Neuroprotective effect of SGLT2 inhibitors.
Open accessThe DOI matches its Crossref record; 122 of the 124 references it cites were found in a registry or an index; 21 of the 97 sources its claims cite were checked for support.
Abstract
Patients with diabetes are at higher risk of cardiovascular diseases and cognitive impairment. SGLT2 inhibitors (Empagliflozin, Canagliflozin, Dapagliflozin, Ertugliflozin, Sotagliflozin) are newer hypoglycemic agents with many pleiotropic effects. In this review, we discuss their neuroprotective potential. SGLT2 inhibitors (SGLT2i) are lipid-soluble and reach the brain/serum ratio from 0.3 to 0.5. SGLT receptors are present in the central nervous system (CNS). Flozins are not fully SGLT2-selective and have an affinity for the SGLT1 receptor, which is associated with protection against ischemia/reperfusion brain damage. SGLT2i show an anti-inflammatory and anti-atherosclerotic effect, including reduction of proinflammatory cytokines, M2 macrophage polarization, JAK2/STAT1 and NLRP3 inflammasome inhibition, as well as cIMT regression. They also mitigate oxidative stress. SGLT2i improve endothelial function, prevent remodeling and exert a protective effect on the neurovascular unit, blood-brain barrier, pericytes, astrocytes, microglia, and oligodendrocytes. Flozins are also able to inhibit AChE, which contributes to cognitive improvement. Empagliflozin significantly increases the level of cerebral BDNF, which modulates neurotransmission and ensures growth, survival, and plasticity of neurons. Moreover, they may be able to restore the circadian rhythm of mTOR activation, which is quite a novel finding in the field of research on metabolic diseases and cognitive impairment. SGLT2i have a great potential to protect against atherosclerosis and cognitive impairment in patients with type 2 diabetes mellitus.
Overview
CoChat Summary
TL;DR
This narrative review summarizes proposed neuroprotective mechanisms of SGLT2 inhibitors in type 2 diabetes, atherosclerosis, and cognitive impairment, including anti-inflammatory, antioxidant, anti-atherosclerotic, AChE-inhibiting, BDNF-increasing, and mTOR-rhythm-restoring effects. It concludes that flozins have potential but that long-term clinical trials are needed to establish clinical relevance.
Key findings
- SGLT2 inhibitors are lipid-soluble and cross the blood-brain barrier, with brain/serum AUC ratios of 0.3 for canagliflozin and dapagliflozin and 0.5 for empagliflozin; SGLT1/SGLT2 are expressed in the CNS, and SGLT1 inhibition is linked to protection against ischemia/reperfusion brain damage.
- Anti-inflammatory and anti-atherosclerotic effects include reduced proinflammatory cytokines, M2 macrophage polarization, JAK2/STAT1 and NLRP3 inflammasome inhibition, and cIMT regression; empagliflozin reduced cIMT by 7.9% (p < 0.0001) after 3 months, and canagliflozin lowered IL-6 by 26.6% (p = 0.010) after 2 years.
- AChE inhibition may contribute to cognitive improvement: canagliflozin Ki = 0.12859 µM with estimated brain concentration ~3 µM, enough to inhibit AChE; empagliflozin Ki = 0.177 µM with brain concentration 0.5 µM, also enough; dapagliflozin Ki = 25.02 µM with brain concentration 0.3 µM, not enough.
- Empagliflozin significantly increased cerebral BDNF in db/db mice and improved cognition; in rat cerebral ischemia/reperfusion, empagliflozin dose-dependently reduced neuronal death, infarct size, and cognitive impairment via HIF-1α/VEGF signaling; SGLT2i may restore circadian mTOR activation and mimic fasting.
Methods
- Narrative review; no primary data collection or systematic search strategy is described.
- Evidence synthesized includes preclinical studies (murine/rat models, ex vivo, in silico) and selected clinical studies on SGLT2 inhibitors, atherosclerosis, inflammation, oxidative stress, mTOR signaling, cerebrovascular dysfunction, and Alzheimer’s disease pathology.
- Main analysis is a qualitative synthesis of mechanisms and reported outcomes, with one ongoing clinical trial noted: dapagliflozin in Alzheimer’s disease (NCT03801642).
Limitations
- The paper states that long-term clinical trials are necessary to establish whether the discussed mechanisms are clinically relevant, since atheroprotective and neuroprotective effects will not be immediate and require long-term SGLT2i intake.
- The paper notes that there is no available data on adverse CNS effects of flozins and that the exact mechanism of cardiovascular benefit has not been fully established.
- Evident from the text: this is a narrative review without systematic methods, and many key findings come from animal, ex vivo, or in silico studies, with limited human cognitive-outcome data.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures




Verification
Claims the paper makes
14 Valid · 1 Not valid · 67 Not checked · 11 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Canagliflozin, Dapagliflozin, Sotagliflozin significantly decrease the composed primary end-point, including cardiovascular mortality and other cardiovascular outcomes [10,11,12].”
§ Neuroprotective Effect of SGLT2 Inhibitors > 1. Introduction
What the sources say
- Supportedcites [10] Mahaffey, 2018, Canagliflozin for Primary and Secondary Prevention of Cardiovascular Events: Results from the CANVAS Program (Canagliflozin Cardiovascular Assessment Study), Circulation source
The Methods (Outcomes) define the composite endpoint as cardiovascular mortality, nonfatal myocardial infarction, or nonfatal stroke. The Results report that canagliflozin significantly reduced that endpoint versus placebo (P=0.02 for superiority). In this sentence with three citations, this source supports the canagliflozin portion; it is not offered as evidence for the other two drugs. (LLM, conf=0.93)
- Partly supportedcites [11] Wiviott, 2019, Dapagliflozin and Cardiovascular Outcomes in Type 2 Diabetes, N. Engl. J. Med. sourceExperimental — accuracy not yet measured
The Cardiovascular and Renal Outcomes section reports that dapagliflozin significantly reduced the composite of cardiovascular death or hospitalization for heart failure. It also says the reduction was driven by fewer heart-failure hospitalizations, with no difference in cardiovascular death; MACE was not significantly reduced. This paper tests dapagliflozin, not the other two drugs named alongside it. Page numbers were not supplied in the passages. (LLM, conf=0.89)
- Supportedcites [12] Bhatt, 2021, Sotagliflozin in Patients with Diabetes and Recent Worsening Heart Failure, N. Engl. J. Med. source
The Methods (p. 2) define the sotagliflozin trial’s composite primary endpoint as cardiovascular death, heart-failure hospitalization, and urgent heart-failure visits. The Efficacy End Points (p. 7) report a significantly lower rate for that composite with sotagliflozin than with placebo (hazard ratio 0.67; P<0.001). This source supports the sotagliflozin portion of the multi-citation sentence; it does not establish a significant reduction in cardiovascular death on its own. (LLM, conf=0.92)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“SGLT1 inhibitors are present in the pyramidal cells of the brain cortex, Purkinje cerebellum cells, hippocampus pyramidal, and granular cells [20].”
§ Neuroprotective Effect of SGLT2 Inhibitors > 2. Neurological Potential of SGLT2 Inhibitors
What the sources say
- Partly supportedcites [20] Poppe, 1997, Expression of the Na+-D-glucose cotransporter SGLT1 in neurons, J. Neurochem. sourceExperimental — accuracy not yet measured
The 'Expression of SGLT1 protein in brain' section and Fig. 7 report SGLT1 protein in cortical pyramidal cells, hippocampal pyramidal and granular cells, and cerebellar Purkinje cells. The citing sentence says 'SGLT1 inhibitors' are present; these passages identify the protein, not inhibitors. (LLM, conf=0.88)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Among commonly used SGLT2 inhibitors, Canagliflozin has the greatest potential for inhibiting SGLT1 receptors.”
§ Neuroprotective Effect of SGLT2 Inhibitors > 2. Neurological Potential of SGLT2 Inhibitors
What the sources say
- Could not verifycites [27] Malhotra, 2015, Sodium glucose co-transporter inhibitors—A new class of old drugs, Int. J. Appl. Basic Med. Res. sourceExperimental — accuracy not yet measured
Our copy names canagliflozin only as 'the prototype drug among the newer agents' with its pharmacokinetics (PHARMACOKINETICS section), and it discusses SGLT-1/SGLT-2 selectivity in general terms (INTRODUCTION, HISTORICAL BACKGROUND, 'On fasting hypoglycemia'), but nowhere states that canagliflozin has the greatest SGLT-1 inhibitory potential among the SGLT2 inhibitors. The one place that could carry it — 'Table 1 New SGLT inhibitors and their status and selectivity' — appears in our copy only as a caption, with no table rows/indexed content, so the comparative selectivity data are not available to us. Searches for 'canagliflozin SGLT1 selectivity', 'selectivity ratio SGLT-1 versus SGLT-2 / IC50', 'Table 1 status and selectivity', and 'SGLT-1 inhibition adverse effects' returned no passage making or contradicting the claim. (LLM, conf=0.65)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“In humans, the serum level of IL-6 dropped by 26.6% (p = 0.010) after 2 years of canagliflozin treatment [35].”
§ Neuroprotective Effect of SGLT2 Inhibitors > 4. Inflammation
What the sources say
- Partly supportedcites [35] Heerspink, 2019, Canagliflozin reduces inflammation and fibrosis biomarkers: A potential mechanism of action for beneficial effects of SGLT2 inhibitors in diabetic kidney disease, Diabetologia sourceExperimental — accuracy not yet measured
The Abstract (Results) and Table 2 of our copy do carry the exact figure: canagliflozin 300 mg/day lowered IL-6 by 26.6% (95% CI −42.0, −7.2), p = 0.010, over 2 years (104 weeks) — but explicitly as a least-squares mean percentage DIFFERENCE VERSUS GLIMEPIRIDE and only at the 300 mg dose (100 mg gave −26.3%, p = 0.011), in people with type 2 diabetes and elevated albuminuria (CANTATA-D/SU subset, n = 296). The citing sentence recasts this between-group contrast as an absolute 'drop' of 26.6% from canagliflozin treatment alone, and drops both the comparator and the dose qualifiers. It also says 'serum', whereas the Methods (Biomarker measurement) state IL-6 was measured by Luminex assay on stored PLASMA samples. The magnitude, p value and 2-year duration are faithful; the attribution of the 26.6% as a within-treatment fall in serum is not supported. (LLM, conf=0.72)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“In previous studies, atherosclerotic lesions in intra- and extra-cranial arteries were associated with cognitive impairment and even dementia [2,66,67].”
§ Neuroprotective Effect of SGLT2 Inhibitors > 3. Atherosclerosis, Cognitive Impairment, and SGLT2i
What the sources say
- Not checked yetcites [2] Dearborn, 2017, Intracranial atherosclerosis and dementia The Atherosclerosis Risk in Communities (ARIC) Study, Am. Acad. Neurol.
- Supportedcites [66] Bos, 2015, Atherosclerotic calcification is related to a higher risk of dementia and cognitive decline, Alzheimer’s Dement. source
The Results section reports that after censoring for stroke, 'both extracranial and intracranial carotid artery calcification remained statistically significantly associated with dementia' (HR 1.32 and 1.34 per SD increase), and that larger calcification volumes were associated with cognitive decline in nondemented persons. The Abstract Conclusions and Discussion (generalized atherosclerosis 'is associated with dementia') state the same association for dementia and cognitive deterioration, matching the cited claim for both intra- and extra-cranial arteries. Minor wording differences (calcification as the atherosclerosis measure; 'cognitive decline' rather than 'cognitive impairment') do not alter the attributed content. (LLM, conf=0.88)
- Partly supportedcites [67] Wingo, 2020, Shared proteomic effects of cerebral atherosclerosis and Alzheimer’s disease on the human brain, Nat. Neurosci. sourceExperimental — accuracy not yet measured
The Abstract/Introduction of our copy supports the intracranial half of the claim: cerebral atherosclerosis is defined as plaques in 'the walls of large arteries in the brain' and is stated to be among the earliest changes associated with development of AD dementia, with ~three times higher risk for AD and nine times higher risk for vascular dementia (Abstract, p. 1). The METHODS confirm the paper's own atherosclerosis assessment covered only intracranial vessels — visual grading of the Circle of Willis (vertebral, basilar, posterior, middle and anterior cerebral arteries and proximal branches), not extracranial arteries (METHODS, ROS/MAP section). No passage in our copy addresses atherosclerotic lesions in extra-cranial arteries and cognition (searched: 'extracranial carotid atherosclerosis', 'carotid artery atherosclerosis systemic extracranial vessels'), so that part of the citing sentence is not supported by this source. (LLM, conf=0.72)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
42 more claims · 2 with their cited sources checked
Create a free account to see all 93 claims and to verify the 76 references still open.
- It may state two separate claims as one. 41
- Probably not a substantial claim. 6
- The quote only partly states this claim. 1
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Similarly, SGLT2is have demonstrated neuroinflammatory and antioxidative effects in addition to cardiovascular prevention.56”
From Cardioprotective Glucose-Lowering Agents and Dementia Risk: A Systematic Review and Meta-Analysis (2025) · cites this paper as [56]
Citations
Cites
1 retracted
Xue, L., Yuan, X., Zhang, S., & Zhao, X. (2021) · Evidence-based Complementary and Alternative Medicine · doi:10.1155/2021/9388562 · Cited by 20
RetractedEmpagliflozin, Cardiovascular Outcomes, and Mortality in Type 2 Diabetes
Zinman, B., Wanner, C., Lachin, J. M., Fitchett, D., Bluhmki, E., Hantel, S., Mattheus, M., Devins, T., Johansen, O. E., Woerle, H. J., et al. (2015) · New England Journal of Medicine · doi:10.1056/nejmoa1504720 · Cited by 12,315
Antiinflammatory Therapy with Canakinumab for Atherosclerotic Disease
Ridker, P. M., Everett, B. M., Thuren, T., MacFadyen, J. G., Chang, W. H., Ballantyne, C., Fonseca, F., Nicolau, J., Koenig, W., Anker, S. D., et al. (2017) · New England Journal of Medicine · doi:10.1056/nejmoa1707914 · Cited by 9,482 · cited 1×
“In the CANTOS study, inhibition of interleukin-1β with canakinumab significantly reduced cardiovascular risk independently of lipid levels [78].”
BackgroundDapagliflozin and Cardiovascular Outcomes in Type 2 Diabetes
Wiviott, S. D., Raz, I., Bonaca, M. P., Mosenzon, O., Kato, E. T., Cahn, A., Silverman, M. G., Zelniker, T. A., Kuder, J. F., Murphy, S. A., et al. (2019) · New England Journal of Medicine · doi:10.1056/nejmoa1812389 · Cited by 6,382
Dapagliflozin in Patients with Chronic Kidney Disease
Heerspink, H. J. L., Stefánsson, B. V., Correa-Rotter, R., Chertow, G. M., Greene, T., Hou, F.-F., Mann, J. F. E., McMurray, J. J. V., Lindberg, M., Rossing, P., et al. (2020) · New England Journal of Medicine · doi:10.1056/nejmoa2024816 · Cited by 5,618 · cited 1×
“The exact mechanism has not been fully established yet, and SGLT2 inhibitors show many additional beneficial effects which contribute to their wider use, even in non-diabetic patients [14].”
BackgroundReactive Oxygen Species in Inflammation and Tissue Injury
Mittal, M., Siddiqui, M. R., Tran, K., Reddy, S. P., & Malik, A. B. (2014) · Antioxidants & Redox Signaling · doi:10.1089/ars.2012.5149 · Cited by 4,897 · cited 4×
“The overproduction of ROS or decrease in the anti-oxidant defense results in oxidative stress, which is a significant contributor to vascular diseases, including atherosclerosis, as it causes endothelial dysfunction, promotes remodeling and further enhances inflammation [94].”
BackgroundInfluentialO'Leary, D. H., Polak, J. F., Kronmal, R. A., Manolio, T. A., Burke, G. L., & Wolfson, S. K. (1999) · New England Journal of Medicine · doi:10.1056/nejm199901073400103 · Cited by 4,685
The NLRP3 Inflammasome: an overview of mechanisms of activation and regulation
Kelley, N., Jeltema, D., Duan, Y., & He, Y. (2019) · International Journal of Molecular Sciences · doi:10.3390/ijms20133328 · Cited by 3,539 · cited 1×
“It is a crucial element of the innate immune system activated not only by microbial infection or cellular damage but also by chronic inflammatory diseases, including atherosclerosis and Alzheimer’s disease [80].”
BackgroundEinarson, T. R., Acs, A., Ludwig, C., & Panton, U. H. (2018) · Cardiovascular Diabetology · doi:10.1186/s12933-018-0728-6 · Cited by 2,312 · cited 1×
“Type 2 diabetes mellitus (T2DM) is a chronic metabolic disease causing a variety of complications, including atherosclerosis which is associated with increased cardiovascular risk contributing to reduced life expectancy [1].”
BackgroundSotagliflozin in patients with diabetes and recent worsening heart failure
Bhatt, D. L., Szarek, M., Steg, P. G., Cannon, C. P., Leiter, L. A., McGuire, D. K., Lewis, J. B., Riddle, M. C., Voors, A. A., Metra, M., et al. (2021) · New England Journal of Medicine · doi:10.1056/nejmoa2030183 · Cited by 1,868 · cited 1×
“It is even called a “dual SGLT1/SGLT2 inhibitor”, however, it is the newest Flozin, and it is not yet used in diabetic patients on a large scale [12].”
Background
Research works printed but not linked: 1
- Esterline, R., Oscarsson, J., and Burns, J. (2020). International Review of Neurobiology, Elsevier Inc.. [1st ed.].
Cited by
Wang, W., Wang, Q., Qi, X., Gurney, M., Perry, G., Volkow, N. D., Davis, P. B., Kaelber, D. C., & Xu, R. (2024) · Alzheimer's & Dementia · doi:10.1002/alz.14313 · Cited by 180
Mone, P., Lombardi, A., Gambardella, J., Pansini, A., Macina, G., Morgante, M., Frullone, S., & Santulli, G. (2022) · Diabetes Care · doi:10.2337/dc21-2434 · Cited by 151
Inflammatory Mechanisms of Diabetes and Its Vascular Complications
Nedosugova, L. V., Markina, Y. V., Бочкарева, Л. А., Кузина, И. А., Petunina, N. A., Yudina, I. Y., & Kirichenko, T. V. (2022) · Biomedicines · doi:10.3390/biomedicines10051168 · Cited by 157
GLP-1RA and SGLT2i Medications for Type 2 Diabetes and Alzheimer Disease and Related Dementias
Tang, H., Donahoo, W. T., DeKosky, S. T., Lee, Y. A., Kotecha, P., Svensson, M., Bian, J., & Guo, J. S. (2025) · JAMA Neurology · doi:10.1001/jamaneurol.2025.0353 · Cited by 116
Mei, J., Li, Y., Niu, L., Liang, R., Tang, M., Cai, Q., Xu, J., Zhang, D., Yin, X., Liu, X., et al. (2024) · Translational Neurodegeneration · doi:10.1186/s40035-024-00431-y · Cited by 98
“Neuroinflammation is closely related to cognitive deficits [5].”
Background
Metadata
- Authors
- Pawlos, A., Broncel, M., Woźniak, E., & Gorzelak-Pabiś, P.
- Affiliations
- Medical University of Lodz (PL)
- Laboratory of Tissue Immunopharmacology, Department of Internal Diseases and Clinical Pharmacology, Medical University of Lodz, Kniaziewicza 1/5, 91-347 Lodz, Poland
- Corresponding author
- Broncel, Marlena
- Venue
- Molecules
- Publisher
- MDPI AG
- Published
- 2021-11-28
- Volume
- 26
- Issue
- 23
- Pages
- 7213
- ISSN
- 1420-3049 (ISSN-L)
- PMC
- PMC8659196
- OpenAlex
- W3214770882
- Semantic Scholar
- d4fddd52e53127993c142ff6d31526f815eb1c00
- Cited by
- 304
- References
- 124
- Citations by year
- 2026: 66 · 2025: 92 · 2024: 71 · 2023: 50 · 2022: 25
- FWCI
- 4.26
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Molecules (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Topics
- Life Sciences › Neuroscience › Neurology › Neuroprotection in Cerebral Ischemia
- Health Sciences › Medicine › Endocrinology, Diabetes and Metabolism › Diabetes Treatment and Management
- Life Sciences › Neuroscience › Neurology › Neurological Disease Mechanisms and Treatments
- Keywords
- SGLT2 inhibitors, neuroprotection, type 2 diabetes, cognitive impairment, cardiovascular disease, anti-inflammatory activity, NLRP3 inflammasome, cerebral blood flow, neurovascular unit, blood-brain barrier, acetylcholinesterase inhibition, brain-derived neurotrophic factor, atherosclerosis
- MeSH terms
- Sodium-Glucose Transporter 2 Inhibitors, Animals, Brain, Diabetes Mellitus, Type 2, Humans, Hypoglycemic Agents, Neuroprotective Agents, Atherosclerosis, Sodium-Glucose Transporter 2, Cognitive Dysfunction
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-04, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-04 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Neuroprotective effect of SGLT2 inhibitors. https://app.cochat.ai/papers/10.3390/molecules26237213
- The paper
- Pawlos, A., Broncel, M., Woźniak, E., & Gorzelak-Pabiś, P. (2021). Neuroprotective effect of SGLT2 inhibitors. Molecules. https://doi.org/10.3390/molecules26237213
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.