Skip to main content
After creating a case and uploading documents, MedMitra’s AI agents generate comprehensive medical insights. This guide shows you how to access and understand these insights.

​
Accessing Case Insights

1

Navigate to Cases

From the dashboard, click on Cases in the navigation menu to view all your patient cases.
2

Check Case Status

Cases display status badges:
  • Processing: AI analysis is still running
  • Completed: Insights are ready to view
  • Failed: An error occurred (rare)
Processing typically takes 30 seconds to 2 minutes depending on the number and size of documents.
3

Open Case Details

Click on any completed case row to view the full case details and AI insights.The case details page opens with comprehensive information displayed in an intuitive layout.

​
Understanding the Case Details Page

The case details page is organized into several key sections:

​
Header Section

At the top of the page, you’ll find:
  • Patient name and case title
  • Status badge showing current case state
  • Overall AI confidence score as a percentage
  • Last updated timestamp
  • Action buttons: Share, Edit, Export PDF

​
AI Insights Overview (Top Cards)

Three prominent cards provide the most critical information at a glance:

Primary Diagnosis

Contains:
  • Primary diagnosis name
  • ICD-10 code
  • Confidence score
  • Brief description
Example:
Diagnosis: Type 2 Diabetes Mellitus
ICD-10: E11.9
Confidence: 94%

Key Findings

Contains:
  • Top 3 most important findings
  • Bullet-point format
  • Extracted from all documents
Example:
  • Elevated fasting glucose (142 mg/dL)
  • HbA1c at 7.8%
  • Microalbuminuria detected

AI Summary

Contains:
  • Comprehensive case overview
  • Integration of all data sources
  • Clinical context
  • Summary confidence score

​
SOAP Notes Section

The Medical Insights Agent generates structured SOAP notes following the standard clinical format:
Patient-reported information:
  • Chief complaint
  • Symptoms description
  • Patient’s perspective
  • Symptom timeline
Displayed in a blue-highlighted card with relevant information extracted from your case summary.
SOAP notes are generated with their own confidence score, displayed at the top of the section.

​
Supporting Evidence

The AI provides transparent reasoning by showing supporting evidence:
  • Each piece of evidence is displayed in a separate card
  • Evidence is extracted directly from uploaded documents
  • Links clinical findings to the proposed diagnosis
  • Helps validate AI conclusions
Example Evidence Items:
“Fasting glucose level of 142 mg/dL (reference: 70-100) indicates impaired glucose metabolism consistent with diabetes diagnosis.”
“Chest X-ray shows clear lung fields with no evidence of active disease, ruling out pulmonary complications.”
“Patient reports polyuria and polydipsia for 3 months, classic symptoms of hyperglycemia.”

​
Patient Information Panel

The right sidebar displays:

​
Demographics

  • Patient name
  • Age
  • Gender
  • Doctor’s notes (your original case summary)

​
Test Results Summary

Laboratory Summary:
  • Concise summary of all lab findings
  • Highlights abnormal values
  • Notes significant trends
Radiology Summary:
  • Overview of imaging findings
  • Key impressions from all images
  • Anatomical areas examined

​
Uploaded Files Section

View all documents associated with the case:
1

File List

Each file shows:
  • File name and type icon
  • File size
  • Category badge (lab or radiology)
  • Download button
2

Individual AI Analysis

Files that have been analyzed by AI show additional information:
  • Findings: Specific observations from this document
  • Impressions: Clinical conclusions
  • Confidence Score: How confident the AI is about this analysis
3

Download Files

Click the download button to retrieve the original document from secure storage.

​
Confidence Scores Explained

MedMitra displays multiple confidence scores to help you assess AI reliability:

Overall Confidence

Range: 0-100%Aggregate confidence across all analysesInterpretation:
  • 80-100%: High confidence
  • 60-79%: Moderate confidence
  • Below 60%: Low confidence, verify carefully

Diagnosis Confidence

Range: 0-100%Confidence in the primary diagnosisFactors:
  • Quality of supporting evidence
  • Consistency across documents
  • Strength of clinical indicators

SOAP Confidence

Range: 0-100%Confidence in SOAP note generationFactors:
  • Completeness of input data
  • Clarity of case summary
  • Document quality

​
Color-Coded Confidence Indicators

Confidence scores are color-coded for quick assessment:
Score RangeColorBadgeInterpretation
80-100%Green✅High confidence - reliable insights
60-79%Yellow⚠️Moderate confidence - verify findings
0-59%Red❌Low confidence - manual review required
Always use clinical judgment when reviewing AI insights, especially for cases with lower confidence scores.

​
Interpreting AI Insights

Follow these guidelines when reviewing AI-generated insights:
1

Start with Confidence Scores

Check the overall and diagnosis confidence scores first. This gives you an immediate sense of how reliable the AI analysis is.
2

Review Primary Diagnosis

  • Evaluate the proposed diagnosis
  • Check the ICD-10 code
  • Read the supporting evidence
  • Consider differential diagnoses
3

Examine SOAP Notes

  • Verify subjective information matches your notes
  • Confirm objective data from lab reports and images
  • Review the clinical assessment
  • Evaluate the proposed plan
4

Cross-Reference Supporting Evidence

  • Check that evidence matches actual document content
  • Verify numerical values and units
  • Ensure interpretations are clinically sound
5

Review Individual Documents

  • Check AI analysis for each uploaded file
  • Verify findings against the original documents
  • Look for any missed or misinterpreted information
6

Make Clinical Decisions

Use AI insights as a decision support tool, not a replacement for clinical judgment:
  • Integrate AI findings with your clinical expertise
  • Consider patient context not captured in documents
  • Order additional tests if needed
  • Consult specialists when appropriate

​
Example: Reading a Complete Case

Let’s walk through a sample case:
Header Information:
  • Patient: John Doe, 45 years old, Male
  • Status: Completed
  • Overall Confidence: 92%
  • Last Updated: March 4, 2024, 10:23 AM
Primary Diagnosis Card:
  • Diagnosis: Type 2 Diabetes Mellitus
  • ICD-10: E11.9
  • Confidence: 94%
  • Description: “Chronic metabolic disorder characterized by hyperglycemia due to insulin resistance”
Key Findings:
  1. Elevated fasting glucose at 142 mg/dL (reference: 70-100 mg/dL)
  2. HbA1c level of 7.8%, indicating poor glycemic control over 3 months
  3. Microalbuminuria detected, suggesting early diabetic nephropathy
SOAP Notes:
  • S: Patient reports increased thirst, frequent urination for 3 months, fatigue
  • O: Fasting glucose 142 mg/dL, HbA1c 7.8%, BMI 32, BP 138/88
  • A: Type 2 Diabetes Mellitus with early nephropathy, uncontrolled
  • P: Start metformin 500mg BID, dietary counseling, recheck HbA1c in 3 months
Clinical Interpretation: The high confidence scores (92% overall, 94% diagnosis) suggest the AI has strong supporting evidence. The diagnosis is well-supported by multiple abnormal values (glucose, HbA1c, albuminuria) and classic symptoms. The SOAP notes integrate the clinical presentation with objective findings logically. This case would be suitable for initiating the AI-recommended treatment plan, with appropriate patient counseling and follow-up.

​
Actions You Can Take

From the case details page, you can:

Export PDF

Generate a comprehensive PDF report including all insights, documents, and analysis

Edit Case

Update patient information, add notes, or modify the case summary

Export Case

Download case details and AI insights for external review

​
Data Structure

For developers, here’s the AI insights data structure: 
interface AIInsights {
  comprehensive_summary: string;
  key_findings: string[];
  patient_context: {
    name: string;
    age: number;
    gender: string;
  };
  doctor_notes: string;
  lab_summary: string;
  radiology_summary: string;
  case_summary_confidence_score: number;
  soap_subjective: string;
  soap_objective: string;
  soap_assessment: string;
  soap_plan: string;
  soap_confidence_score: number;
  primary_diagnosis: string;
  icd_code: string;
  diagnosis_description: string;
  diagnosis_confidence_score: number;
  supporting_evidence: string[];
  overall_confidence_score: number;
}
The case details component is located at frontend/app/cases/[caseid]/page.tsx:42-490.

​
Best Practices

Recommended workflow:
  1. Quick scan: Check confidence scores and primary diagnosis
  2. Deep dive: Review SOAP notes section by section
  3. Verification: Cross-reference with actual documents
  4. Decision: Make clinical decisions incorporating AI insights
  5. Documentation: Export or save for patient records
High confidence scenarios:
  • All confidence scores above 80%
  • Strong supporting evidence from multiple sources
  • Clear, consistent findings across documents
  • Standard clinical presentations
Exercise caution when:
  • Confidence scores below 60%
  • Conflicting information in documents
  • Complex or unusual presentations
  • Limited uploaded documents
If confidence scores are low:
  1. Review uploaded documents for quality issues
  2. Check if important documents are missing
  3. Ensure case summary is comprehensive
  4. Consider uploading additional test results
  5. Manually review all AI interpretations
  6. Consult with specialists if needed
  7. Order additional diagnostic tests

​
Troubleshooting

Possible causes:
  • Case still processing (check status badge)
  • Processing failed (status shows ‘failed’)
  • No documents uploaded
Solutions:
  • Wait for processing to complete
  • Check case status
  • Refresh the page
  • Contact support if status is ‘failed’
Possible causes:
  • Poor quality documents
  • Incomplete case summary
  • Insufficient medical data
  • Ambiguous clinical presentation
Solutions:
  • Upload higher quality documents
  • Add more detailed case summary
  • Upload additional test results
  • Include more clinical context
Possible causes:
  • Document parsing errors
  • AI misinterpretation
  • Corrupted files
Solutions:
  • Verify document content manually
  • Re-upload problematic documents
  • Report discrepancies to support
  • Use clinical judgment to override AI

​
Privacy and Security

Your case insights are protected:
  • Only you can view your cases and insights
  • All data transmission is encrypted
  • Insights are stored securely in Supabase
  • Access logs are maintained for auditing
  • HIPAA compliance measures in place
Never share your account credentials. Use the built-in sharing features (when available) to collaborate with colleagues.

​
Next Steps

Understand SOAP Notes

Deep dive into how AI generates SOAP notes

Learn About Diagnosis Support

Explore how AI suggests diagnoses with ICD codes

AI Agents Overview

Learn about the AI agents powering insights

API Reference

Access insights programmatically via API
Pro tip: Use keyboard shortcuts to navigate between sections quickly. The case details page is fully keyboard-accessible!

Build docs developers (and LLMs) love

Get started for freeTalk to us