Cash Leakage Analysis

Identify patterns of revenue loss across denials with AI-powered batch analysis.

Overview

Cash leakage occurs when denials are left unworked or not appealed when viable. This workflow helps identify where money is being lost and provides actionable insights.

Tools Used

Tool	Purpose
batch_classify_denials	Analyze multiple denials, group by category/payer/code

Example: Analyze Seeded Denials

The database contains 4 seeded denials. Let's analyze them for patterns.

User prompt:

Analyze all our denials and identify where we're losing money.

What happens:

The batch_classify_denials tool groups the seeded denials:

{
  "total_denials": 4,
  "total_amount": 1265.0,
  "groups": [
    {
      "group_key": "AUTHORIZATION",
      "count": 1,
      "total_amount": 500.0,
      "percentage": 39.5,
      "avg_amount": 500.0,
      "top_codes": ["CO-197"]
    },
    {
      "group_key": "MEDICAL_NECESSITY",
      "count": 1,
      "total_amount": 500.0,
      "percentage": 39.5,
      "avg_amount": 500.0,
      "top_codes": ["CO-50"]
    },
    {
      "group_key": "CODING_ERROR",
      "count": 1,
      "total_amount": 250.0,
      "percentage": 19.8,
      "avg_amount": 250.0,
      "top_codes": ["CO-4"]
    },
    {
      "group_key": "PATIENT_RESPONSIBILITY",
      "count": 1,
      "total_amount": 15.0,
      "percentage": 1.2,
      "avg_amount": 15.0,
      "top_codes": ["PR-1"]
    }
  ],
  "insights": [
    "39% of denials ($500) are AUTHORIZATION issues - implement pre-service auth checks",
    "39% of denials ($500) are MEDICAL_NECESSITY - ensure clinical documentation supports services",
    "20% of denials ($250) are CODING_ERROR - pre-submission validation can prevent these"
  ]
}

Expected Response:

Analysis of 4 denials totaling $1,265:

Category	Count	Amount	%	Action
Authorization	1	$500	40%	Appeal (CO-197 on CLM-2024-001235)
Medical Necessity	1	$500	40%	Appeal with clinical documentation
Coding Error	1	$250	20%	Already rebilled (CO-4)
Patient Responsibility	1	$15	1%	Written off per policy

Key Findings:

• 80% of denial dollars ($1,000) are appealable
• The CO-4 coding error was preventable with modifier validation
• $15 PR-1 correctly written off (below $25 threshold)

Seeded Data Summary

Denial	Claim	Amount	Status	Resolution
CO-197	CLM-2024-001235	$500	Appeal submitted	Pending
CO-50	CLM-2024-001235	$500	Appeal pending	In progress
CO-4	CLM-2024-001237	$250	Rebilled	Paid $200
PR-1	CLM-2024-001236	$15	Written off	Below threshold

Workflow Diagram

Analytics Features

• Grouping - By category, payer, or denial code
• Percentage breakdowns - See where money is concentrated
• Actionable insights - Specific recommendations per pattern
• Concentration risk - Identifies if one payer/category dominates

Next Steps

• Denial Triage - Act on identified denials
• Coding Validation - Prevent future coding errors
• Appeals Management - Track appeal outcomes