Figure 1. Publications on Tourette syndrome. The number of new publications on Tourette syndrome or other tic disorders each year was estimated from PubMed. The colored line is a LOWESS (locally weighted scatterplot smoothing) curve from the primary data.
PubMed was searched on 29 Jun 2015 using the search string “("Tic Disorders"[MeSH] OR Tourette NOT Tourette[AU]) AND datestring
” for various values of datestring
. For datestring=1800[PDAT]:1950[PDAT]
, the search returned zero articles. Subsequent searches used datestring=year[PDAT]:next[PDAT]
for each year from 1950 through 2014, where next
means year + 1
. (This strategy addresses PubMed’s listing both electronic and subsequent paper publication dates for about 250 TS publications since 2005.) The graph was generated by matplotlib in python (see supplementary material).
f1000research_color=tuple([value/255.0 for value in (203,97,45)])
f1000research_h1 =tuple([value/255.0 for value in (55,67,76)])
f1000research_h2 =tuple([value/255.0 for value in (33,97,128)])
import numpy as np
import csv
with open('TS_publications_20150629.csv', 'rb') as datafile:
data2 = np.recfromcsv(datafile, names=True)
years = data2['year'][0:-1]
pubs = data2['pubsfinal'][0:-1]
import statsmodels.api as sm
b = sm.nonparametric.lowess(pubs,years,frac=1./3)
import matplotlib.pyplot as plt
%matplotlib inline
plt.scatter(years[1:],pubs[1:],color='black')
plt.plot(b[1:,0],b[1:,1],color=f1000research_color,linewidth=3)
plt.axis((1950,2015,0,250))
plt.xlabel('Year',fontsize=16)
plt.ylabel('Publications (PubMed)',fontsize=16)
plt.legend(['smoothed','new publications'], loc='upper left')
plt.savefig('TS_publications_2014_v2.png', dpi=600, transparent=True)
plt.show()
print "year\tpubs" ; print "====\t===="
for index in range(0,years.size):
print "{0:>}\t{1:>4}".format(years[index],pubs[index])