2.1 Analysis of the total number of publications
The variable of the total number of publications in the two time periods was found to obey a normal distribution through P-P chart analysis. Therefore, the comparison of the two means found that P = 0.000128 < 0.01(Table 1), and the difference is statistically significant. The average number of publications increased from 1271.0 in 2000–2009 to 2013.9 in 2010–2019. From the perspective of the average annual growth rate, the growth rate from 2000 to 2009 was 7.999%, which was greater than the 5.348% from 2010 to 2019, indicating that the total number of migraine publications from 2000 to 2009 increased faster than that from 2010 to 2019, and the total number of publications has also increased from 855 in 2000 to 2,757 in 2019 (Fig. 1), which can also be proved from the linear regression analysis results. The linear regression equation for 2000–2009 is Y = 102.2X + 708.7 (Fig. 1 and Table 1), the Pearson r correlation coefficient = 0.9348, P < 0.0001, indicating that it is greater than 99.5% Confident that the correlation coefficient reaches 0.9348, X and Y are linearly related; while the linear regression equation for 2010–2019 is Y = 101.7X + 1455 (Fig. 1 and Table 1), the Pearson r correlation coefficient = 0.8713, and P = 0.001 < 0.01, indicating that it is greater than 99.5% Confident that the correlation coefficient reaches 0.8713, X and Y are linearly related.
Table 1
Comparison of the mean that the total number of publications per year in two different periods
Group
|
TNP (x̄ ± s)
|
The average annual growth rate
|
F
|
P-Value
|
2000–2009
|
1271.0000 ± 331.15086
|
7.999%
|
0.007
|
0.000128
|
2010–2019
|
2013.9000 ± 353.36240
|
5.348%
|
TNP = Total Number of Publications |
Y = 102.2∗X + 708.7 |
Y = 101.7∗X + 1455 |
2.2 Distribution of Countries/Territories/Institutions and Cluster analysis
Comparing the high-yielding countries/territories in the two time periods, it is found that the USA, Italy, Germany, and England are the four countries/territories with the most published migraine studies in the two periods, and the number of papers published in the two periods has been increasing (Table 2). Besides, the number of migraine studies published in China from 2010 to 2019 also jumped to fifth place. However, from the perspective of centrality, the countries/territories engaged in migraine research from 2000 to 2009 are mainly Belgium, Netherlands, Hungary, and their centrality are 1.20, 1.12, and 0.74 respectively; and the most important countries from 2010 to 2019 are Pakistan (centrality = 1.08), U Arab Emirates (centrality = 1.08) and Ireland (centrality = 0.88).
Table 2
Countries/Territories/Institutions contributed to publications on migraine from 2000 to 2009 and 2010 to 2019
Different time periods
|
Country/ Territories
|
Frequency
|
Centrality
|
Country/ Territories
|
Institution
|
Frequency
|
Centrality
|
Institution
|
2000–2009
|
USA
|
4274
|
1.20
|
Belgium
|
Albert Einstein Coll Med
|
322
|
1.13
|
Houston Headache Clin
|
Italy
|
1303
|
1.12
|
Netherlands
|
Harvard Univ
|
278
|
0.73
|
Albert Einstein Coll Med
|
Germany
|
1198
|
0.74
|
Hungary
|
Univ Roma La Sapienza
|
236
|
0.69
|
Mayo Clin
|
England
|
1048
|
0.67
|
Poland
|
Univ Copenhagen
|
213
|
0.69
|
Duke Univ
|
Spain
|
522
|
0.65
|
England
|
Leiden Univ
|
192
|
0.64
|
Leiden Univ
|
Netherlands
|
508
|
0.65
|
Croatia
|
New England Ctr Headache
|
171
|
0.64
|
Diamond Headache Clin Ltd
|
France
|
465
|
0.64
|
Mexico
|
Mayo Clin
|
152
|
0.57
|
Univ Essen Gesamthsch
|
Canada
|
432
|
0.55
|
Ecuador
|
Univ Calif Los Angeles
|
151
|
0.57
|
Lipton
|
Denmark
|
330
|
0.47
|
Greece
|
Inst Neurol
|
142
|
0.51
|
Thomas Jefferson Univ
|
Turkey
|
328
|
0.41
|
Bulgaria
|
Thomas Jefferson Univ
|
135
|
0.45
|
Ortho McNeil Janssen Sci Affairs LLC
|
Brazil
|
297
|
0.30
|
Colombia
|
UCL Natl Hosp Neurol & Neurosurg
|
120
|
0.45
|
Univ Erlangen Nurnberg
|
Japan
|
296
|
0.28
|
USA
|
Univ Essen Gesamthsch
|
109
|
0.39
|
New England Ctr Headache
|
Australia
|
257
|
0.27
|
Chile
|
Univ Cincinnati
|
106
|
0.39
|
Michigan Head Pain & Neurol Inst
|
Belgium
|
242
|
0.20
|
Switzerland
|
Univ Liege
|
98
|
0.37
|
Drexel Univ
|
Sweden
|
232
|
0.20
|
Finland
|
Merck & Co Inc
|
95
|
0.36
|
UCL Natl Hosp Neurol & Neurosurg
|
2010–2019
|
USA
|
7107
|
1.08
|
Pakistan
|
Albert Einstein Coll Med
|
592
|
1.32
|
Amgen Inc
|
Italy
|
2382
|
1.08
|
U Arab Emirates
|
Univ Copenhagen
|
530
|
0.96
|
Vedanta Res
|
Germany
|
1516
|
0.88
|
Ireland
|
Mayo Clin
|
487
|
0.94
|
Albert Einstein Coll Med
|
England
|
1493
|
0.65
|
Malaysia
|
Harvard Univ
|
355
|
0.90
|
Montefiore Med Ctr
|
Peoples R China
|
866
|
0.60
|
Saudi Arabia
|
Leiden Univ
|
334
|
0.82
|
Montefiore Headache Ctr
|
Denmark
|
832
|
0.53
|
Norway
|
Kings Coll London
|
324
|
0.77
|
Novartis Pharma AG
|
Spain
|
825
|
0.39
|
Switzerland
|
Univ Calif San Francisco
|
277
|
0.73
|
Sapienza Univ Rome
|
Turkey
|
759
|
0.39
|
Estonia
|
Harvard Med Sch
|
250
|
0.62
|
Sapienza Univ
|
Netherlands
|
716
|
0.28
|
Belgium
|
Thomas Jefferson Univ
|
243
|
0.58
|
Univ Pavia
|
Canada
|
672
|
0.28
|
New Zealand
|
Univ Calif Los Angeles
|
236
|
0.57
|
Charite
|
France
|
630
|
0.25
|
Wales
|
Univ Pavia
|
223
|
0.54
|
Harvard Univ
|
Australia
|
531
|
0.18
|
USA
|
Univ Roma La Sapienza
|
205
|
0.50
|
Univ Roma La Sapienza
|
Brazil
|
527
|
0.18
|
Netherlands
|
Montefiore Med Ctr
|
199
|
0.45
|
Univ Mississippi
|
Norway
|
434
|
0.18
|
Greece
|
Amgen Inc
|
175
|
0.42
|
Univ Cincinnati
|
Switzerland
|
429
|
0.17
|
Scotland
|
Univ Sao Paulo
|
168
|
0.40
|
Cincinnati Childrens Hosp Med Ctr
|
Note: For specific abbreviations, please refer to the abbreviations at the end of the article. |
In 2000-2009 and 2010-2019, two network graphs containing 93 nodes and 88 countries/territories or 66 nodes and 73 countries/territories links were generated respectively (Figure 2), and two cluster maps with 117 nodes and 121 links (Modularity Q = 0.8154, Silhouette = 0.4925) or with 113 nodes and 115 links (Modularity Q = 0.8261, Silhouette = 0.6720) of institutions were emerged, along with timeline view (Figure 3). Rely on the cluster analysis of institutions, researchers can find the same research category that different institutions are engaged in. This point is mainly based on the cluster view Figure 3(A) and Table 3. When the cluster view can’t judge which institutions focus on the same research in detail, it will utilize the cluster timeline view Figure 3(B) to analyze. The whole modularity Q= 0.8154> 0.3 in 2000-2009, and whole modularity Q= 0.8261>0.3 in 2010-2019, which indicates that the results of the divided community structure are significant. 10 categories were obtained by cluster analysis in 2000-2009, which were #0 akershus study (S=0.913> 0.7, owns 16 institutions), #1 comparative trial (S=0.949>0.7, owns 13 institutions), #2 spectrum study (S=0.918>0.7, owns 13 institutions), #3 acute treatment (S=0.909>0.7, owns 12 institutions), #4 naproxen sodium 500mg (S=0.710>0.7, owns 11 institutions), #5 menstrual migraine (S=0.772>0.7, owns 11 institutions), #6 migraine prophylaxis (S=0.918>0.7, owns 9 institutions), #7 auditory evoked cortical potential (S=0.882>0.7, owns 8 institutions), #8 potential anti-migraine agent (S=0.952>0.7, owns 8 institutions), #9 red ear syndrome (S=0.941>0.7, owns 8 institutions), and 11 categories were gotten in 2010-2019, including #0 headache care, #1 juvenile patient, #2 chronic migraine prevention, #3 temporomandibular disorder, #4 patient-reported outcome, #5 ampp study, #6 systematic review, #7 depressive disorder, #8 pressure algometry data, #9 chronic migraine and #10 childhood maltreatment. The silhouette of all the clustering results was greater than 0.7, which indicates that the results of all categories of cluster analysis are very reliable. Also, the most important research categories in the two periods were a potential anti-migraine agent and headache care.
Table 3
Institutions engaged in migraine that Details of knowledge clusters from 2000 to 2009 and 2010 to 2019
Different periods
|
Cluster ID
|
Size
|
Silhouette
|
Mean (Year)
|
Label (LLR)
|
Label (MI)
|
2000–2009
|
0
|
16
|
0.913
|
2003
|
akershus study
|
self-reported muscle tension
|
1
|
13
|
0.949
|
2002
|
comparative trial
|
self-reported muscle tension
|
2
|
13
|
0.918
|
2003
|
spectrum study
|
self-reported muscle tension
|
3
|
12
|
0.909
|
2002
|
acute treatment
|
visual cortex excitability
|
4
|
11
|
0.710
|
2003
|
naproxen sodium 500 mg
|
premenstrual syndrome
|
5
|
11
|
0.772
|
2002
|
menstrual migraine
|
self-reported muscle tension
|
6
|
9
|
0.918
|
2002
|
migraine prophylaxis
|
ventroposteromedial nucleu
|
7
|
8
|
0.882
|
2001
|
auditory evoked cortical potential
|
self-reported muscle tension
|
8
|
8
|
0.952
|
2000
|
potential anti-migraine agent
|
feline brain
|
9
|
8
|
0.941
|
2003
|
red ear syndrome
|
plasma level
|
2010–2019
|
0
|
13
|
0.919
|
2012
|
headache care
|
pediatric patient
|
1
|
12
|
0.956
|
2013
|
juvenile patient
|
two-year follow-up point
|
2
|
12
|
0.956
|
2014
|
chronic migraine prevention
|
prospective multicentre descriptive study
|
3
|
11
|
0.936
|
2012
|
temporomandibular disorder
|
psychiatric comorbidities
|
4
|
10
|
0.916
|
2014
|
patient-reported outcome
|
two-year follow-up point
|
5
|
9
|
0.842
|
2012
|
ampp study
|
two-year follow-up point
|
6
|
9
|
0.805
|
2011
|
systematic review
|
two-year follow-up point
|
7
|
9
|
0.969
|
2010
|
depressive disorder
|
two-year follow-up point
|
8
|
9
|
0.930
|
2011
|
pressure algometry data
|
medieval headache classification
|
9
|
8
|
0.942
|
2012
|
chronic migraine
|
two-year follow-up point
|
10
|
7
|
0.910
|
2012
|
childhood maltreatment
|
brain excitability
|
2.3 Distribution of authors and Cluster analysis
The distribution of the main scholars engaged in migraine is shown in Table 4. From 2000 to 2009, the most prolific author was Lipton RB (Frequency = 279), followed by Goadsby PJ (Frequency = 250), Diener HC (Frequency = 243), Bigal ME (Frequency = 189), and Silberstein SD (Frequency = 167), the most critical author was also Lipton RB (centrality = 1.22), followed by Dahlof C (centrality = 0.74), O’Quinn S (centrality = 0.74), Cady R (centrality = 0.70) and Diamond M (centrality = 0.67). In 2010–2019, compared with 2000–2009, the most productive author is still Lipton RB (Frequency = 472), followed by Goadsby PJ (Frequency = 370), Buse DC (Frequency = 256), Ashina M (Frequency = 238) and Ferrari MD (Frequency = 226), and the most vital author is Klatt J (centrality = 0.89), followed by Buse DC (centrality = 0.88), Mikol DD (centrality = 0.80), Hareendran A (centrality = 0.80) and Lipton RB (centrality = 0.78), therefore, they are considered to be the most important researcher in the Author’s network relationship.
Table 4
Authors/Cited References contributed to publications on migraine from 2000 to 2009 and 2010 to 2019
Different periods
|
Author
|
Frequency
|
Centrality
|
Author
|
Cited Reference
|
Frequency
|
Centrality
|
Cited Reference
|
|
Lipton RB
|
279
|
1.22
|
Lipton RB
|
Silberstein SD (2004)
|
1690
|
1.16
|
Welch KMA (2001)
|
|
Goadsby PJ
|
250
|
0.74
|
Dahlof C
|
Lipton RB (2001)
|
543
|
1.11
|
Weiller C (1995)
|
|
Diener HC
|
243
|
0.74
|
O'Quinn S
|
Goadsby PJ (2002)
|
536
|
0.98
|
Solomon GD (1997)
|
|
Bigal ME
|
189
|
0.70
|
Cady R
|
Ferrari MD (2001)
|
322
|
0.95
|
May A (1998)
|
|
Silberstein SD
|
167
|
0.67
|
Diamond M
|
Ophoff RA (1996)
|
300
|
0.95
|
Goadsby PJ (1993)
|
|
Ferrari MD
|
163
|
0.51
|
Massiou H
|
Silberstein SD (2000)
|
285
|
0.95
|
Goadsby PJ (1994)
|
|
Olesen J
|
145
|
0.48
|
Diener HC
|
Hadjikhani N (2001)
|
257
|
0.94
|
Rapoport AM (1997)
|
2000–2009
|
Schoenen J
|
129
|
0.45
|
Dowson AJ
|
De FM (2003)
|
248
|
0.94
|
Longmore J (1997)
|
|
Bussone G
|
115
|
0.45
|
MacGregor EA
|
Hu XH (1999)
|
248
|
0.92
|
Vandenbrink AM (1998)
|
|
Dodick DW
|
106
|
0.44
|
Reed ML
|
Olesen J (2004)
|
245
|
0.86
|
Ayata C (2006)
|
|
Nappi G
|
105
|
0.36
|
Stewart WF
|
Kruit MC (2004)
|
242
|
0.81
|
Pascual J (2000)
|
|
Evers S
|
99
|
0.35
|
Lines CR
|
Lipton RB (2001)
|
242
|
0.78
|
Perry CM (1998)
|
|
Pascual J
|
86
|
0.35
|
McCarroll KA
|
Brandes JL (2004)
|
236
|
0.78
|
Mathew NT (1997)
|
|
Tepper SJ
|
84
|
0.34
|
Kolodner K
|
Tfelt-hansen P (2000)
|
231
|
0.59
|
Scher AI (2003)
|
|
Rapoport AM
|
75
|
0.33
|
Lipton R
|
Bolay H (2002)
|
224
|
0.57
|
Hadjikhani N (2001)
|
|
Lipton RB
|
472
|
0.89
|
Klatt J
|
Bes A (2013)
|
1711
|
0.89
|
Goadsby PJ (2017)
|
|
Goadsby PJ
|
370
|
0.88
|
Buse DC
|
Silberstein SD (2004)
|
1307
|
0.75
|
Dodick DW (2014)
|
|
Buse DC
|
256
|
0.80
|
Mikol DD
|
Stovner LJ (2007)
|
481
|
0.73
|
Silberstein SD (2012)
|
|
Ashina M
|
238
|
0.80
|
Hareendran A
|
Lipton RB (2007)
|
467
|
0.71
|
Tepper S (2017)
|
|
Ferrari MD
|
226
|
0.78
|
Lipton RB
|
Olesen J (2006)
|
384
|
0.71
|
Hepp Z (2015)
|
|
Olesen J
|
217
|
0.78
|
Turkel CC
|
Vos T (2012)
|
335
|
0.70
|
Sun H (2016)
|
|
Dodick DW
|
207
|
0.77
|
Zhang F
|
Silberstein SD (2012)
|
283
|
0.69
|
Lipton RB (2007)
|
2010–2019
|
Silberstein SD
|
182
|
0.73
|
Manack A
|
Diener HC (2010)
|
283
|
0.69
|
Ho TW (2008)
|
|
Diener HC
|
180
|
0.71
|
DeGryse RE
|
Schurks M (2009)
|
261
|
0.64
|
Ho TW (2010)
|
|
Tassorelli C
|
163
|
0.66
|
Lanteri-Minet M
|
Aurora SK (2010)
|
257
|
0.64
|
Munakata J (2009)
|
|
Martelletti P
|
156
|
0.64
|
Barbanti P
|
Dodick DW (2010)
|
247
|
0.57
|
Natoli JL (2010)
|
|
Wang SJ
|
131
|
0.62
|
Martelletti P
|
Evers S (2009)
|
242
|
0.52
|
Ho TW (2008)
|
|
Terwindt GM
|
129
|
0.49
|
Diener HC
|
Pietrobon D (2013)
|
240
|
0.49
|
Olesen J (2009)
|
|
Schoenen J
|
116
|
0.39
|
Tassorelli C
|
Akerman S (2011)
|
238
|
0.46
|
Aurora SK (2011)
|
|
Grazzi L
|
112
|
0.39
|
Nappi G
|
Blumenfeld AM (2011)
|
220
|
0.46
|
Zhang XC (2011)
|
2 cluster maps with 142 nodes and 145 links (Fig. 4(A), Modularity Q = 0.8357, Silhouette = 0.5962) or with 140 nodes and 135 links (Fig. 4(C), Modularity Q = 0.8469, Silhouette = 0.6471) of authors were emerged by Citespace, along with timeline view (Fig. 4(B) and Fig. 4(D)). The modularity Q > 0.3, which indicates that the results of the divided community structure are prominent. Then through the cluster analysis, the researcher can quickly locate the same research content and the relationship between different scholars. The results of cluster analysis are shown in Table 5. During 2000 to 2009, the content of cluster analysis included 12 categories, mainly including #0 population-based survey (S = 0.919 > 0.7, Contains 17 authors), #1 menstrual migraine (S = 0.795 > 0.7, Contains 17 authors), #2 migraine prevention (S = 0.909 > 0.7, Contains 14 authors), #3 migraine patient (S = 0.957 > 0.7, Contains 13 authors), #4 disintegrating tablet (S = 0.934 > 0.7, Contains 12 authors), #5 strong point (S = 0.973 > 0.7, Contains 12 authors), #6 medication overuse (S = 0.812 > 0.7, Contains 10 authors), #7 cardiovascular disease (S = 0.976 > 0.7, Contains 9 authors), #8 chronic daily headache (S = 0.978 > 0.7, Contains 8 authors), #9 human isolated blood vessel (S = 0.926 > 0.7, Contains 7 authors), #10 alternating hemiplegia (S = 0.953 > 0.7, Contains 5 authors), #11 reducing headache recurrence (S = 0.985 > 0.7, Contains 4 authors). In 2010–2019, a total of 11 cluster analysis results were obtained, namely chronic migraine, following symptomatic treatment, patient-reported outcome, eurolight questionnaire, American migraine prevalence, preempt clinical program, medication-overuse headache, genome-wide association study, America symptom, chronic migraine, population-based study. The silhouette of all the clustering results was greater than 0.7, which indicates that the results of all categories of cluster analysis are very credible. In addition, the most important categories in the two periods were reducing headache recurrence and following symptomatic treatment.
Table 5
Authors engaged in migraine that Details of knowledge clusters from 2000 to 2009 and 2010 to 2019
Different periods
|
Cluster ID
|
Size
|
Silhouette
|
Mean (Year)
|
Label (LLR)
|
Label (MI)
|
|
0
|
17
|
0.919
|
2003
|
population-based survey
|
identifying functional disability
|
|
1
|
17
|
0.795
|
2001
|
menstrual migraine
|
causing cadasil
|
|
2
|
14
|
0.909
|
2002
|
migraine prevention
|
post-marketing experience
|
|
3
|
13
|
0.957
|
2002
|
migraine patient
|
identifying functional disability
|
|
4
|
12
|
0.934
|
2003
|
disintegrating tablet
|
identifying functional disability
|
2000–2009
|
5
|
12
|
0.973
|
2004
|
strong point
|
mtdna a3243g mutation
|
|
6
|
10
|
0.812
|
2002
|
medication overuse
|
identifying functional disability
|
|
7
|
9
|
0.976
|
2003
|
cardiovascular disease
|
identifying functional disability
|
|
8
|
8
|
0.978
|
2002
|
chronic daily headache
|
identifying functional disability
|
|
9
|
7
|
0.926
|
2001
|
human isolated blood vessel
|
identifying functional disability
|
|
10
|
5
|
0.953
|
2003
|
alternating hemiplegia
|
genetic relation
|
|
11
|
4
|
0.985
|
2004
|
reducing headache recurrence
|
functional assessment measure
|
|
0
|
13
|
0.956
|
2014
|
chronic migraine
|
adolescent perspective
|
|
1
|
13
|
0.986
|
2012
|
following symptomatic treatment
|
adolescent perspective
|
|
2
|
12
|
0.858
|
2015
|
patient-reported outcome
|
adolescent perspective
|
|
3
|
12
|
0.944
|
2011
|
eurolight questionnaire
|
adolescent perspective
|
2010–2019
|
4
|
12
|
0.805
|
2014
|
american migraine prevalence
|
adolescent perspective
|
|
5
|
11
|
0.859
|
2011
|
preempt clinical program
|
adolescent perspective
|
|
6
|
11
|
0.966
|
2011
|
medication-overuse headache
|
adolescent perspective
|
|
7
|
11
|
0.944
|
2012
|
genome-wide association study
|
adolescent perspective
|
|
8
|
10
|
0.970
|
2013
|
america symptom
|
adolescent perspective
|
|
9
|
10
|
0.979
|
2012
|
chronic migraine
|
adolescent perspective
|
|
10
|
8
|
0.969
|
2011
|
population-based study
|
adolescent perspective
|
2.4 Distribution of Cited reference and Cluster analysis
The knowledge base and research progress of migraine are described using high co-citation references and main article clusters respectively, and the co-citation frequencies of references and their importance in network nodes are shown in Table 4. The reference with the highest co-citations in 2000–2009 is Silberstein SD (2004)[19] ( Frequency = 1690), followed by Lipton RB (2001) ( Frequency = 543), Goadsby PJ (2002) ( Frequency = 536), Ferrari MD (2001) ( Frequency = 322) and Ophoff RA (1996) ( Frequency = 300), and the highest centrality is Welch KMA (2001)[20] ( Centrality = 1.16) and is regarded as the most important reference in the field, followed by Weiller C (1995) ( Centrality = 1.11), Solomon GD (1997) ( Centrality = 0.98), May A (1998) ( Centrality = 0.95) and Goadsby PJ (1993) ( Centrality = 0.95). Between 2010 and 2019, the reference with the highest co-citations is Bes A (2013) [21] ( Frequency = 1711), followed by Silberstein SD (2004) ( Frequency = 1307), Stovner LJ (2007) ( Frequency = 481), Lipton RB (2007) ( Frequency = 467) and Olesen J (2006) ( Frequency = 384), and the highest centrality is Goadsby PJ (2017)[22]( Centrality = 0.89), and is considered as the most momentous reference in the field, followed by Weiller C (1995) ( Centrality = 1.11), Dodick DW (2014) (Centrality = 0.75), Silberstein SD (2012) ( Centrality = 0.73) and Tepper S (2017) ( Centrality = 0.71).
Two cluster maps with 149 nodes and 159 links (Fig. 5(A), Modularity Q = 0.8481, Silhouette = 0.7417) or with 143 nodes and 141 links (Fig. 5(C), Modularity Q = 0.8467, Silhouette = 0.4431) of references was generated by Citespace, along with timeline view (Fig. 5(B) and Fig. 5(D)). For the cluster analysis (Table 6) between 2000 and 2009, the twelve largest clusters (small clusters were automatically filtered), included #0 comparative review, #1 chronic migraine, #2 economic outcome, #3 delayed treatment, #4 contrasting patient preference, #5 headache sufferers study, #6 migraine headache, #7 oral contraceptive, #8 chronic daily headache, #9 trigeminovascular system, #10 using rizatriptan, #11 to-left shunt. The silhouette of all the clustering results was greater than 0.7, which indicates that all clustering results are dependable. Moreover, the category with the highest silhouette value is #11 to-left shunt, which shows that the category is the most convincing in this cluster analysis. From 2010 to 2019, 11 cluster analysis results were acquired, which were #0 placebo-controlled phase (S = 0.970 > 0.7, possesses 14 references), #1 excitatory-inhibitory balance (S = 0.899 > 0.7, possesses 12 references), #2 chronic migraine (S = 0.947 > 0.7, possesses 12 references), #3 new analysis (S = 0.955 > 0.7, possesses 12 references), #4 familial hemiplegic migraine (S = 0.966 > 0.7, possesses 11 references), #5 placebo effect (S = 0.909 > 0.7, possesses 11 references), #6 emerging drug (S = 0.904 > 0.7, possesses 11 references), #7 high frequency migraine (S = 0.957 > 0.7, possesses 11 references), #8 calcium channel function (S = 0.848 > 0.7, possesses 10 references), #9 key player neuropeptide (S = 0.827 > 0.7, possesses 10 references), #11 familial migraine (S = 0.895 > 0.7, possesses 7 references). All the clustering silhouettes are greater than 0.7, indicating that the results are very reliable. Furthermore, the category with the highest silhouette value is #0 placebo-controlled phase, which prompts that the category is the most persuasive in this cluster analysis.
Table 6
Cited reference engaged in migraine that Details of knowledge clusters from 2000 to 2009 and 2010 to 2019
Different periods
|
Cluster ID
|
Size
|
Silhouette
|
Mean (Year)
|
Label (LLR)
|
Label (MI)
|
|
0
|
15
|
0.911
|
1997
|
comparative review
|
migraine
|
|
1
|
14
|
0.979
|
2004
|
chronic migraine
|
migraine
|
|
2
|
14
|
0.912
|
1996
|
economic outcome
|
migraine
|
|
3
|
13
|
0.974
|
1999
|
delayed treatment
|
migraine
|
|
4
|
13
|
0.876
|
2001
|
contrasting patient preference
|
migraine
|
2000–2009
|
5
|
12
|
0.976
|
2001
|
headache sufferers study
|
migraine
|
|
6
|
12
|
0.892
|
1997
|
migraine headache
|
calcium channelopathies
|
|
7
|
12
|
0.962
|
2002
|
oral contraceptive
|
migraine
|
|
8
|
12
|
0.979
|
2000
|
chronic daily headache
|
migraine
|
|
9
|
11
|
0.813
|
2003
|
trigeminovascular system
|
underlying mechanism
|
|
10
|
10
|
0.869
|
1995
|
using rizatriptan
|
migraine
|
|
11
|
8
|
0.981
|
2002
|
to-left shunt
|
migraine
|
|
0
|
14
|
0.970
|
2010
|
placebo-controlled phase
|
to-resting glucose uptake ratio
|
|
1
|
12
|
0.899
|
2009
|
excitatory-inhibitory balance
|
to-resting glucose uptake ratio
|
|
2
|
12
|
0.947
|
2010
|
chronic migraine
|
to-resting glucose uptake ratio
|
|
3
|
12
|
0.955
|
2010
|
new analysis
|
to-resting glucose uptake ratio
|
|
4
|
11
|
0.966
|
2005
|
familial hemiplegic migraine
|
to-resting glucose uptake ratio
|
2010–2019
|
5
|
11
|
0.909
|
2007
|
placebo effect
|
to-resting glucose uptake ratio
|
|
6
|
11
|
0.904
|
2016
|
emerging drug
|
to-resting glucose uptake ratio
|
|
7
|
11
|
0.957
|
2012
|
high frequency migraine
|
to-resting glucose uptake ratio
|
|
8
|
10
|
0.848
|
2009
|
calcium channel function
|
facial expression
|
|
9
|
10
|
0.827
|
2014
|
key player neuropeptide
|
pharmacologic characterization
|
|
11
|
7
|
0.895
|
2007
|
familial migraine
|
to-resting glucose uptake ratio
|
2.5 Distribution of Keywords and Cluster analysis
Table 7 shows keywords contributed to publications on migraine, the top five keywords in 2000–2009 are migraine (Frequency = 5308), headache (2413), prevalence (1223), double blind (844), and sumatriptan (754). But from the perspective of centrality, the most vital keyword in the network is migraine (centrality = 1.45), followed by sumatriptan (centrality = 0.97), prevalence (0.96), headache (0.85), aura (centrality = 0.65). Thus, compared with the first stage, migraine, headache, prevalence, double blind were also in the top four, only increasing in number. And the most critical keywords are the same as the previous period, which is migraine (centrality = 1.56). The next four different keywords are double blind (centrality = 1.07), prevalence (centrality = 0.71), aura (centrality = 0.42), and gene related peptide (centrality = 0.38). Therefore, these keywords constitute the main content of migraine research.
Table 7
Keyword contributed to publications on migraine from 2000 to 2009 and 2010 to 2019
Different periods
|
Keyword
|
Frequency
|
Centrality
|
Keyword
|
|
migraine
|
5308
|
1.45
|
migraine
|
|
headache
|
2413
|
0.97
|
sumatriptan
|
|
prevalence
|
1223
|
0.96
|
prevalence
|
|
double blind
|
844
|
0.85
|
headache
|
|
sumatriptan
|
754
|
0.65
|
aura
|
|
pain
|
641
|
0.61
|
rizatriptan
|
|
aura
|
564
|
0.58
|
efficacy
|
2000–2009
|
epidemiology
|
534
|
0.39
|
united states
|
|
efficacy
|
500
|
0.35
|
serotonin
|
|
population
|
479
|
0.29
|
patent foramen ovale
|
|
children
|
461
|
0.25
|
stroke
|
|
tension type headache
|
454
|
0.20
|
mutation
|
|
united states
|
435
|
0.20
|
risk factor
|
|
familial hemiplegic migraine
|
434
|
0.20
|
burden
|
|
stroke
|
400
|
0.15
|
familial hemiplegic migraine
|
|
migraine
|
7831
|
1.56
|
migraine
|
|
headache
|
3903
|
1.07
|
double blind
|
|
prevalence
|
2453
|
0.71
|
prevalence
|
|
double blind
|
1467
|
0.42
|
aura
|
|
pain
|
1413
|
0.38
|
gene related peptide
|
|
epidemiology
|
881
|
0.32
|
patent foramen ovale
|
|
children
|
870
|
0.28
|
burden
|
2010–2019
|
chronic migraine
|
858
|
0.28
|
stroke
|
|
aura
|
802
|
0.22
|
placebo controlled phase
|
|
population
|
746
|
0.17
|
pain
|
|
efficacy
|
734
|
0.17
|
depression
|
|
depression
|
727
|
0.17
|
disability
|
|
burden
|
700
|
0.17
|
cortical spreading depression
|
|
tension type headache
|
664
|
0.17
|
topiramate
|
|
quality of life
|
648
|
0.17
|
onabotulinumtoxina
|
Figure 6 and Table 8 exhibit keyword co-occurrence cluster analysis, 2 knowledge mapping with 76 nodes and 84 keywords links (modular Q = 0.7884, Mean silhouette = 0.8275) or with 69 nodes and 74 keywords links (modular Q = 0.7791, Mean silhouette = 0.8579) and cluster view are created. The modularity Q > 0.3, which indicates that the results of the divided community structure are remarkable. The timeline view shows that new keywords appear almost every year. In the cluster map Fig. 6(A), the eight largest clusters (small clusters were automatically filtered) between 2000 and 2009, contained #0 spectrum study (S = 0.985 > 0.7, contains 13 keywords), #1 trigeminal ganglion cell (S = 0.917 > 0.7, contains 12 keywords), #2 common experience (S = 0.850 > 0.7, contains 11 keywords), #3 cytokine level (S = 0.962 > 0.7, contains 11 keywords), #4 sumatriptan tablet (S = 0.965 > 0.7, contains 9 keywords), #5 familial hemiplegic migraine (S = 0.925 > 0.7, contains 8 keywords), #6 to-left shunt (S = 0.905 > 0.7, contains 7 keywords), #7 chronic migraine (S = 0.939 > 0.7, contains 4 keywords). By 2010–2019, cluster analysis has obtained nine research categories, which are #0 chronic migraine (S = 0.962 > 0.7, contains 12 keywords), #1 chronic migraine (S = 0.946 > 0.7, contains 11 keywords), #2 tension-type headache (S = 0.962 > 0.7, contains 9 keywords), #3 transient global amnesia (S = 0.936 > 0.7, contains 9 keywords), #4 familial hemiplegic migraine (S = 0.982 > 0.7, contains 8 keywords), #5 migraine education (S = 0.955 > 0.7, contains 6 keywords), #6 chronic migraine (S = 0.925 > 0.7, contains 5 keywords), #7 laser-evoked potential (S = 0.936 > 0.7, contains 4 keywords), #8 anxiety symptom (S = 0.974 > 0.7, contains 4 keywords). The silhouette of all the clustering results was greater than 0.7, it states clearly that all clustering results are authentic. Hence, after comparing the silhouette values of cluster analysis, it is found that spectrum study and familial hemiplegic migraine are the most important research contents.
Table 8
Keywords engaged in migraine that Details of knowledge clusters from 2000 to 2009 and 2010 to 2019
Different periods
|
Cluster ID
|
Size
|
Silhouette
|
Mean (Year)
|
Label (LLR)
|
Label (MI)
|
|
0
|
13
|
0.985
|
2002
|
spectrum study
|
prothrombotic risk factor
|
|
1
|
12
|
0.917
|
2000
|
trigeminal ganglion cell
|
prothrombotic risk factor
|
|
2
|
11
|
0.850
|
2001
|
common experience
|
prothrombotic risk factor
|
2000–2009
|
3
|
11
|
0.962
|
2000
|
cytokine level
|
prothrombotic risk factor
|
|
4
|
9
|
0.965
|
2000
|
sumatriptan tablet
|
prothrombotic risk factor
|
|
5
|
8
|
0.925
|
2002
|
familial hemiplegic migraine
|
prothrombotic risk factor
|
|
6
|
7
|
0.905
|
2002
|
to-left shunt
|
prothrombotic risk factor
|
|
7
|
4
|
0.939
|
2004
|
chronic migraine
|
prothrombotic risk factor
|
|
0
|
12
|
0.962
|
2011
|
chronic migraine
|
occipital nerve block
|
|
1
|
11
|
0.946
|
2011
|
chronic migraine
|
occipital nerve block
|
|
2
|
9
|
0.962
|
2010
|
tension-type headache
|
occipital nerve block
|
|
3
|
9
|
0.936
|
2010
|
transient global amnesia
|
occipital nerve block
|
2010–2019
|
4
|
8
|
0.982
|
2011
|
familial hemiplegic migraine
|
triggering migraine
|
|
5
|
6
|
0.955
|
2010
|
migraine education
|
occipital nerve block
|
|
6
|
5
|
0.925
|
2012
|
chronic migraine
|
occipital nerve block
|
|
7
|
4
|
0.936
|
2010
|
laser-evoked potential
|
occipital nerve block
|
|
8
|
4
|
0.974
|
2010
|
anxiety symptom
|
alcohol dependence
|
Burst is used to reflecting hotspots within a certain time node. The red bar indicates the duration of the citation explosion and the start and end year that the citation explosion. That is to say: the red bar indicates some frequently cited keywords, and the green bar is rarely cited in terms of keywords[23], it was found that the hotspot of migraine between 2000 and 2009 was “stimulation”. And from 2010 to 2019, the hotspot of migraine was “triptan”. The specific results are shown in Fig. 7.