Maternal Health Risk

An exploratory dashboard on maternal health risk, built entirely in R Markdown.

Live site · Report · Risk Explorer

---

Overview

This repository is a static R Markdown website (rmarkdown::render_site) built on the Kaggle Maternal Health Risk Data Set — 1,014 pregnancies monitored across hospitals, community clinics, and maternal health cares via IoT-based risk sensors, each labeled with a clinician-assigned risk level (low / mid / high) from six vitals: age, systolic and diastolic blood pressure, blood glucose, body temperature, and heart rate.

The site has three pages: a landing page, a full exploratory analysis, and an interactive tool that lets a visitor enter their own vitals and see how they compare against the cohort. It's designed to knit and deploy the way a course or personal data-science project typically does — .Rmd → rmarkdown::render_site() → static HTML → GitHub Pages — with no backend server.

Pages

Page	File	Contents
Home	index.Rmd	Dataset overview, variable definitions, cohort summary stats
Report	report.Rmd	Distributions by risk level, correlation structure, a data-quality note, and a baseline multinomial logistic classifier
Risk Explorer	explore.Rmd	Interactive k-NN comparison tool — enter your own vitals and see which risk band they resemble most

All analysis code in report.Rmd is folded by default (code_folding: hide); click Code above any table or figure to expand the exact R that produced it.

Project structure

.
├── _site.yml              # site-wide config: navbar, theme, output options
├── index.Rmd               # → index.html
├── report.Rmd               # → report.html
├── explore.Rmd               # → explore.html
├── assets/
│   ├── custom.css          # theme overrides on top of the flatly bootswatch theme
│   └── header.html         # <head> includes (webfont)
└── data/
    └── Maternal Health Risk Data Set.csv

Getting started

Prerequisites

• R (≥ 4.1) and RStudio

• The following packages:

  install.packages(c(
    "tidyverse", "plotly", "nnet", "broom",
    "jsonlite", "janitor", "scales"
  ))

Data path

Every .Rmd loads the dataset with:

read_csv("data/Maternal Health Risk Data Set.csv")

This assumes the CSV lives in the data/ folder shipped with this repo. If you relocate it, that's the one line to update — in index.Rmd, report.Rmd, and explore.Rmd. Nothing else in the project depends on file location.

Build

Open the project folder in RStudio (it will recognize _site.yml as a website project) and either:

• click Build → Build Website in the Build pane, or
• run rmarkdown::render_site() from the console

Both knit all three .Rmd files and write index.html, report.html, and explore.html into the project root. Open index.html in a browser to preview locally.

Deploying to GitHub Pages

1. Create a repository — <your-username>.github.io if this should be your root personal site, or any other name for a project page.
2. Commit and push the whole project, including the rendered .html files, _site.yml, assets/, and data/ — GitHub Pages serves the built HTML, not the .Rmd source, so the .html files need to be in the repo.
3. In Settings → Pages, set the source to the branch you pushed to (root directory).
4. Update the GitHub icon link under navbar: right: in _site.yml to point at this repo.

Re-run Build Website and push again after every edit to an .Rmd or _site.yml — GitHub Pages only reflects what's already been knit.

How the risk explorer works

explore.Rmd computes cohort means and standard deviations in R and embeds the cohort (plus those summary stats) as JSON directly into the page via a knitr chunk. A small vanilla-JS widget then standardizes whatever a visitor enters, finds the 15 nearest cohort records by Euclidean distance, and has them vote on a risk level — the same standardization approach used in report.Rmd's correlation and modeling sections. Everything runs client-side; nothing entered in the form is sent anywhere, which is also what makes it work on GitHub Pages without a live R session.

Tech

R Markdown · rmarkdown::render_site() · tidyverse · plotly · nnet · vanilla JS · Chart.js · GitHub Pages

Data source & disclaimer

Data: Maternal Health Risk Data Set, Kaggle, collected via IoT-based risk monitoring across hospitals, community clinics, and maternal health cares.

This project is exploratory and educational. The risk explorer describes which historical records a set of inputs most resembles — it is not a diagnostic tool and was not validated for clinical use.