walkingModule.R

library("synapseClient")
synapseLogin()

library("plyr")
library("dplyr")
library("ggplot2")
library("jsonlite")
library("parallel")
library("tidyr")
library("lubridate")
library("stringr")
library("doMC")
library("data.table")

#devtools::install_github("Sage-Bionetworks/mpowertools")
library("mpowertools")

WALK_ACTIVITY_TABLE_SYNID = 'syn7222425'
actv_walk_syntable <- synTableQuery(paste0('SELECT * FROM ', WALK_ACTIVITY_TABLE_SYNID))
actv_walk <- actv_walk_syntable@values
actv_walk$idx <- rownames(actv_walk)

attr(actv_walk$createdOn, 'tzone') <- 'UTC'
actv_walk <- actv_walk %>% mutate( createdOn = as.character(createdOn))


# # select users 
# selected_users_walk <-   fread(synGet("syn8113982")@filePath, header=T, data.table=F) %>%
#   .$healthCode %>% as.character()
# actv_walk <- actv_walk %>% dplyr::filter(healthCode %in% selected_users_walk)
# to enable download of only filtered data
# actv_walk_syntable@values <- actv_walk




######################
# Download JSON Files
######################
#download outbound walking json file
outbound_Walking_json_files <- synDownloadTableColumns(actv_walk_syntable, "deviceMotion_walking_outbound.json.items")
outbound_Walking_json_files <- data.frame(outbound_Walking_json_fileId =names(outbound_Walking_json_files),
                                          outbound_Walking_json_file = as.character(outbound_Walking_json_files))

actv_walk  <- merge(actv_walk,outbound_Walking_json_files, 
                      by.x="deviceMotion_walking_outbound.json.items", 
                      by.y="outbound_Walking_json_fileId", all=T)

#download pedometer data
outbound_pedometer_json_files <- synDownloadTableColumns(actv_walk_syntable, "pedometer_walking_outbound.json.items")
outbound_pedometer_json_files <- data.frame(outbound_pedometer_json_fileId =names(outbound_pedometer_json_files),
                                            outbound_pedometer_json_file = as.character(outbound_pedometer_json_files))

actv_walk <- merge(actv_walk, outbound_pedometer_json_files,
                   by.x="pedometer_walking_outbound.json.items",
                   by.y="outbound_pedometer_json_fileId")
actv_walk <- actv_walk %>% mutate(outbound_pedometer_json_file=as.character(outbound_pedometer_json_file))


actv_walk <- actv_walk %>% dplyr::select(-accel_walking_outbound.json.items,
                                         -accel_walking_return.json.items,
                                         -accel_walking_rest.json.items,
                                         -deviceMotion_walking_rest.json.items,
                                         -pedometer_walking_return.json.items,
                                         -deviceMotion_walking_return.json.items
                                         )

actv_walk$outbound_Walking_json_file <- as.character(actv_walk$outbound_Walking_json_file )
actv_walk$outbound_pedometer_json_file <- as.character(actv_walk$outbound_pedometer_json_file )

#############
# Feature Extraction
##############
registerDoMC(detectCores()-2)

#extract pedometere features
pedoFeatures <- ddply(.data=actv_walk, 
                      .variables=colnames(actv_walk), 
                      .parallel=T,
                      .fun = function(row) { 
                        tryCatch({ mpowertools::getPedometerFeatures(row$outbound_pedometer_json_file)},
                                 error = function(err){
                                   print(paste0('Unable to process ', row$outbound_pedometer_json_file))
                                   stop(err) })  
                      })
pedoFeatures['error_pedometer_features'] = pedoFeatures$error
pedoFeatures$error <- NULL
pedoFeatures$outbound_pedometer_json_file <- NULL


# 
# f <- synDownloadTableFile("syn7222425", "35780_4", "deviceMotion_walking_outbound.json.items")
# r <- mpowertools::getWalkFeatures(f)
# 
# 
# f <- synDownloadTableFile("syn7222425", "20315_2", "deviceMotion_walking_outbound.json.items")
# mpowertools::getWalkFeatures(f)
# 
# dat <- jsonlite::fromJSON(f)
# dat <- mpowertools:::ShapeGaitData(dat)
# x <- dat$x
# y <- dat$y
# z <- dat$z
# aa <- sqrt(x^2 + y^2 + z^2)
# aj <- sqrt(diff(x)^2 + diff(y)^2 + diff(z)^2)
# outX <- mpowertools:::SingleAxisFeatures(x, dat$timestamp, varName = "X")
# 
# mpowertools:::
# View(x)
# 

#extract walking features
walkFeatures <- ddply(.data=actv_walk, 
                      .variables=colnames(actv_walk), 
                      .parallel=T,
                      .fun = function(row) { 
                        tryCatch({ 
                          res <- mpowertools::getWalkFeatures(row$outbound_Walking_json_file)
                         },
                          error = function(err){
                          print(paste0('Unable to process ', row$outbound_Walking_json_file))
                          stop(err) })  
                      })


# walkFeatures
walkFeatures['error_walking_features'] = walkFeatures$error
walkFeatures$error <- NULL
walkFeatures$pedoJsonPath <- NULL
walkFeatures$walkingJsonPath <- NULL
walkFeatures <- merge(pedoFeatures, walkFeatures, all.x=T, all.y=T)


#############
# Final Data
#############
OUTPUT_FOLDER_ID <- "syn7231638"
OUTPUT_FILE <- "WalkFeatures.tsv"
write.table(walkFeatures, OUTPUT_FILE, sep="\t", row.names=F, quote=F, na="")
synStore(File(OUTPUT_FILE, parentId=OUTPUT_FOLDER_ID),
         used=WALK_ACTIVITY_TABLE_SYNID,
         executed="https://github.com/Sage-Bionetworks/mPowerAnalysis/blob/master/featureExtraction/restModule.R")
unlink(OUTPUT_FILE)