Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 68 of the 70 references it cites were found in a registry or an index; none of the 80 sources its claims cite is checked for support yet.

Altered brown fat thermoregulation and enhanced cold-induced thermogenesis in young, healthy, winter-swimming men

Open access

Abstract

The Scandinavian winter-swimming culture combines brief dips in cold water with hot sauna sessions, with conceivable effects on body temperature. We study thermogenic brown adipose tissue (BAT) in experienced winter-swimming men performing this activity 2-3 times per week. Our data suggest a lower thermal comfort state in the winter swimmers compared with controls, with a lower core temperature and absence of BAT activity. In response to cold, we observe greater increases in cold-induced thermogenesis and supraclavicular skin temperature in the winter swimmers, whereas BAT glucose uptake and muscle activity increase similarly to those of the controls. All subjects demonstrate nocturnal reduction in supraclavicular skin temperature, whereas a distinct peak occurs at 4:30-5:30 a.m. in the winter swimmers. Our data leverage understanding of BAT in adult human thermoregulation, suggest both heat and cold acclimation in winter swimmers, and propose winter swimming as a potential strategy for increasing energy expenditure.

Overview

CoChat Summary

TL;DR

In young, healthy men, experienced winter swimmers had lower core temperature and no BAT glucose uptake at thermal comfort, but greater cold-induced thermogenesis and higher supraclavicular skin temperature during cooling than matched controls. The findings suggest both cold and heat acclimation in winter swimmers and raise winter swimming as a potential strategy to increase energy expenditure.

Key findings

  • At thermal comfort, most controls showed modest BAT FDG uptake, whereas winter swimmers had no uptake in any subject; during cooling, BAT SUVmean increased comparably in both groups.
  • Resting energy expenditure during 30 min cooling was higher in winter swimmers (3,044 ± 337.2 kcal/24 h) than controls (2,560 ± 348.1 kcal/24 h; p = 0.01), with no group difference at thermal comfort (2,038 ± 96.0 vs 2,005 ± 209.6 kcal/24 h; p = 0.69). Cold-induced thermogenesis was substantially higher in winter swimmers; the text reports energy expenditure increased by 500–1,000 kcal/24 h, versus a previous estimate of ~20 kcal/24 h for human BAT.
  • Winter swimmers had lower core temperature over 24 h and higher supraclavicular skin temperature; all subjects showed nocturnal reduction in supraclavicular skin temperature, but winter swimmers had a distinct peak at 4:30–5:30 a.m.
  • Controls had higher total fat percentage (18.2 ± 4.3% vs 12.0 ± 4.6%, p = 0.01) and gynoid fat (23.3 ± 5.4% vs 16.6 ± 6.1%, p = 0.03); winter swimmers had lower plasma glucose at 120 min after glucose ingestion and lower thyroxine, alkaline phosphatase, and transferrin, with higher potassium.

Methods

  • Randomized-order crossover design: a cooling test-day with PET/MRI, a thermal comfort test-day with PET/MRI, and a second cooling day 3–7 days later for resting energy expenditure and abdominal subcutaneous white adipose tissue biopsies; conducted in cold months in Denmark (March–May 2017 and October 2017–February 2018; diurnal study October 2017–January 2018).
  • Participants were young healthy men aged 18–35 years with BMI 19–25 kg/m²: winter swimmers (n = 8; ≥2–3 cold-water sessions/week and at least second winter-swimming season) and controls (n = 8) matched for age, BMI, and VO2max; one BAT-negative winter swimmer was excluded, leaving n = 7 winter swimmers and n = 8 controls for main analyses; diurnal subset was n = 5 winter swimmers and n = 6 controls.
  • Main measures included FDG-PET/MRI for BAT glucose uptake (primary outcome), infrared thermography and iButtons for skin temperature, indirect calorimetry for resting energy expenditure, EMG for muscle activity, DEXA, oral glucose tolerance test, cold pressor test, blood biomarkers, and adipose tissue qPCR; analyses used unpaired t tests, repeated-measures two-way ANOVA or mixed models, and Sidak post-tests, with p < 0.05 considered significant.

Limitations

  • The paper states that causality cannot be concluded: winter swimming per se may not cause the observed differences, because people with a certain thermophenotype could be more prone to winter swimming; groups also differed in tissue fat percentage and 120-min plasma glucose, and unmeasured lifestyle habits could differ.
  • The paper states low sample size risks type 1 and type 2 errors; four BAT-activity methods have limitations (PET measures glucose tracer uptake, not thermogenesis; MRI water percentage is indirect; IRT and iButtons are confounded by skin perfusion/insulation and require precise positioning); PET threshold 2.0 g/mL instead of recommended 1.2 g/mL may underestimate BAT; and PET/MRI overlay was not possible due to breathing artifacts.
  • The paper states the diurnal study lacked sleep measurements, had short familiarization and similar eating/sleep schedules, non-constant room temperature, and no chronotype determination; only men were studied; VAS-based cooling/thermal comfort could be suboptimal given possible adaptation, though no group differences in thermal perception or water temperature were observed. Text-evident: main analyses exclude one BAT-negative winter swimmer, and the diurnal subset is small (n = 5 winter swimmers, n = 6 controls).

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1016/j.xcrm.2021.100408

Cites

68 of 68 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Graphical abstract
Graphical abstract CC BY-NC-ND · PubMed Central
Figure 1 Subject characterization and experimental design Subjects (winter swimmers [WS], n = 7; controls [C], n = 8) performed an oral glucose tolerance test (OGTT). n represents the number of human individuals in each group and is consistent throughout this figure unless otherwise stated. (A) Plasma glucose levels before (time point 0) and following glucose ingestion. (B) Plasma glucose levels 2 h after glucose ingestion. (C) Plasma insulin levels before (time point 0) and following glucose ingestion. (D) Pulse before and during cooling and after reheating. (E) Systolic blood pressure in response to cooling during a cold pressor test. (F) Diastolic blood pressure in response to cooling. (G) Experimental design depicting the measurements performed during the three experimental days. Data are presented as mean ± SD. Differences were assessed using two-way repeated-measures ANOVA with Sidak’s multiple comparison tests to assess specific differences except in (B), where an unpaired t test was used. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.
Figure 1 Subject characterization and experimental design Subjects (winter swimmers [WS], n = 7; controls [C], n = 8) performed an oral glucose tolerance test (OGTT). n represents the number of human individuals in each group and is consistent throughout this figure unless otherwise stated. (A) Plasma glucose levels before (time point 0) and following glucose ingestion. (B) Plasma glucose levels 2 h after glucose ingestion. (C) Plasma insulin levels before (time point 0) and following glucose ingestion. (D) Pulse before and during cooling and after reheating. (E) Systolic blood pressure in response to cooling during a cold pressor test. (F) Diastolic blood pressure in response to cooling. (G) Experimental design depicting the measurements performed during the three experimental days. Data are presented as mean ± SD. Differences were assessed using two-way repeated-measures ANOVA with Sidak’s multiple comparison tests to assess specific differences except in (B), where an unpaired t test was used. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.CC BY-NC-ND · PubMed Central
Figure 2 Infrared thermography of supraclavicular BAT heat production at cooling and a thermal comfort (TC) state Subjects (WS, n = 7; C, n = 8) were cooled using a Blanketrol III system by controlling the temperature of circulating water in the blanket. n represents the number of human individuals in each group and is consistent throughout this figure unless otherwise specifically stated. (A) Water temperature in the blanket during cooling. (B) Estimate of the subjects’ temperature perception using a VAS score. (C) Core temperature as estimated by rectal thermometers. On a separate day, subjects were observed during the thermal comfort state protocol. (D) Water temperature during a thermal comfort state. (E) Estimate of the subjects’ temperature perception using a VAS score. (F) Core temperature as estimated by rectal thermometers. (G) Supraclavicular BAT skin temperature during cooling measured by IRT. The effect of the group and cooling were assessed during time points of 45–105 min. Sidak’s post-test was used to assess differences between the time point of 45 min (thermal comfort state) and the time points during cooling. (H) Sternum skin temperature during cooling as measured by IRT and analyzed as described in (G). (I) Supraclavicular BAT skin temperature in relation to the temperature at the sternum in response to cooling. Area under the curve (AUC) between groups is displayed to the right. (J) Supraclavicular BAT skin temperature as measured by IRT during a thermal comfort state and analyzed as described in (G). (K) Sternum skin temperature during a thermal comfort state as measured by IRT and analyzed as described in (G). (L) IRT recording of the temperature changes in response to a thermal comfort state in supraclavicular BAT in relation to the temperature at the sternum. Because of experimental error, measurements were not available for all subjects at every time point. The time points at which data from fewer than three subjects were recorded were excluded. (M) Representative IRT images of a control subject (upper panel) and a winter swimmer (lower panel). The scale bar represents temperature in °C. (N) MRI assessing the distance between the activated BAT and the surface of the skin in the area where IRT images were obtained. Relative IRT values in (I) and (L) were calculated as described in the STAR Methods section. Data are presented as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S1.
Figure 2 Infrared thermography of supraclavicular BAT heat production at cooling and a thermal comfort (TC) state Subjects (WS, n = 7; C, n = 8) were cooled using a Blanketrol III system by controlling the temperature of circulating water in the blanket. n represents the number of human individuals in each group and is consistent throughout this figure unless otherwise specifically stated. (A) Water temperature in the blanket during cooling. (B) Estimate of the subjects’ temperature perception using a VAS score. (C) Core temperature as estimated by rectal thermometers. On a separate day, subjects were observed during the thermal comfort state protocol. (D) Water temperature during a thermal comfort state. (E) Estimate of the subjects’ temperature perception using a VAS score. (F) Core temperature as estimated by rectal thermometers. (G) Supraclavicular BAT skin temperature during cooling measured by IRT. The effect of the group and cooling were assessed during time points of 45–105 min. Sidak’s post-test was used to assess differences between the time point of 45 min (thermal comfort state) and the time points during cooling. (H) Sternum skin temperature during cooling as measured by IRT and analyzed as described in (G). (I) Supraclavicular BAT skin temperature in relation to the temperature at the sternum in response to cooling. Area under the curve (AUC) between groups is displayed to the right. (J) Supraclavicular BAT skin temperature as measured by IRT during a thermal comfort state and analyzed as described in (G). (K) Sternum skin temperature during a thermal comfort state as measured by IRT and analyzed as described in (G). (L) IRT recording of the temperature changes in response to a thermal comfort state in supraclavicular BAT in relation to the temperature at the sternum. Because of experimental error, measurements were not available for all subjects at every time point. The time points at which data from fewer than three subjects were recorded were excluded. (M) Representative IRT images of a control subject (upper panel) and a winter swimmer (lower panel). The scale bar represents temperature in °C. (N) MRI assessing the distance between the activated BAT and the surface of the skin in the area where IRT images were obtained. Relative IRT values in (I) and (L) were calculated as described in the STAR Methods section. Data are presented as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S1.CC BY-NC-ND · PubMed Central
Figure 3 PET/MRI scanning of subjects in response to perception-based cooling or a thermal comfort state PET/MRI scanning was part of the perception-based cooling or thermal comfort state protocols described in Figures 1 and 2, and data are shown from WS (n = 7) and C (n = 8). n represents the number of human individuals in each group and is consistent throughout this figure unless otherwise specifically stated. (A) Representative images of a subject (winter swimmer) at a thermal comfort state (left) and during cooling (right). (B) Glucose tracer (FDG) uptake in BAT measured as mean standardized uptake value (SUVmean) in the regions shown in (A). (C) SUVmean uptake multiplied by the volume of BAT with SUV above or equal to 2.0 during a TC state or during cooling as a measure of BAT metabolic volume (BMV). (D) Representative image showing glucose tracer uptake in the perirenal BAT (left), which was quantified for each subject and compared between groups and in response to a TC state and cooling (right). (E) Resting energy expenditure (REE) in winter swimmers and controls in response to a thermal comfort state or cooling. The actual analysis was performed for 30 min, and the energy consumption during 24 h represents a calculated value if the energy consumption remains constant during this period. (F) Cold-induced thermogenesis, calculated as the delta increase in energy expenditure induced by cooling. (G) Lean body mass as measured by DEXA scan as an estimation of muscle mass. (H) Electromyography (EMG) measurements recorded during 5 min at a thermal comfort state and during 5 min of cooling. Values are the area under the curve, including negative peaks, calculated per individual. (I) A subset of subjects displayed glucose tracer uptake in intercostal and/or neck muscles following cooling (left). The signal in intercostal muscles was graded 1–4 depending on intensity, and 0.5 was added if the subject also had a signal in the neck. The scale bars thus represent estimated arbitrary values for cold-induced glucose uptake in the intercostal and neck muscles. (J) Example of an MRI scan on a cooling day. Delineation of the region of interest (ROI) for calculation of the water percentage in areas with BAT (upper and middle panels), subcutaneous WAT (upper and lower panels), and skeletal muscle (middle panel). (K) Water percentage in WAT, BAT, and skeletal muscle as calculated from MRI data. (L) Water percentage in BAT in response to cooling. (M–O) Gene expression analysis of LIPE, PLIN1, and ELOVL6 in WAT between groups and in response to cooling. Differences between groups and treatments were assessed using mixed models and subsequent post-tests. Differences in AUC were assessed with an unpaired t test. When not otherwise specified, data are presented as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. See also Figure S2.
Figure 3 PET/MRI scanning of subjects in response to perception-based cooling or a thermal comfort state PET/MRI scanning was part of the perception-based cooling or thermal comfort state protocols described in Figures 1 and 2, and data are shown from WS (n = 7) and C (n = 8). n represents the number of human individuals in each group and is consistent throughout this figure unless otherwise specifically stated. (A) Representative images of a subject (winter swimmer) at a thermal comfort state (left) and during cooling (right). (B) Glucose tracer (FDG) uptake in BAT measured as mean standardized uptake value (SUVmean) in the regions shown in (A). (C) SUVmean uptake multiplied by the volume of BAT with SUV above or equal to 2.0 during a TC state or during cooling as a measure of BAT metabolic volume (BMV). (D) Representative image showing glucose tracer uptake in the perirenal BAT (left), which was quantified for each subject and compared between groups and in response to a TC state and cooling (right). (E) Resting energy expenditure (REE) in winter swimmers and controls in response to a thermal comfort state or cooling. The actual analysis was performed for 30 min, and the energy consumption during 24 h represents a calculated value if the energy consumption remains constant during this period. (F) Cold-induced thermogenesis, calculated as the delta increase in energy expenditure induced by cooling. (G) Lean body mass as measured by DEXA scan as an estimation of muscle mass. (H) Electromyography (EMG) measurements recorded during 5 min at a thermal comfort state and during 5 min of cooling. Values are the area under the curve, including negative peaks, calculated per individual. (I) A subset of subjects displayed glucose tracer uptake in intercostal and/or neck muscles following cooling (left). The signal in intercostal muscles was graded 1–4 depending on intensity, and 0.5 was added if the subject also had a signal in the neck. The scale bars thus represent estimated arbitrary values for cold-induced glucose uptake in the intercostal and neck muscles. (J) Example of an MRI scan on a cooling day. Delineation of the region of interest (ROI) for calculation of the water percentage in areas with BAT (upper and middle panels), subcutaneous WAT (upper and lower panels), and skeletal muscle (middle panel). (K) Water percentage in WAT, BAT, and skeletal muscle as calculated from MRI data. (L) Water percentage in BAT in response to cooling. (M–O) Gene expression analysis of LIPE, PLIN1, and ELOVL6 in WAT between groups and in response to cooling. Differences between groups and treatments were assessed using mixed models and subsequent post-tests. Differences in AUC were assessed with an unpaired t test. When not otherwise specified, data are presented as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. See also Figure S2.CC BY-NC-ND · PubMed Central
Figure 4 Diurnal thermoregulation and circulating hormones in adult humans A subset of the subjects included in the preceding experiments were invited for a subsequent experiment to study diurnal thermoregulation at a thermal comfort state (winter swimmers, n = 5; controls, n = 6). n represents the number of human individuals in each group and is consistent throughout the figure unless otherwise specifically stated. (A) Experimental setup. Subjects entered and spent the night at the laboratory before the recordings started and were then followed for 24 h. (B) Number of steps by the two groups. (C) Systolic blood pressure. (D) Diastolic blood pressure. (E) Core temperature recorded by an ingested temperature pill. Means without SD are shown. Area under the curve between groups is shown to the right. (F) Sternum temperature recorded by iButtons. Means without SD are shown. Area under the curve between groups is shown to the right. (G) BAT temperature recorded by using iButtons placed over supraclavicular BAT as detected from the IRT images during cooling. Means without SD are shown. Area under the curve between groups is shown to the right. (H) BAT temperature comparing the average during the day (1:01–2:01 p.m.), nightlow (4:01–4:11 a.m.), and nighthigh (4:49–4:59 a.m.). (I) Plasma cortisol over 24 h. Because of lack of samples, n for cortisol analysis was n = 3 winter swimmers and n = 3 control subjects. (J) Plasma IL-6 over 24 h. (K) Plasma leptin over 24 h. (L) Plasma adiponectin over 24 h. Differences between groups and treatments were assessed using two-way ANOVA or mixed models and subsequent post-tests. Differences in AUC was assessed with an unpaired t test. Data are presented as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S3.
Figure 4 Diurnal thermoregulation and circulating hormones in adult humans A subset of the subjects included in the preceding experiments were invited for a subsequent experiment to study diurnal thermoregulation at a thermal comfort state (winter swimmers, n = 5; controls, n = 6). n represents the number of human individuals in each group and is consistent throughout the figure unless otherwise specifically stated. (A) Experimental setup. Subjects entered and spent the night at the laboratory before the recordings started and were then followed for 24 h. (B) Number of steps by the two groups. (C) Systolic blood pressure. (D) Diastolic blood pressure. (E) Core temperature recorded by an ingested temperature pill. Means without SD are shown. Area under the curve between groups is shown to the right. (F) Sternum temperature recorded by iButtons. Means without SD are shown. Area under the curve between groups is shown to the right. (G) BAT temperature recorded by using iButtons placed over supraclavicular BAT as detected from the IRT images during cooling. Means without SD are shown. Area under the curve between groups is shown to the right. (H) BAT temperature comparing the average during the day (1:01–2:01 p.m.), nightlow (4:01–4:11 a.m.), and nighthigh (4:49–4:59 a.m.). (I) Plasma cortisol over 24 h. Because of lack of samples, n for cortisol analysis was n = 3 winter swimmers and n = 3 control subjects. (J) Plasma IL-6 over 24 h. (K) Plasma leptin over 24 h. (L) Plasma adiponectin over 24 h. Differences between groups and treatments were assessed using two-way ANOVA or mixed models and subsequent post-tests. Differences in AUC was assessed with an unpaired t test. Data are presented as mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S3.CC BY-NC-ND · PubMed Central

Verification

Claims the paper makes

0 Valid · 2 Not valid · 64 Not checked · 190 Own results and interpretations

81 more claims

Create a free account to see all 256 claims and to verify the 80 references still open.

See all claims — free account
  • It may state two separate claims as one. 74
  • Probably not a substantial claim. 137
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

4 of 4

Citations

Cites

Show all 70 references

Cited by

See all 80

Metadata

Authors
Søberg, S., Löfgren, J., Philipsen, F. E., Jensen, M., Hansen, A. E., Ahrens, E., Nystrup, K. B., Nielsen, R. D., Sølling, C., Wedell-Neergaard, A.-S., Berntsen, M., Loft, A., Kjær, A., Gerhart-Hines, Z., Johannesen, H. H., Pedersen, B. K., Karstoft, K., & Scheele, C.
Affiliations
  • University of Copenhagen (DK)
  • Rigshospitalet (DK)
Venue
Cell Reports Medicine
Publisher
Elsevier BV
Published
2021-10
Volume
2
Issue
10
Pages
100408
ISSN
2666-3791 (ISSN-L)
OpenAlex
W3205221523
Cited by
80
References
70
Citations by year
2026: 11 · 2025: 21 · 2024: 23 · 2023: 15 · 2022: 9 · 2021: 1
FWCI
3.50
Type
journal article
Language
en
Open access
Licence
Locations
  • Cell Reports Medicine (publishedVersion, cc-by-nc-nd)
  • SSRN Electronic Journal (publishedVersion)
  • Europe PMC (PubMed Central) (publishedVersion, other-oa)
  • PubMed (publishedVersion)
  • PubMed Central (publishedVersion, cc-by-nc-nd)
  • Research at the University of Copenhagen (University of Copenhagen) (publishedVersion, cc-by-nc-nd)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • European Research Council (639382)
  • Novo Nordisk Foundation (NNF17OC0029918, NNF18CC0034900)
  • Rigshospitalet (E-22793-05)
  • Horizon 2020
  • TrygFonden (101390, 20045)
  • Horizon 2020 Framework Programme
  • Gentofte Hospital
  • Københavns Universitet
Topics
  • Health Sciences › Medicine › Physiology (medical) › Adipose Tissue and Metabolism
  • Health Sciences › Medicine › Physiology (medical) › Thermoregulation and physiological responses
  • Health Sciences › Medicine › Critical Care and Intensive Care Medicine › Thermal Regulation in Medicine
Keywords
brown adipose tissue, thermoregulation, cold-induced thermogenesis, winter swimming, cold acclimation, thermal comfort, core body temperature, muscle activity, energy expenditure, sauna, circadian rhythm
MeSH terms
Adult, Adipose Tissue, Brown, Circadian Rhythm, Cold Temperature, Hormones, Humans, Magnetic Resonance Imaging, Male, Perception, Seasons, Skin Temperature, Swimming, Thermography, Thermogenesis, Positron-Emission Tomography, Young Adult

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.

A language model drew the claims from the paper on 2026-11-18, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-18 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

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: Altered brown fat thermoregulation and enhanced cold-induced thermogenesis in young, healthy, winter-swimming men. https://app.cochat.ai/papers/10.1016/j.xcrm.2021.100408
The paper
Søberg, S., Löfgren, J., Philipsen, F. E., Jensen, M., Hansen, A. E., Ahrens, E., Nystrup, K. B., Nielsen, R. D., Sølling, C., Wedell-Neergaard, A.-S., Berntsen, M., Loft, A., Kjær, A., Gerhart-Hines, Z., Johannesen, H. H., Pedersen, B. K., Karstoft, K., & Scheele, C. (2021). Altered brown fat thermoregulation and enhanced cold-induced thermogenesis in young, healthy, winter-swimming men. Cell Reports Medicine. https://doi.org/10.1016/j.xcrm.2021.100408

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.