https://github.com/sepmein/stp_snt_2023
The analysis coding for São Tomé and Príncipe subnational tailoring analysis 2023
https://github.com/sepmein/stp_snt_2023
Last synced: 4 months ago
JSON representation
The analysis coding for São Tomé and Príncipe subnational tailoring analysis 2023
- Host: GitHub
- URL: https://github.com/sepmein/stp_snt_2023
- Owner: sepmein
- Created: 2023-10-02T09:03:57.000Z (about 2 years ago)
- Default Branch: master
- Last Pushed: 2023-11-08T06:01:34.000Z (almost 2 years ago)
- Last Synced: 2025-01-22T20:30:10.253Z (9 months ago)
- Language: R
- Size: 38.7 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# STP SNT analysis 2023
The analysis coding for São Tomé and Príncipe subnational tailoring analysis 2023.
## Project Management
### Track Progress
Clickup was used to track the progress: [ClickUp STP Analysis Tracker](https://app.clickup.com/9010032161/v/li/900202049906).
GitHub Repository:
[STP Analysis GitHub Repository](https://github.com/sepmein/stp_snt_2023).
[Code Issues and Improvements](https://github.com/sepmein/stp_snt_2023/issues).
## Replicate of this analysis
This analysis will use R `targets` package to ensure reproducibility.
### Packages to be installed
```r
install.packages(c("targets", "tarchetypes"))
```
## Data Management
Database relationships
```mermaid
---
title: STP SNT 2023
---
erDiagram
adm1["adm1 distrito"] {
uuid id PK, UK
string adm1 PK, UK
}
adm2["adm2 localidade"] {
uuid id PK, UK
string adm1 PK
string adm2 PK
}
hf["US"] {
uuid id PK, UK
string adm1 PK
string adm2 PK
string hf PK
string type
string owner
float lat
float long
string comment
}
hf_raw {
string adm1 PK, FK
string adm2 PK, FK
string hf PK, FK, UK
string type
string owner
float lat
float long
string comment
}
pop_adm1 {
string adm1 PK, FK
int year PK
string sex
string age_group
int pop
}
pop_adm2 {
string adm1 PK, FK
string adm2 PK, FK, UK
int year PK
string sex
int pop
}
routine {
string adm1 PK, FK
string hf PK, FK
string month PK
int year PK
string group PK
int susp
int test
int conf
int negative
int treat
int maladm
int severe
int alldth
int maldth
}
elimination["Casos por localidade"] {
int no
string no_order
string adm1 PK, FK
string adm2 PK, FK, UK
int cases_2018
int cases_2019
int cases_2020
int cases_2021
int cases_2022
string elimination
}
passive_cases_2022["Passiva Casos"] {
string month PK
string hf PK, FK
string adm1 PK, FK
string adm2 PK, FK
int age
string sex
}
reative_cases_2022["Vig. reactiva"] {
string month PK
string hf PK, FK
string adm1 PK, FK
string adm2 PK, FK
int age
string sex
}
intervention["Intervenções de rotina"] {
string adm1 PK, FK
string hf PK, FK
int year PK
string month PK
int anc1
int anc2
int anc3
int anc4
int ipt1
int ipt2
int ipt3
int ipt4
int itn_p
int itn_v
int stock
}
irs["PID"] {
string adm1 PK, FK
string adm2 PK, FK
int year PK
int cycle_1
int cycle_2
int treated
int closed
int refused
int not_treated
int irs_planned
float cover
int uninhabited
int other
int total
int pop
int insecticide
}
itn_campaign["MTILD (Massa)"] {
string adm1 PK, FK
int itn
}
itn_routine {
string adm1 PK, FK
int year PK
int itn_u5
int int_p
int itn_other
int itn
}
lsm["gerenciamento de fonte de larvas"] {
string adm1 PK, FK
string adm2 PK, FK
float area
float scanned
int water_bodies
int sprayed
int issued
int unsprayed
date last_spray
int sampled
int positive
float average_anopheles
}
vector["Ab. vectores"] {
string adm1 PK, FK
string adm2 PK, FK
int year PK
string month PK
int mos_adult_inside
int mos_adult_outside
int lav_1
int lav_2
int lav_3
int lav_4
int pupa
}
resistance["resistência a inseticidas"] {
string adm1 PK, FK
string adm2 PK, FK
int year PK
string month PK
string insecticide
float mortality
string resistance
}
adm1 ||--o| adm2: contains
adm2 ||--o| hf: contains
hf_raw ||--|| hf: match
adm1 ||--o| pop_adm1: has
adm1 ||--o| itn_campaign: has
adm1 ||--o| itn_routine: has
adm2 ||--o| pop_adm2: has
adm2 ||--o| elimination: has
adm2 ||--o| irs: has
adm2 ||--o| lsm: has
adm2 ||--o| vector: has
adm2 ||--o| resistance: has
hf ||--o| passive_cases_2022: has
hf ||--o| reative_cases_2022: has
hf ||--o| intervention: has
hf ||--o| routine: has
```
### Match adm1, adm2 and hf
Approaches used:
```mermaid
flowchart LR
A[adm1 shapefile] -->|Get unique adm1| B(Unique adm1 in shapefile)
C[routine db] -->|Get unique adm1| D(Unique adm1 in routine db)
B --> E{fuzzy join by string distance}
D --> E
E --> F{Remove the exact same match}
F --> G{Manually check and fix}
G --> H{Match function for routine db}
H --> I[Aligned adm1 for shapefile and routine]
```
```mermaid
flowchart LR
A[Fix ADM1] --> B(Fix ADM2)
B --> C(Fix HF)
C
```
#### adm1
In adm1_from_itn_route, there is an extra adm1 called "CNE", which might stands for National Center for Endemic Diseases (CNE) (Sao Tome and Principe). Not sure why, set CNE to NA firstly
#### hf
1. hf in routine database
mismatch after fuzzy matching
- POSTO DE QUARTEL
- P. SAÚDE DE V. D´AMÉRICA
- POSTO DE QUARTEL
- HOSPITAL DOUTOR QUARESMA DIAS DA GRAÇA
- POSTO DE NOVA APOSTÓLICA
- POSTO DE ÁGUA ARROZ
- POSTO SAÚDE DE PINHEIRA ROÇA
- P. S. DE RIBEIRA AFONSO should it be Ribeira Afonso I or Ribeira Afonso II(Sta Infancia)?
- P. S. IRMÃS C. DE R. AFONSO should it be Ribeira Afonso I or Ribeira Afonso II(Sta Infancia)?
### GIS
#### Shapefiles
##### Align shapefiles
> Problem
![align_shapefiles]()
![align_shapefiles]()
Green shapefile was provided by NMCP with detailed village level of information. Orange one was from WHO GISHUB. The unaligned borders between the two was noticed.
A aligned shapefile with adm1/adm2/adm3 level of information should be created before the analysis.
> Solution
I used the crs reprojection with the geo-referencing in QGIS to manually aligned the two islands.
##### Fix same localites names with same and different adm1
During the process, I found some of the localites has the exact same name, some of them located in the same district, others don't.
The shapefile country shared with us has duplicated records with the exact same adm1 and adm2. There are records with identical adm1 and adm2 values, some of which are adjacent to each other, while others are not. Attached is the list detailing these duplicates.
I have conducted a preliminary analysis and before proceeding further, I need confirmation from the country office on the following points:
- Merging Adjacent Duplicates:
Is it permissible to merge the records that are adjacent to each other, considering they share the same adm1 and adm2 values?
- Review of Non-Adjacent Duplicates:
For the records that are not adjacent but have identical adm1 and adm2 values, country need to review and confirm if these instances are due to naming errors or map inaccuracies? Additionally, it would be helpful to receive guidance on how to proceed with these records.
##### the creation of the Health facility shapefile, the decision of Unit of analysis
> Problem
We have a shapefile with adm2 level. I need to create a hf level shapefile. There are two approaches:
1. The first approach
- For each adm2, I calculated the closest HF.
- Assign a HF value to each adm2
- For each HF group, union all the adm2s
The problem is that it is that for each HF group, there will be multiple adm1s. Country will be deciding the unit of analysis.
2. The second approach
- Create adm1 shp based on adm2 shp
- Within each adm1, for each adm2, calculate the closest HF
- Assign a HF value to each adm2
- For each HF group, union all the adm2s
##### HFs without any data
1. Uba Budo in Cantagalo.
2. Hospital Principe in Principe.
##### HFs with incorrect GPS
1. Porto Alegre POINT (6.63361 0.03556)
2. Santana POINT (6.74718 0.25813)
Resides outside of the adm1.