Full-text Review Explorer

A lightweight, self-hosted systematic review research tool. Provides thematic analysis, paper reviews, AI-powered search, and classification of research papers.

Comparison with Covidence

Feature	Covidence	Full-text Review Explorer
Hosting	Cloud SaaS	Local, fully client-side
Import formats	RIS, PubMed, CSV, reference managers	CSV (Covidence export), RIS
Duplicate detection	Automatic + manual resolution	Automatic (title + DOI matching)
Screening stages	Title/abstract, full-text (configurable)	Full-text review with status tracking
Review statuses	Include, exclude, maybe	Included, excluded, unreviewed
Exclusion reasons	Per-paper, configurable	Per-paper with presets + custom reasons
Thematic classification	Tags only	Hierarchical themes + flat tags
Tag management	Basic labels	CRUD with global registry, batch operations
Inline editing	Limited	All metadata fields editable inline
Search	Basic keyword	Title, abstract, author search
AI-powered chat	None	OpenAI-compatible chat panel with paper search
Paper discovery via AI	None	LLM references papers as clickable cards
Batch operations	Bulk actions	Multi-select with batch tag add/remove
Conflict resolution	Dual independent reviewers	Single reviewer
Data export	CSV, PRISMA flow diagram	Data in localStorage (no built-in export)
Collaboration	Multi-user with roles	Single user, local only
Pricing	Subscription-based	Free, open-source
Data privacy	Stored on Covidence servers	Stays in your browser

Features

Review Workflow

• Full-text review - Track papers through included, excluded, and unreviewed statuses with one-click cycling.
• Exclusion reasons - Preset reasons (Not LLM-based MAS, Duplicate, Full text unavailable, Not primary research) plus free-text custom reasons.
• Move back to screening - Remove papers from the review entirely with a confirmation dialog.

Thematic Classification

• Hierarchical themes - 10 predefined research themes with support for sub-themes. Full CRUD: create, rename, delete, and assign to papers.
• Tags - Flat tag system with a global registry. Create, assign, and remove tags from papers.
• Batch operations - Multi-select papers with Shift+click range selection, then batch add or remove tags.

AI-Powered Chat

• Research chat panel - Slide-in panel with an OpenAI-compatible LLM (DeepSeek, OpenAI, Z.AI, local models, etc.).
• Paper search via natural language - Ask questions like "Find papers about oncology" and get clickable paper cards in responses.
• Streaming responses - Tokens appear in real-time as the LLM generates them.
• Configurable - Set Base URL, API Key, and Model via .env file or the in-app settings panel.

Paper Management

• Inline editing - Edit any paper field (abstract, journal, year, authors, DOI, notes) directly in the detail modal.
• Paper detail modal - Full view with executive summary, key findings, system architecture, and single-LLM comparison sections.
• Clickable DOIs - Direct links to published papers via doi.org.

Data Import

• CSV import - Import from Covidence CSV exports with automatic column mapping.
• RIS import - Import from RIS reference files with standard field mapping.
• Duplicate detection - Papers matched by title (case-insensitive) or DOI are automatically skipped.
• Auto-theme creation - Themes referenced in imported files are created automatically.

Persistence

• All data stored in localStorage - papers, themes, and tags persist across sessions.
• Seed data - Ships with a built-in dataset of 76 research papers and 10 predefined themes.
• Automatic migration - Legacy data formats are upgraded seamlessly on load.

Tech Stack

Technology	Purpose
React 19	UI framework
Vite 6	Build tool and dev server
Tailwind CSS 4	Styling
Framer Motion	Animations
Lucide React	Icons
OpenAI SDK	AI chat integration
PapaParse	CSV parsing
TypeScript	Type safety

Getting Started

Prerequisites

• Node.js v18+
• npm

Setup

# Clone the repository
git clone https://github.com/LaansDole/mas-health-explore.git
cd mas-health-explore

# Install dependencies
npm install

Chat API Configuration

Copy the example env file and add your API credentials:

cp .env.example .env

Edit .env with your OpenAI-compatible API details:

CHAT_BASE_URL=https://api.deepseek.com
CHAT_API_KEY=your-api-key-here
CHAT_MODEL=deepseek-chat

Any OpenAI-compatible provider works:

Provider	Base URL	Example Model
DeepSeek	https://api.deepseek.com	deepseek-chat
Z.AI (GLM)	https://api.z.ai/api/paas/v4/	glm-5.1
OpenAI	https://api.openai.com/v1	gpt-4o
Local (Ollama)	http://localhost:11434/v1	llama3

You can also configure these in the chat panel settings UI at runtime. UI overrides are saved in localStorage and take precedence over .env values.

Run

npm run dev

Open http://localhost:3000 in your browser.

Other Commands

npm run build       # Production build
npm run preview     # Preview production build
npm run lint        # TypeScript type checking (tsc --noEmit)
npm run clean       # Remove build artifacts