroimaging research. The steepest declines are in demographic and animal-model tags: Male (−16.2 pp), Female (−12.6 pp), and Animals (−12.5 pp), along with terms associated with classical electrophysiology studies such as Synaptic Transmission and Rats. Together these shifts reflect a field-wide transition away from animal-based electrophysiology toward human clinical and imaging studies.

The full time series for the 20 most-published MeSH terms, expressed as a percentage of all annual publications in the database. The top 5 terms by recent share — Humans (~80 %), Male, Female, Animals, and Adult — are highlighted; the remaining 15 are shown in gray and can be toggled via the legend. Viewing all terms exposes striking competitive dynamics: Animals has fallen from a peak of ~54 % to ~32 %, while Synaptic Transmission dropped from ~25 % to ~4 % and Motor Neurons from ~16 % to ~5 %. Meanwhile, human clinical and imaging terms remain dominant or are slowly growing. The story is less about absolute publication volume and more about a decades-long reorientation of the field from animal electrophysiology toward human-centered disease and imaging research. Toggle individual terms by clicking the legend.

Taken together, these six analyses point to a consistent, decades-long transformation in spinal neuroscience. In absolute terms the field has never been larger — Spinal Cord Injuries output is at an all-time high and the co-authorship network has grown by more than 50× since the 1970s. But the character of the research has shifted fundamentally: animal-model and classical electrophysiology terms (Animals, Rats, Synaptic Transmission, Motor Neurons) have lost substantial share of the literature, while human, clinical, and imaging terms (Multiple Sclerosis, Brain, White Matter, MRI) are on the rise. The citation landscape reinforces this: the single most-cited paper in the dataset is a clinical diagnostic consensus document, not a bench-science discovery. Meanwhile, the growing breadth — but declining per-pair paper output — of the collaboration network mirrors the broader trend toward large, slow-burn multi-site studies over rapid single-experiment publications. The database and SQL framework built here make it straightforward to re-run or extend any of these analyses as the literature continues to evolve.

One important caveat: because the underlying data are drawn exclusively from PubMed, which is curated by the NIH and skews toward clinical and translational literature, the apparent shift toward human research may be at least partly an artifact of the data source rather than a pure reflection of the broader field. Animal-model and basic-science work published in journals with weaker PubMed indexing coverage could be systematically under-represented, inflating the relative share of clinical terms. A fuller picture would require cross-referencing against additional sources such as Scopus or Web of Science.