vignettes/AMR.Rmd
AMR.Rmd
Using the sample() function, we can randomly select items from all objects we defined earlier. To let our fake data reflect reality a bit, we will also approximately define the probabilities of -bacteria and the antibiotic results, using the random_rsi() +bacteria and the antibiotic results, using the random_sir() function.
sample()
random_rsi()
random_sir()
sample_size <- 20000 @@ -360,10 +360,10 @@ function. size = sample_size, replace = TRUE, prob = c(0.50, 0.25, 0.15, 0.10) ), - AMX = random_rsi(sample_size, prob_RSI = c(0.35, 0.60, 0.05)), - AMC = random_rsi(sample_size, prob_RSI = c(0.15, 0.75, 0.10)), - CIP = random_rsi(sample_size, prob_RSI = c(0.20, 0.80, 0.00)), - GEN = random_rsi(sample_size, prob_RSI = c(0.08, 0.92, 0.00)) + AMX = random_sir(sample_size, prob_sir = c(0.35, 0.60, 0.05)), + AMC = random_sir(sample_size, prob_sir = c(0.15, 0.75, 0.10)), + CIP = random_sir(sample_size, prob_sir = c(0.20, 0.80, 0.00)), + GEN = random_sir(sample_size, prob_sir = c(0.08, 0.92, 0.00)) )
Using the left_join() function from the dplyr package, we can ‘map’ the gender to the patient ID @@ -380,12 +380,12 @@ data set:
left_join()
dplyr
Now, let’s start the cleaning and the analysis!
data <- data %>% mutate(bacteria = as.mo(bacteria))
We also want to transform the antibiotics, because in real life data -we don’t know if they are really clean. The as.rsi() +we don’t know if they are really clean. The as.sir() function ensures reliability and reproducibility in these kind of -variables. The is.rsi.eligible() can check which columns -are probably columns with R/SI test results. Using mutate() +variables. The is_sir_eligible() can check which columns +are probably columns with SIR test results. Using mutate() and across(), we can apply the transformation to the formal <rsi> class:
as.rsi()
as.sir()
is.rsi.eligible()
mutate()
is_sir_eligible()
across()
<rsi>
-is.rsi.eligible(data) +is_sir_eligible(data) # [1] FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE FALSE -colnames(data)[is.rsi.eligible(data)] +colnames(data)[is_sir_eligible(data)] # [1] "AMX" "AMC" "CIP" "GEN" data <- data %>% - mutate(across(where(is.rsi.eligible), as.rsi))
is.rsi.eligible(data) +is_sir_eligible(data) # [1] FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE FALSE -colnames(data)[is.rsi.eligible(data)] +colnames(data)[is_sir_eligible(data)] # [1] "AMX" "AMC" "CIP" "GEN" data <- data %>% - mutate(across(where(is.rsi.eligible), as.rsi))
is_sir_eligible(data) # [1] FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE FALSE -colnames(data)[is.rsi.eligible(data)] +colnames(data)[is_sir_eligible(data)] # [1] "AMX" "AMC" "CIP" "GEN" data <- data %>% - mutate(across(where(is.rsi.eligible), as.rsi))
Finally, we will apply EUCAST rules on our antimicrobial results. In Europe, most medical microbiological laboratories already apply these rules. Our package @@ -621,9 +621,9 @@ takes into account the antimicrobial susceptibility test results using # Basing inclusion on all antimicrobial results, using a points threshold of # 2 # Including isolates from ICU. -# => Found 10,664 'phenotype-based' first isolates (53.3% of total where a +# => Found 12,360 'phenotype-based' first isolates (61.8% of total where a # microbial ID was available) -
So only 53.3% is suitable for resistance analysis! We can now filter +
So only 61.8% is suitable for resistance analysis! We can now filter on it with the filter() function, also from the dplyr package:
filter()
@@ -634,25 +634,25 @@ on it with the data_1st <- data %>% filter_first_isolate() # Including isolates from ICU.
data_1st <- data %>% filter_first_isolate() # Including isolates from ICU.
So we end up with 10,664 isolates for analysis. Now our data looks +
So we end up with 12,360 isolates for analysis. Now our data looks like:
head(data_1st)
data_1st %>% freq(genus, species)
Frequency table
Class: character -Length: 10,664 -Available: 10,664 (100%, NA: 0 = 0%) +Length: 12,360 +Available: 12,360 (100%, NA: 0 = 0%) Unique: 4
Shortest: 16 Longest: 24
proportion_SI()
data_1st %>% resistance(AMX) -# [1] 0.5381658
Or can be used in conjunction with group_by() and summarise(), both from the dplyr package:
group_by()
summarise()
@@ -1152,25 +1152,25 @@ own: Hospital A -0.5460547 +0.5845627 Hospital B -0.5252972 +0.5830477 Hospital C -0.5231231 +0.5790359 Hospital D -0.5592912 +0.5767790 Of course it would be very convenient to know the number of isolates responsible for the percentages. For that purpose the -n_rsi() can be used, which works exactly like +n_sir() can be used, which works exactly like n_distinct() from the dplyr package. It counts all isolates available for every group (i.e. values S, I or R): @@ -1178,7 +1178,7 @@ all isolates available for every group (i.e. values S, I or R): group_by(hospital) %>% summarise( amoxicillin = resistance(AMX), - available = n_rsi(AMX) + available = n_sir(AMX) )
Of course it would be very convenient to know the number of isolates responsible for the percentages. For that purpose the -n_rsi() can be used, which works exactly like +n_sir() can be used, which works exactly like n_distinct() from the dplyr package. It counts all isolates available for every group (i.e. values S, I or R):
n_rsi()
n_sir()
n_distinct()
@@ -1178,7 +1178,7 @@ all isolates available for every group (i.e. values S, I or R): group_by(hospital) %>% summarise( amoxicillin = resistance(AMX), - available = n_rsi(AMX) + available = n_sir(AMX) )
The AMR package contains functions to extend this -ggplot2 package, for example geom_rsi(). It +ggplot2 package, for example geom_sir(). It automatically transforms data with count_df() or proportion_df() and show results in stacked bars. Its simplest and shortest example:
AMR
ggplot2
geom_rsi()
geom_sir()
count_df()
proportion_df()
ggplot(data_1st) + - geom_rsi(translate_ab = FALSE)
Omit the translate_ab = FALSE to have the antibiotic codes (AMX, AMC, CIP, GEN) translated to official WHO names @@ -1363,15 +1363,15 @@ additional functions from our package, we can create this:
translate_ab = FALSE
# group the data on `genus` ggplot(data_1st %>% group_by(genus)) + # create bars with genus on x axis - # it looks for variables with class `rsi`, - # of which we have 4 (earlier created with `as.rsi`) - geom_rsi(x = "genus") + + # it looks for variables with class `sir`, + # of which we have 4 (earlier created with `as.sir`) + geom_sir(x = "genus") + # split plots on antibiotic - facet_rsi(facet = "antibiotic") + - # set colours to the R/SI interpretations (colour-blind friendly) - scale_rsi_colours() + + facet_sir(facet = "antibiotic") + + # set colours to the SIR interpretations (colour-blind friendly) + scale_sir_colours() + # show percentages on y axis - scale_y_percent(breaks = 0:4 * 25) + + scale_y_percent(breaks = 0:4 * 25) + # turn 90 degrees, to make it bars instead of columns coord_flip() + # add labels @@ -1383,12 +1383,12 @@ additional functions from our package, we can create this: # (is now y axis because we turned the plot) theme(axis.text.y = element_text(face = "italic"))
To simplify this, we also created the ggplot_rsi() +
ggplot_rsi()
To simplify this, we also created the ggplot_sir() function, which combines almost all above functions:
ggplot_sir()
data_1st %>% group_by(genus) %>% - ggplot_rsi( + ggplot_sir( x = "genus", facet = "antibiotic", breaks = 0:4 * 25, @@ -1410,16 +1410,18 @@ classes) <mic> and <disk>: mic_values <- random_mic(size = 100) mic_values # Class 'mic' -# [1] >=256 0.005 64 0.025 0.5 0.0625 0.025 128 1 >=256 -# [11] >=256 0.025 0.025 0.001 0.25 1 0.005 2 >=256 64 -# [21] 0.01 2 0.001 64 32 2 0.0625 0.005 0.125 0.001 -# [31] 0.5 2 0.25 0.0625 0.125 128 0.0625 0.0625 2 32 -# [41] 128 0.002 >=256 0.25 32 64 0.01 0.025 16 0.005 -# [51] 0.002 0.125 0.5 64 0.002 >=256 0.5 0.5 0.0625 1 -# [61] 4 >=256 0.005 2 128 128 32 0.25 1 4 -# [71] 8 4 0.002 128 0.025 32 8 0.5 0.001 0.025 -# [81] 8 32 1 0.025 0.5 0.001 0.25 0.002 0.002 64 -# [91] 1 0.002 0.025 128 128 2 0.01 0.002 16 4
<mic>
<disk>
mic_values <- random_mic(size = 100) mic_values # Class 'mic' -# [1] >=256 0.005 64 0.025 0.5 0.0625 0.025 128 1 >=256 -# [11] >=256 0.025 0.025 0.001 0.25 1 0.005 2 >=256 64 -# [21] 0.01 2 0.001 64 32 2 0.0625 0.005 0.125 0.001 -# [31] 0.5 2 0.25 0.0625 0.125 128 0.0625 0.0625 2 32 -# [41] 128 0.002 >=256 0.25 32 64 0.01 0.025 16 0.005 -# [51] 0.002 0.125 0.5 64 0.002 >=256 0.5 0.5 0.0625 1 -# [61] 4 >=256 0.005 2 128 128 32 0.25 1 4 -# [71] 8 4 0.002 128 0.025 32 8 0.5 0.001 0.025 -# [81] 8 32 1 0.025 0.5 0.001 0.25 0.002 0.002 64 -# [91] 1 0.002 0.025 128 128 2 0.01 0.002 16 4
# base R: plot(mic_values)
disk_values <- random_disk(size = 100, mo = "E. coli", ab = "cipro") disk_values # Class 'disk' -# [1] 17 29 22 22 20 26 28 25 21 24 26 31 19 28 26 18 21 24 23 23 31 22 26 31 18 -# [26] 20 29 18 21 29 25 31 23 29 26 24 28 28 23 22 21 18 17 20 21 30 29 27 29 26 -# [51] 31 29 22 29 24 19 26 27 30 23 30 22 30 23 31 26 19 23 29 22 24 28 21 17 17 -# [76] 25 24 30 21 30 17 27 19 19 25 23 26 26 22 25 30 18 24 29 20 29 20 24 25 24
# base R: plot(disk_values, mo = "E. coli", ab = "cipro")
S<2PNjj^N-(X5lHgw&Az{XI1y&(3=K zO*-G^B_EWdAFgV37mD2X2UP>&+b0`uyYgzQA^%(_&qiB*Wt7MXspCJkl#e{Gz|lQ4zW&a zJxw^xt^G(bbka?VIPK5>UYP}HHiYJ-^g&_(^Iv*L@{|>$GXu>7)}C!qO&zwPd$6iV zsXbn85pP+UA5$)$>#j2z-%lfB#^P@w#nsBt@bAfD`-E$!M~`2hXtRvT6w}{GJ6<$& zQZKz)yHiOuNv~gfqmT#$0+mV9R`R-<{Xt2othx<|jab={LA2=d?ssHlXZuoKo%P-o z83Z6*TeQYzt@q}2f&Hv#^O*|;MgA|%b;hC)-z9Br;OD9jn3MR)13oObdb1wVjkBGH zRxYR(&c4S5bl|(d=61Bx?Bj#_^{5dyjfo5dQA6p_*DCLn+Rew0(ttx3(BXUfu$Uxc z+QCe_AA?VQgw{2aX7^&&#K}^Vvs%Mz9TUs)tf`(*_;)P@ci|UbBBMqFMbNPRLc73M zb_U3E@QYN{#>KoathO67vhNV^8#$x>T>WO}XvrE#!=-sE^SvSW5!5gR8*V6n`@}PD zO$N7|A6uz4$$SzgDz0L@pnlBi^1s245N-JdCHn+K0 zGJZl1JwO5!u{~>_E@r9$Bsb$2d6KAAUHJnGstAo81y^8f3mCfMxY+ir*Thjk&n5NN znPSJfI7m8b5l9L|z2dc)0_c2@Qs@b4KGz8jHU3PWy+-QYPIaDP$Ur#)68E7XI^KFg z;?9_EtoDaXVP@FZfmLb?eR=NtlrHJ67ygt@4}x4O3c$7;++Aj$2IlDqv(5ib->%jJ z({IAKi}vW#n8q^7@y}L#)Z!VQUTTTGaj=J=wWVZKM%xoA)ECjSJ9YRL&bFT^3rQ z5uYm9438lqhq}X*m15h0qwwAl$04XAwA3mV(w-*YVMf*7mk+Bg!&eHx`)SO9-rXw)u^dkWYm!ULI7C)Ds_oeTYi9!vtHKCT*G zzBcgb6>yYRB6pUiCDXy)e~p_2H4G1!u%TtT4&6h>@r78- z+Rv6bCw9;aKiFA@HsNTiq>q{g^K8YJeCjMM9VdUOZK&}uol@CGHA4fr{ zW!esyp16{Xvs!-_-;OZus67~*%-sShxWEy11!d7oLmNx^5i-}A#~`kY$g%y+JWxu5 z6o%{*xow(`<1B#C{mhB{y4KFFi@o0b#A(Lqn0p|XGrY7ctraP39No~#WtiR6zo9t| zpQF$P`LOe{AAOu+SQCN~EfgIv&PwM+H#?ID6FYHlRp>#;x09_ag+{*2x8dd9a z%Tc>dYwf~vure8GAp*#txy?-!($t>0oYnfE>oMDv*;l;oR-IubLnD;d!bP!E6^GA;Ip zPcGdLlLRAq+7|I>&r&^7_iGpTP)RkKVlZuQe}?SaHx&Fs73Cw?q*A1ybpP*XZ&VFq|skcBf^S>rcXmeW8PkS9Ix-dn*ai(rpgs>Nvu zusR6M`xknjfyxa;l8I$WS@%~@uY4yS%hGN_h8)Y48s<#N2fms<`*NR_L#A{sOS#@f zofhsx2*vn4?pN=-`an==*wul{@b<_l8_#ygnYzBE-`<>7NIdV)P%mp}HYgog5D@ee zyu^*~;bq-NSy7W?O-QXmQWv*B*svof>!519#%ZUeqA<$Fyj{{B>+ORA7Y?wVwaxpu!$+dG~6g@YM+)bmF;Y^Tn|B;Zzr|0$p*aSre zaYdBOwIF@+W*IJvcv~;WRE76@)-QhuOZ-~Uei##%2Xb?1VV2mzBKVl_eCNWa#*)o$ z<39$gt{fgnI+r<>XBe&}zlrfIY1L1~_~r6)0RvcuFzHS*qSTHzF#7|_U&#{;%}T8T z1wh6#e@YCIdCtVPqYP4AhyH^a-zixG;z*Lzi=S3&vn;x~*;6n>t9=}JM5-$+2W%eP zzO=SE+BPzqMbs3a7M)^dwdj{0+SLk|%?v#^%9Y=B;Ipu5L$0cS%yB(N{7{oG)}4Of z)&A4Xo0e)G8F|;CGb7;_18*GnyRJAh;oaNCh~8gOz4zx$3F~2g>4^})K?PpPw|O73 zUeHw_c4ahkJt{(gP=t^K2q9+&Ki_kHbMGH_=FHqPbIzPS1B5qk+3(tGJ};Hx3K%lNTl&IX1L)l6<3Il``?W`N-3HDGs|OMc;-4)aPncO%aT zE}7=_xVg6jo>OOSDyguhP`#A2)1&|U>;GEd|80Tmc}s~VKUe?A;Y6{imLBds_pn9J ze7GTyk^@~>zun*hgTm*c$1w4-1W|h>=RNQ0DqYRL)N()|!_TLu>)8IiV|W@5o7=yi zubUkXX9WeTIqmr`hfDUI0f)-o*%zH=eJ4cQcO5)n0dWIWr>&%I6bA1RElgaC#k7g^ zB=h!Ik}xHho1&B?-x>^;oaaI9Q6cwAjJHUL^jtIQxFvHkF6P^i!x{K%r{3jI6jwz$ z4fS1TzzbCcx03Qn=hH%H5452=AN5V>2g8zSnN+SbnYp#$tre%50P!^$*bfg{cQaRFb1VK-oswd+8uIHyS|CdB zPtF*r)N-cisZ&|tYhYp%?ejPGfg$zCk#D5(xEi^{iNA1tVJ3!=m8K19!JWoe2X z9v^XL9)=5-o6yo^#+c`G{ZS2RB8y99%l#GBzrFO)9!{?7sb@vp-(%jn&lC&GOSd@+ zy5|jE<^n$A?Eiv!eis0625Bpqm01=hdusnEi5lbGl1*zD&rl|9y7>C zP_>5QZku1GqNtCLnCIsXri3iRm985-=~i?X_jxU{nmiu%J68sO^{+(_=cr7FSTnUz zRI~H5!)juO-`rP%$mZ(Rmz-4_g*vNei4EEe;83ycB`TGoA@e91$+=rlPtGrmGL)d6 zj5JyI811Gh+f{zqxf2_w6H(%Go++y-VjZV{^YRO$z9;qa@>86TTk9k)Ulvt`ZJ=gq zf7cQ|rr)>!*Q2iM*_@+=Pm;CxAzj9+LKP_{Ph-8<+FDhSSftb9xx|1!iX*dALX~8< z$gw*CDmDY=?XL&4!s8)vu!Zx0Ll#jE=Eh?|S7| zUY6e|H^-<@51IaQc=Qp|?2@ghT(bM)3N^{M7W3;o-dcFzT^(ziND@jPZtjEc&-9H(B`eB@I*s?yh|*gtT(h7gA?nWStUHEYfj0eyy>q zX^FFHdj1rE6;>t{21EQwP`DeVW!D;`_7D&m*yn~QM;0J7ou-bQSnH*9H#c9U=&`GUj>xiT z)OSA31@^})O|ss$;&7|cx(-MBUD=ncB&Jg&@9!HxauzWv4c@K@7aA)zCCR9B0s?aQLL zz%U+8_&v$;%(3UFr4%%N*rV5`y7uM7DET8sU!^lht5)D4B3HX!-HOgvw;;PBA+PTe z>RNqQP)1(iQBxI+TB4CYrZRuv5!WnTW&YfscH$Kd-*+16#@FWeYsLL1j^lW=rv<0iAo<_Ke458KB z=Y}IL8~{LoePr)Y?0;%HspEV@id)?Q_$Ca_Euo+`>xZ(_Eom&m$9>w^}8 z#_rk&AVbvX1J3k|TGyH%d7`2F2Od^;Q9o;i^f%X`mNd%+^wf61LGdQ!6a?Zxq&2iP zkwsO`$wPhH5f#annM*YV)#qJuy7!qi$Fx`aj5LY#Y=$MRaiA3uJhe71j$dpZ(Bm^3 zyZno3)~o4!4XIsJgxp}fNfHtSZ3lY?tzugs`TOkwq8#2+DLu&q?GW^&5{tc)*orDj z+W7p#DCJUZ(|^XYAn?ZIpgV^A2Q^=D;BFz~n4EAGUHhJ>{QK{Mf&3`+n;&))TWjd8Z()h{o`sUab+{) zvF_(HaK6Cbwjw-Ni?W9+ErjHg;4t*326={6ic(n!%lTPNu{^cA9dGi~Kc@2T1L-lg zSsg*CX;rJsNmPCz_f!p)mIVHwD>lphg2Wb2cAo}%NHQ^ae6Cb>7UF9$fIL^_orh7d zrHUT79_jRyhPI0c-PJd;{mkmvoHr?!`ywIk9AIgozQ<6Q=2ynCG%Twgl7YG;><;PUWH}dJlf6 zf@2xBt0Tzr;34Rr5e>b6!cQy*F({et)XG?Bn_5g$_8SLoVk()c0D~sK|XL8M{sV8bmFb4o%C{9MR zBlUfT>c(^3m)+Abk3>V6zo4mwYqfvEaWC%rn1spEIx338u1`BZYEY~Gs8oiwbfI2tTNN6mN{cC2+#Ouu_#}xcg1j^rivLRb;YPN|2o#ySAjP*}`n`@|{*5pdNzr^Sl+_jQ)PsE5> z7APk)7#CFdp(=nl2lnC_(I2Qry@v}bXm=vJ=-wXo#gg|t;slLX@1|#z4^8Xo2I;3t zZqVIeF28eLqd5297Om`cJl)WpwRi=2_6DtaFzPARmmtZfwJZ z8*N4d8<76N?jKb~oWsNjmsghjF-VEP%&KIaIAzNTsM6}Q5KKVOmNGm@B)45d(L=P7 zEb$=HC4dNjnhn7_J3mVRnm5r1gKAZM6j7B^KTg37j_-{UZ_6Kgx| zYn>-d+~90m)F!SIDhCuEjd*ipGGaUZ_De$2<^o9?v z1sb++0a07Axqgx9Vb#jc)4%0I&fb<&TPb&XT2|@Yd;19{!dX3I?DP1o56@eo8)4-l zm1V2=l#{oiNfo*?8g)XX-iisMg*OvD&M?+HIf(ab&LEAIFC2mHi!wj!LtpL9jA_vL zT7@q<9bV*KgbFOS>-4*M%{R_{>X(fICVJssTdE{|EBNSjOX@69{Mx`U+pI}eUdf&# z;ar>bGZOg)d*lYM-x3Y^5kI*RaThP*N2f%YPFjMI(o{i*2bv7ja#AxgEi$VmK5xT^ zwj;|70LRe-x>=MI0Q7L=isb~E}XsaKw zeKouAkAEEHno4$Ij#B;TN>-;T%$vht=+F>(H(O$ov|1S6#$APZb3l2BB#5FufD=9= zVl_9=jMfGyHjBlmkyo$0n*Ewj`!ql-9I73u7y%dE!) zMoUxIs=8<8V8;BHQhXb`Ykg!kFH5KeZ8c^4i7;U3jeCpFi@j1#GR%@TEwBd$x(RA7 zpJjs7D*WVxry|=`GDXpbNhxr|O_wv}x^mT@?ehfqo|3#&!p{5TBV^&&_6BZP*DD>* ze=uo5T6ORlmGSc+2^AOhTW$9#z7sQoOZF|r&;TeFR9zhuAR3yP3V4@QRaDKtm6T`P zsKtoNsr9O;j4o?Pwz-ic$Rk?=_qwX(@Co?JscW#6tp>iTv*;r=M?QuzT&)cVm#g;5 z35SW!nW8;5Oi6^@wyVP-P8J;+_T~!B7T6V^JA4xU0*yaW0j1rQ@!1xKRbSRcv|FF! z!SoszaPMg61*ceph2-BlX76clWQXnk%xR%B?cP6+-p>9?PF5lpz3!67YbA&;KW7(4 z=|xFZXJ{aTRNXp)_Q(3pC+U=2keqwYPS-QMuR6GHof|3YPG&dNMuigK^(v!G`Q)Q&AsHdpYj#XyVTTy=*ICE+FhqF9jZ zTZDyswirC6>)h3+n?KyVhYIWBXe}O&;ExaceP^aW^tOX8b^NM!9{-C(Wvp z5XflP@=*`><4B^pN%hFPC++}c4x3EkMvJ%Yi|KVvv(YCF5xl3L_=n&59Ian@_xdL^ z%@}R?VnsN&sGJINunRI}og+70dbO3A+nl5Rm`d-mu7`npG6xH*le=fzb1(<^$E+of zua7FIg+w^*dRbT~x6qnSO$8@`mh()Ij zev|t!pdWydeNgeRR_6y^)tfp?l^)u@d-S`)s#FrGU}?QT@`JUi1~{Uj(1ljgz)8-x^A}$nqc4O=4Q31>p>&?^%gV#h{tq`}S;o|Swj zLmDUbc=*$aMx%1O+9B6T=LaFa zAZQ1uc(biEdFs<3FIea?IHEP4u(57A!=Y5{Zv4wx8DeW+S|28(U(%d?2lnRi$-5ux zg+J9rbXsea(uDen`yk=3E^vU%t=5&}xm+4@5Ej6>YuU2gt?!O1(G?h)v1KO>>3+IcrBC{?P?7zIUyuphvyFZTx%-#A>ae& zKtXkST1J8uDm{54eulSJcKk{gB9L^0hTmVWM*MhKu;$mMDhJl`Jd zBGx&sXizj9I4nut_DrO$C{c{&+aAB|Mhfsu3#?0*r-#Lid1Yi+jvAItAzfwH(f4@^ zJfEV%^u|MH|L}a$SqrcO9_ZMbbs##U+79K?xdyBJ9U2hr6e?10=yJ(V04^%$^VvJe zohbrel)LXCw(5NR*j5CFCX$x-l|fS{jie`cyjC%~rs|LZpHQ3jTf%-rLLuR{{H(g| zBky?1xWPy>1?RD(A7syy!f=)uTbvU!6U`0!NU_1nCxxluNH#D3WN2NB9`JwC-9D%6 z^wj)N(&Mjpfn-zAsc7glcam@f3#PHZkU2J#G>ZlxBwpDbR6rDK~Q=wAd zt$?b)>d|hGQ}a}TF*RM*<`L*m%V;^ap3&$uj-dIM3dU|8N^Ne$hpA0%E`<0$p5y-X ztt|yL!Qv_XuUt1a)0pGZH4Cj=GxUx0lI8UWRBp8pJiiS2?yJQ0HG zgup6p#|335+XV#Z9(y%$je5BxJ?zoQ@|CBZf<3rP@CoYndjGq*{qH9yDntIv>(wH; z?u*H1S2!Kl7}FLTj>{f?*_vv*x_w85L_*Hi~a{YF5&6 zzw)sMm0oEF%UKA8G5hak2tYPPTICkQ`=bQi-<-Q5cRqBR;OlJVZQ=6FMF5W1RGJvw zE;@?5Q%<}pC^hGpe=W#SZS3<3`aJc9OLp$PR+#PVhZbqn zkt{0sfqLSEadT2|&_fAoUVb9DcZsXU`ft&jAYn9z;GIjy2nqoh^!n_5i zP+5(4h|?qpiR)}JFZ~@gg_q0*`Eh_1n|YB^PP?ws;hT}DFpcHnVYfV!gXRlE?%7Jw z(!&bpK8MOet+4ctknPzcLu8HUiM;e&9g@H~d5xKa=)g$(L^L--7QKgaYDDCQR`n2AQoFIDRP&T_1u`Wd0@mxzq7X6=SgnVL3W|lgMUktb)|arSsJmSf0dnWs!tB`3{eVcm;$JV_6qRbo8_@zm(Fgv|NHyPoz^xlMP2q|up{ zCN52`*}DrCkDv{`Dhn1Xoz+Q2KHpO$3<8VpGI4T)VZ+3_rfzzmyNw`hITueia{TE2 zBVKGDvvU6AKX;nEQ0iUmi}UCFRp&dqm+9rEOX!Ci2uXjLND2SoGw3k}GRfE0COzJ4 zv;5j9IV&m4XPoL4yXh5I5&5BTLvGm_3ZJT+A`HCwGSq^P*X)&%ilDw$6ommnW5rs!&hQpE+%IIwc>viI|LT zO|n(&+H*O^2t78qF!(z;3T02~c*d(7(3P}Vab`mfmU!Daw6B!-wfMs@h1c|#`#*BI z%7gx6OXZz^>OO$^DNxl&xC+whb~jX;#Ny-B@(hRR2k1|?-^+!W{MxgR!LHOI7qbyq zxq+1ll;^yjZM-Zf$41IIs*mp15^LvA;{2{W@;T!clAHGcd(G>_Mj5zu+t-i9oNTkf zt)o!GUmFf-+{CvN_%3?zE$62SQLl7SvY{(w`h|=1xDF_(C!xi6C$oMJ`eMkog9D|Fg(1b+tVK4pG7( zTFev;7i&t=2ItJ)92?zk_}%g9Tbd}j>3*CdETW)n5mJsfIaKJZZZDIl$Rs%(_oz2e z)Z7uU-e-%n>0C2Q2>UA2RGD&Cse z96MX?Fd`q_bH1kt&)A5uwVcha5I>(1Ss$D7F#Ius*kt)8t`_eud?L2Y; z`uEFX9c#OHrSRRO9~tRz5{4{N{%p$R3720b2BWtb%INxQZ-RbzV1VsIaw-Yf9RD5D z+k8BFYcB*tv_&^nIToGXwZ0bHeYrjed}4i_GXm)TI#obo5I-45dI3>gkXDE7RJwQF z00sZNZ+GPeWu1an^o5Lz+jX8Cg)w+`?s`u6Z*eo{QD8^H&XND4e8fVqDM{NdN? z%366zq`#((%d)@u&cMCsG~4$}8?|0QeX3i-Z`XRU@7Th9)9@&po6zk*@Xf!?fq08G zi?Y+=jnbj%TVxg#9KGVa65~>&r7esf5YKAX>S1BV}m! zP=Td!9O=r^tMSdnM%pBO^ug6@55GL)EQ~m>V4i0kR6=@(>^fA84HaYXhwO)#*s}}| z!Q7qcM%Nz#9iWz#mAybG%<}AhW^Fy#UwXH#t9g5qs(6ah5#;ZRcrU2&5alG6mjHzz(z2R-02_3gE8 zgaj$oyGildL<5I_Z{Hdij92B%M#q*Hf5=!^Gyc!dmdFC@>LQAC_4uQ~Kw?^e8t7dW zd;gG4%&oJ5Nl8gRb(j|2FXqRu2$_3v4+GnaO=goN`^61;{(}l0par3Zp z`8=4cZ1KRx@+O6(v6owL%NAc`>#e;0BZ6sbX^9RagwXsDKb~;RxR@C3ZCd{oQ;PaF zlNSn^lWNBfcIx+PDl?-Z2sjOLsMa~GxqXN!Ek?F!Rv7+41KH#}+xL#3M614syVnJ* zQT?ld+-n{0-}1xVe6BhmY3f<^Ztq$%3e7XaY;sz*@%^-H(_16K&6@G4a&AbyAeig9 zSMfjs^Zo&1aKv1Fx7VwiDKfTKhKlW(ZHG=@saT)SFSgR^w1k|x-g9g}-0xmClTi8K zDn-q0FiY3NYo5#0&VR(YYoLf965O6|n%VHh%6m+b#?XV zi#iy1;TIrER6UU@)Dua-Elkps{Jt%Nm}>E=U#yDta=v{wz7LGB_GqmPKwvEI*|SHV_xN$up)%*L)Ls?ZczV>Rn~%)3@;j~3 zUs!JIke1;(k1vMJAj7t%K6y7Ys@l=wNmXo`UR}x3j__#p4}sIu)90A>rT&YfJgg_z z`Vfb-E5blXR4!m$MUIY~xsk316Gy93WW%7=F1sg_>vGVW#8EcAIDalrmS*1uw`t8ZzGKH2 zs;wF#M>+jw`+nv*cO?tVcUJ>QRRIWA*EaA`8);GD*8jD75PerO?%J%oZL!MdU%JR% zllL60U=-E-iY9e-cAgrcU-23rz7OCt+katrz~Axti+FJB zYfDMycIZ=Fks$>6qp#J;3DUc!t6&qMOTQEJmzb+vvi66~ZGNs!cWDcTZLOf5AKOj#8y3pC^v3G^JZysq06wi6H|O(-%K+(jyY#%d+a{SmyIT_lXAZ+qNAhhVna@bAu&FoZLd$$ zR=Uyh%o4Z$_b@@w;hVqsbeh0pxNK`4IA*+m|K4x8nYn$lH)PHxyFOZne)Sw9G*)$q zri79?eqyaMHAa;Sa=2xm8)TnBU|`@RkN&k%b3#*ycOr&`?>aaGO0*>>w{48+#P3w6jT|i_L@MX zi!=h}BYrQm4hUYJY}Bu3P{(K)u&~fC{h?~`?2Oab9zRe}zAy;DGi5T2$%v`GKO*-4 zg!gV3bBlaHws~tc=#7GR)5NXeIckd@fRxcd(9OvUp*Tav(0=`kN**2c`I{y_%4#sI%(b<%4q+XVrPZeT4^KnRe6qS5Q3H)<;{xi z7+L_}tF;E&pkd=?C_SY)Om-wUbh|RGxTfan06=m3`hex_cLe`zYcmPnyhO~_RxrN* z3_$kb?sZQvKFGH|y^656{-f@~H1ISIZEfxP-V9i?9gAPYBtNJ{?1i1TEc?wxtW^eK z7B&t%wUG$yGP^-0LIJs^`^H+V4qkz52ZPRi08a~bc6XnH0#_Vx`1b0gYxh4FZwjlw zGH`VK`3-pPS|R8;?KSx1d-D?|>$^gsb+rt$Ujr%trbE+%mLl}8t$g)o=)>V1Sr_kq z3!WcOhQk(kxgx83)0AnWTmnjoV`II*2c}hmKG z0zA;Tun17dS{4q-!9CSIHR8O~vD>$ACjr&EVSvigORGzcg`nhz9h6B0<)LvH zO%Wpy#xP9vog?8G;QV%__eaW_VIWM+S!k3H6B82#wXFs|g^+jcy+#2iXFTW-P2m@G zYRskT-u?D!q`It3u30Zl$pad)W@r!l;yl2u?6*A0IOgi`bqF{j0YO_*u|(Mgue5rF zjwa>_k+|kTnB);S=4^QM7<5|L#hf4Iyxkj%gl6WO&k-EVn<$yO+4rg3iofH|oL2C< zKMCAZkrn;*b`;{OQeiI$ga9a?69dvyBX!n2Dj0xQhp=2UfR`AOGVNTGe z{*LT)#5{u~(Bo7?Hn&2hrr#?1jA{-5YO0?m5(APbHytjWs`Nf{5MCO%8#rlQJ;n%L zxAN^YkrleBd@?q}*?^fBL9o+j!r68o&j1vif9lHp8tEi|4okDM-c1`bOSEwxeXMNL z`ui>*_GqJEs62x*LhgV3`0@4lYD!{aflfsu7}0(M36qn+5deQ(BXY6F)q-$fTyX2l zB#Nk1Kbr2eGS1fX1dl~R6@#b?Q&K?t1942Z2B0394q__*X^t$eX%39zgf)um+MiEI zM(@uzz@fkAn-??da1Xy5YJ^azW8V1{6%`Q~S3XL%*G2pfza2cH!GC17bdTR(?g4(T z6b7hGL)x+o6J%)EU&jsGdZ1Hww|m;pil?v56ss!JzL_5cG(!D;b&zfE1&S`Tx%C1+ zmqxw+V%$koSTjmM9h8%kWu{>IIuk{2K!&1x{RpuBMspJk1x{co&j;rt*0-!^%(*v zJ*^zFxh5aBIhP&6%7#YUjs@H=V*CB&%MHA>0?1}N+4ext9sn_0e|(N;PQC0gB+p`J zd-Ce)S=6ORzuULfaPjvDYoy}lp%e@tF>!uevZffL#JJq0$KOH|H}w8?Ob^P$&TjIP zcfHXOrwO$*|1p~Gi1d^5|*qyI0i$!v4Y95z69gpAm5y73VUpbCKp>#8X zS343?uXrL{=D+9aX6a{4&Vb%4FN}FNs|14-DlJniI3UR>>(5kf6S zt!2X)bA%K-#9X$hsOag_z%zRUYX<>x4uWLnp?Y2$SLP;wv$o9$Zq>)5^k5dtY<(B0MmsOb);ZO5@VS~`vrVcMwX8&squ#niP1DN2?3z0(G`l=ov0%*w?&{65$uC6gE@Yy9r-`T#KPxJCMkb0ZdvbietV>3@ zTVGL)(gF?s^BGP6W5D$}`tuYG zSRJrD=~`@Ap5+O+vgxgo-+Vsu+G_j4Zmz2i>Xw!xOB#S?4i`75*-6zOEmqa*)Pre2 zQvw-L3->`5`P~GRnZCQ(drJ%+%1Ftg)I3!4$MemBJbngHce6A+V8RbC1iv?4voaI| zI71DN`16^gcxkM9IRhwi+$TYt5y%Aqyk|D`X_576syY4-@kP~SH1*LDllypri9n3YkN}!&2GwdMye(Z-JNK7 zeh#1F4i*1M0Ku`rX95G^1vWIjPy~LUsr%;AxdhC#v-Mj0090NQB-K{VPcBJ50o<&+ zgfAxagUnF^E$6H?$Vw|H>nl?*yJ|1n72w@hA8)SBx$0%8GP-XCF1C8#XBD>Y0;XWt zKmNG~zxjeo0HF_pgC|y%>wQ5pW6`@@Tt7Y75QGjM6vkl3_CJ6B&LF1@Wc4v1Noljj zvh<>`M(o>4R!+>yjvT+|7>*^407kvdl3^taJ|~p43_(S#1WN-+0%VU)+RnVs10=SF2gt+z0gZ zxi_ArOih6^&g(sg@`MH*rx4VZHgrHsIG&jUfAR)lI#!hy|}j_X$o1 z(7%3&*gUuxnFavrAJ@qc+5{}kZ^)ASE%-II2pH9hwn-``{Cpd#xxXkwE89yJoz5M^ zRy;4a$4Ci(OAcgWk7P`eqkXZ