MoveApps Survival Estimator (R-SDK)

Overview

This application is a specialized frontend designed for wildlife tracking analysis. It automates Kaplan–Meier survival analysis, providing a standardized, GUI-driven interface to visualize survival curves and life tables from tracking data.

While designed to simulate a node within the MoveApps R-SDK workflow, this web version runs entirely in the browser using React, simulating the statistical logic typically performed by R packages (survival, move2).

Note: This version is a static demo. The "Biological Interpretation" feature serves a pre-written example report instead of connecting to a live AI API, ensuring the app is safe to run without configuration.

Features

• Data Ingestion: Upload CSV files simulating move2 tracking data.
• Survival Analysis: Client-side calculation of Kaplan-Meier estimates (Survival Probability, Standard Error, 95% CI).
• Interactive Visualization: Dynamic survival curves with censoring marks and confidence intervals.
• Risk Table: "Number at risk" tabulation aligned with time steps.
• Interpretation Example: Displays a mock publication-quality ecological insight report (simulating Google Gemini output).
• Configuration: Flexible controls for time units, event columns (mortality), and grouping variables (e.g., Sex, Age).

Technical Stack

• Frontend: React 18, Tailwind CSS, Lucide React (Icons).
• Visualization: Recharts.
• Markdown Rendering: react-markdown.
• Build-less Execution: Uses ES Modules via importmap for immediate browser execution.

Setup & Usage

1. Running: Serve the directory using a simple HTTP server (e.g., python3 -m http.server, npx serve, or VS Code Live Server).
2. Data: The app loads with mock data. Upload a CSV with columns for duration/time, status (binary), and grouping factors to test with your own data.

---

Development History & Prompts

Initial Request: Build Design Specification

User Prompt:

Build Design: MoveApps Survival Estimator (R-SDK) ... (See full prompt in history) ...

Implementation: The system was implemented as a React Web App simulating the R logic.

• survivalLogic.ts: Recreated survfit logic (KM estimator, Greenwood's formula).
• SurvivalChart.tsx: Used Recharts to mimic ggsurvfit aesthetics.

Refinement 1: Fix Arithmetic Error

Issue: Sort comparison in RiskTable.tsx caused a TypeScript arithmetic error due to implicit types. Fix: Explicitly typed parameters in the sort function (a: number, b: number) => a - b.

Refinement 2: UI/UX Polish for AI Report

User Prompt:

can we format pretty this Biological Interpretation (Gemini)?

Implementation:

• Added react-markdown to the importmap in index.html.
• Updated App.tsx to render the Gemini response as Markdown with custom styling.

Refinement 3: Remove API Dependency

User Prompt:

the ai-features are connected to my account? If so just create a png example and remove code that would utilize my API.

Implementation:

• Removed GoogleGenAI import and API key logic from services/geminiService.ts.
• Replaced dynamic generation with a static Markdown string to mock the feature safely.
• Removed @google/genai from index.html.