read-flu-data.R

library(tidyverse)
library(RSocrata)

fetch_flu <- function(temporal_resolution = "weekly", 
                      na.rm = TRUE){

    locations <- readr::read_csv(file = "https://raw.githubusercontent.com/cdcepi/FluSight-forecast-hub/main/auxiliary-data/locations.csv", col_select= 1:4)
    
    #read data from healthdata.gov, 
    health_data = RSocrata::read.socrata(url = "https://healthdata.gov/resource/g62h-syeh.json") %>% 
        # dplyr::filter(date >= as.Date("2022-02-02")) # filtering for when flu reporting became mandatory
        dplyr::filter(date >= as.Date("2021-10-01")) # filtering for the full previous season
    
    #remove  VI and AS as they are not included for FluSight, keep only necessary vars and add epiweek and epiyear 
    recent_data = health_data %>% 
        dplyr::filter(!state %in% c("VI", "AS")) %>% 
        dplyr::select(state, date, previous_day_admission_influenza_confirmed) %>% 
        dplyr::rename("value" = "previous_day_admission_influenza_confirmed") %>% 
        dplyr::mutate(date = as.Date(date), 
                      value = as.numeric(value),
                      epiweek = lubridate::epiweek(date), 
                      epiyear = lubridate::epiyear(date))
    
    #summarize US Flu 
    us_data = recent_data %>% dplyr::group_by(date, epiweek, epiyear) %>% 
        dplyr::summarise(value = sum(value, na.rm = na.rm)) %>% 
        dplyr::mutate(state = "US") %>% 
        dplyr::ungroup()
    
    #bind state and US data
    full_data = rbind(recent_data, us_data) %>% 
        dplyr::left_join(., locations, by = join_by("state" == "abbreviation"))
    
    #convert counts to weekly and calculates weekly rate 
    weeklydat = full_data %>% 
        dplyr::group_by(state,epiweek,epiyear, location, location_name, population) %>% 
        dplyr::summarise(value = sum(value, na.rm = na.rm), date = max(date), num_days = n()) %>% 
        dplyr::ungroup() %>% 
        dplyr::filter(num_days == 7L) %>% 
        dplyr::select(-num_days, -epiweek, -epiyear) %>% 
        dplyr::mutate(weekly_rate = (value*100000)/population )
    
    #if daily data is ever wanted, this returns correct final data
    if(temporal_resolution == "weekly"){
        final_dat = weeklydat %>% 
            dplyr::select(date, location, location_name, population, value, weekly_rate) %>% 
            dplyr::arrange(desc(date))
    } else{
        final_dat = full_data 
    }
    return(final_dat)
}