ClaimAudit

byBhupendra Kumar

Agentic AI based application using langgraph and fastapi , and reactJS ui as follows- -> dashboard homepage , have upload excel file button , on uploading that ill will save all rows in a table of sqlite db [AppData.db] to table allAuditClaims, [each row have claimID, rxClaimNumber, and questions] -> we can have more than 1 batch of claims in form of excel , so i also have a table claimsBatch , for each excelsheet at one row in claimsBatch table, -> already had a table rule , with some sot of rules with some category, -> have one claims tabel with all claims details ,(have rxclaimnumber also) display data their

LandingLoginRules
Landing

Comments (0)

No comments yet. Be the first!

System Requirements

System Requirement Document
Page 1 of 4

ClaimAudit

Introduction

The ClaimAudit project is designed to streamline the insurance claim auditing process by automating the application of rules to claims using advanced AI techniques. This document outlines the system requirements for the ClaimAudit application, which aims to enhance efficiency and accuracy in auditing insurance claims.

System Overview

ClaimAudit is an AI-based application that leverages multi-agent systems and LLMs to process and categorize insurance claims. The system is designed to handle the ingestion of Excel files containing claim details and questions, apply predefined rules to these claims, and generate human-like explanations for auditing results. The application utilizes FastAPI for backend operations, ReactJS for the user interface, and SQLite for data storage.

Functional Requirements

  • As a User, I should be able to upload Excel files containing claim details and questions.
  • As a System, I should store uploaded data in the SQLite database, categorizing it by batches.
  • As a System, I should identify rule categories from questions using an LLM.
  • As a System, I should apply appropriate rules to each claim based on identified categories, focusing solely on internal auditing processes.
  • As a System, I should evaluate claims using an LLM to generate explanations for rule compliance.
  • As a User, I should be able to view comprehensive reports of auditing results on the dashboard.
Page 2 of 4

User Personas

  • Internal Auditor: Responsible for reviewing the auditing results and ensuring compliance with internal standards, including managing Excel file imports.
  • System Administrator: Manages system configurations and oversees data integrity.

Visuals Colors and Theme

  • primary: #2A9D8F (Teal)
  • primary_light: #A8DADC (Light Teal)
  • secondary: #E76F51 (Coral)
  • accent: #F4A261 (Sandy Orange)
  • highlight: #E9C46A (Golden Yellow)
  • bg: #F1FAEE (Off White)
  • surface: rgba(233, 245, 248, 0.8)
  • text: #264653 (Dark Slate)
  • text_muted: #6D6875 (Muted Purple)
  • border: rgba(38, 70, 83, 0.2)

Signature Design Concept

The ClaimAudit homepage will feature an interactive "Claim Galaxy" concept. Users will navigate through a 3D galaxy where each star represents a claim batch. Clicking on a star will zoom into a cluster of claims, allowing users to explore individual claims as planets. Each planet will display claim details and auditing results. The galaxy will be rendered using @react-three/fiber and @react-three/drei, providing a dynamic and engaging user experience. Hovering over stars will highlight connections between claims and their respective rules, creating an immersive exploration of the auditing process.

Page 3 of 4

Interaction Model & Motion Direction

The landing page will utilize a "parallax" interaction model, creating a sense of depth as users scroll through the galaxy of claims. Decorative layers will move at different speeds, enhancing the storytelling aspect of the application. Internal pages, such as the dashboard and reports, will adopt a "static" interaction model to prioritize readability and data presentation.

Non-Functional Requirements

  • The system should handle concurrent uploads and processing of multiple Excel files efficiently.
  • The application must ensure data security and integrity during storage and processing.
  • The user interface should be responsive and accessible across various devices and screen sizes.

Tech Stack

  • Frontend: ReactJS
  • Backend: FastAPI
  • Database: SQLite
  • AI Models: LLM for category identification and claim evaluation
  • AI Tools: Langchain for LLM routing
  • Local Orchestration: Docker
  • Server-side Orchestration: Kubernetes
Page 4 of 4

Assumptions and Constraints

  • The system assumes that Excel files are formatted consistently with required claim details and questions.
  • The application is constrained to operate within the capabilities of the selected AI models and tools.
  • The system must comply with relevant data protection regulations and standards.

Glossary

  • LLM: Large Language Model, used for natural language processing tasks.
  • Multi-Agent System: A system architecture where multiple agents perform specific tasks to achieve a common goal.
  • FastAPI: A modern web framework for building APIs with Python.
  • ReactJS: A JavaScript library for building user interfaces.
  • SQLite: A lightweight, disk-based database engine.
Landing design preview
Landing: View Overview
Login: Sign In
Dashboard: View Uploads
Dashboard: Upload Excel
Dashboard: Confirm Batch
Dashboard: Track Status
Dashboard: View Results