R语言的meta分析-HIV相关性分析内含数据集和教程.zip

# load pre-processed data #data <- read.csv('studies_R.csv') data <- readRDS('studies_R.rds') # output: # table of studies that match the search criteria: author, year, title # series of bar charts showing % of studies that address key populations: # MSM, PWID, Pregnant women, TB, Infants, Children, Orphans, Street children, Adolescents, CSW, Disabled, Prisoners, Migrants # scatter plot of study subjects (N) vs year, colored by cascade step # map of number of studies by country # word clouds: # title # setting # inclusion/excludion criteria # intervention # comparator # comments # filter by: # Study design # Country # Year of study # Study size # Key populations -- check boxes, if unchecked then either-or, if checked then require 'yes' # Cascade step(s) targeted -- check boxes (or if want to allow excluded -- radio buttons Targeted, Excluded, All studies) # Mediators (aka Behavioral Target) -- pull-down menu: All, Not reported, Reported (any type), or behavioral_target strings # Measure -- check boxes for Implementation Measures # Target (imp_target -- implementationt target) - check boxes for capability, opportunity, motivation # Outcomes (implementationt outcomes) - check boxes for capability (imp_outcome_c), opportunity (imp_outcome_o), motivation (imp_outcome_m) library(shiny) library(rworldmap) library(countrycode) library(wordcloud) library(RColorBrewer) library(tm) library(rmeta) world_country_names <- countrycode::countrycode_data$country.name world_country_regex <- countrycode::countrycode_data$regex world_country_codes <- countrycode::countrycode_data$iso3c # the country regex for Korea and Vietnam are missing, add them: world_country_regex[world_country_names == "Korea"] <- "korea" world_country_regex[world_country_names == "Republic of Vietnam"] <- "vietnam" # Define a server for the Shiny app shinyServer(function(input, output) { output$wordclouds <- renderPlot({ source('filterData.R', local=TRUE) if (nrow(data) != 0) { cloud_palette <- brewer.pal(8,"Dark2") par(mfrow=c(2,3)) # Setting wordcloud(paste(data$setting, collapse=" "), rot.per=.15, colors=cloud_palette) # Study Objective wordcloud(paste(data$objective, collapse=" "), colors=cloud_palette) # Inclusion/Exclusion Criteria wordcloud(paste(data$iecriteria, collapse=" "), colors=cloud_palette) # Intervention wordcloud(paste(data$intervention, collapse=" "), colors=cloud_palette) # Alternatively, for just the name of the intervention, not free text field: # wordcloud(paste(data$interv_name_text, collapse=" ")) # Stages of Care Cascade Targeted -- sparse # wordcloud(paste(data$cas_comments, collapse=" ")) # Implementation Actor wordcloud(paste(data$imp_action_desc, collapse=" "), colors=cloud_palette) # Implementation Measures - a bit sparse #wordcloud(paste(data$imp_meas_comment, collapse=" ")) # Implementation Target - a bit sparse #wordcloud(paste(data$imp_target_comment, collapse=" ")) # Implementation Outcome - a bit sparse #wordcloud(paste(data$imp_outcome_comment, collapse=" ")) # Implementation Measures, Targets, and Outcomes wordcloud(paste(c(as.character(data$imp_target_comment),as.character(data$imp_meas_comment), as.character(data$imp_outcome_comment)), collapse=" "), colors=cloud_palette) # Behavioral target -- sparse and full of long words -- not using # wordcloud(paste(data$behav_target_desc, collapse=" ")) } }) output$worldmap <- renderPlot({ source('filterData.R', local=TRUE) # Make sure we have data present before drawing the map if (nrow(data) != 0) { count_country <- function(country_name){length(grep(country_name,data$country, perl=TRUE, ignore.case = TRUE))} country_count <- sapply(world_country_regex, count_country) # add Jamaica that was misspelled Jamacia country_count[world_country_names == "Jamaica"] <- count_country("Jamacia") + country_count[world_country_names == "Jamaica"] # inspect country counts # rbind(world_country_names,unname(country_count)) country_data = data.frame(countrycode=world_country_codes, value=country_count) n = joinCountryData2Map(country_data, nameJoinColumn = "countrycode") mapCountryData(n, nameColumnToPlot = "value", addLegend=TRUE, missingCountryCol = "grey", mapTitle = 'Number of Studies per Country', catMethod="fixedWidth", colourPalette = brewer.pal(7,'YlGnBu')) } }) output$africaMap <- renderPlot({ source('filterData.R', local=TRUE) # Make sure we have data present before drawing the map if (nrow(data) != 0) { count_country <- function(country_name){length(grep(country_name,data$country, perl=TRUE, ignore.case = TRUE))} country_count <- sapply(world_country_regex, count_country) # add Jamaica that was misspelled Jamacia country_count[world_country_names == "Jamaica"] <- count_country("Jamacia") + country_count[world_country_names == "Jamaica"] # inspect country counts # rbind(world_country_names,unname(country_count)) country_data = data.frame(countrycode=world_country_codes, value=country_count) n = joinCountryData2Map(country_data, nameJoinColumn = "countrycode") mapCountryData(n, nameColumnToPlot = "value", addLegend=TRUE, missingCountryCol = "grey", mapTitle = "Number of Studies per Country, Africa Region", mapRegion = 'Africa', catMethod="fixedWidth", colourPalette = brewer.pal(7,'YlGnBu')) } }) ##### output$forest <- renderPlot({ # filter studies based on user inputs source('filterData.R', local=TRUE) par(mfrow = c(4,2),cex.axis=1, cex.main=1.5,cex=1) # Testing fe<-meta.MH (cas_testing_ttotal, cas_testing_ctotal, cas_testing_tcases, cas_testing_ccases, data=data, names=author_year, na.action = na.omit) #re<-meta.DSL (cas_testing_ttotal, cas_testing_ctotal, cas_testing_tcases, cas_testing_ccases, data=data, names=author_year, na.action = na.omit) # Fail gracefully if fe doesnt contain anything if (length(fe$names) != 0 && fe$logOR != Inf && fe$logOR != -Inf){ metaplot(fe$logOR, fe$selogOR, nn=fe$selogOR^-2,fe$names, summn=fe$logMH,sumse=fe$selogMH, sumnn=fe$selogMH^-2, logeffect=TRUE, colors=meta.colors(box = "royalblue", lines="darkblue", summary="orange"),ylab="") title(main='Testing') } # Linkage fe<-meta.MH (cas_linkage_ttotal, cas_linkage_ctotal, cas_linkage_tcases, cas_linkage_ccases, data=data, names=author_year, na.action = na.omit) if (length(fe$names) != 0 && fe$logOR != Inf && fe$logOR != -Inf){ metaplot(fe$logOR, fe$selogOR, nn=fe$selogOR^-2,fe$names, summn=fe$logMH,sumse=fe$selogMH, sumnn=fe$selogMH^-2, logeffect=TRUE, colors=meta.colors(box = "royalblue", lines="darkblue", summary="orange"),ylab="") title(main='Linkage') } # Retention fe<-meta.MH (cas_retne_ttotal, cas_retne_ctotal, cas_retne_tcases, cas_retne_ccases, data=data, names=author_year, na.action = na.omit) if (length(fe$names) != 0 && fe$logOR != Inf && fe$logOR != -Inf){ metaplot(fe$logOR, fe$selogOR, nn=fe$selogOR^-2,fe$names, summn=fe$logMH,sumse=fe$selogMH, sumnn=fe$selogMH^-2, logeffect=TRUE, colors=meta.colors(box = "royalblue", lines="darkblue", summary="orange"),ylab="") title(main='Retention') } # Staging fe<-