Skip to main content

​
Overview

After completing a surgical simulation, Justina’s AI system analyzes your performance and provides detailed feedback. The platform tracks every movement, event, and outcome to generate comprehensive reports.

​
How Results are Generated

1

Simulation Completion

When you click “TERMINAR CIRUGÍA”, a FINISH event is sent via WebSocket.
2

Data Persistence

The backend saves your complete surgery session to the database, including:
  • All movement coordinates
  • Surgical events (tumor touches, hemorrhages)
  • Start and end timestamps
  • Session duration
public void endSurgery() {
    this.endTime = LocalDateTime.now();
    this.durationInSeconds = Duration.between(
        startTime, 
        endTime
    ).getSeconds();
}
3

AI Notification

The backend notifies the AI system via dedicated WebSocket channel:
{
  "event": "NEW_SURGERY",
  "surgeryId": "550e8400-e29b-41d4-a716-446655440000"
}
4

AI Analysis

The AI system retrieves the trajectory, analyzes performance, and submits a score and feedback.
5

Results Display

The frontend polls for results and displays them automatically once available.

​
Viewing Results During Simulation

​
In-Scene Results Panel

After completing a simulation, results appear automatically in a blue overlay panel: 
function mostrarAnalisisUI(score: number, feedback: string) {
  if (analysisPanel && scoreText && feedbackText) {
    scoreText.text = `PUNTUACIÓN IA: ${score.toFixed(1)}/100`;
    feedbackText.text = `FEEDBACK: ${feedback}`;
    analysisPanel.isVisible = true;
  }
}
Panel Contents:
  • Title: “RESULTADO DE IA JUSTINA”
  • Score: Numerical score out of 100
  • Feedback: Detailed AI-generated commentary
  • Close Button: Dismiss the panel
The results panel appears on the left side of the simulation screen. If results don’t appear immediately, the system is waiting for AI analysis (polling every 5 seconds).

​
Automatic Polling

The frontend automatically checks for AI results: 
function iniciarPollingAnalisis(surgeryId: string) {
  console.log("Starting polling for AI analysis...");

  pollingInterval = setInterval(async () => {
    const data = await consultarTrayectoria(surgeryId);

    if (data && data.score !== null && data.feedback !== null) {
      console.log("AI Analysis received!");
      mostrarAnalisisUI(data.score, data.feedback);
      clearInterval(pollingInterval);
    } else {
      console.log("Waiting for AI analysis...");
    }
  }, 5000); // Every 5 seconds
}

​
Accessing Historical Results

​
Using the Reports Module

From the dashboard, access historical simulation data:
1

Navigate to Reports

Click the “Reportes” card on the dashboard.
{
  id: "reports",
  title: "Reportes",
  description: "Visualiza y analiza los resultados de simulaciones anteriores",
  icon: FileText,
  color: "green",
  available: true
}
2

Browse Simulations

View a list of your completed surgical simulations with basic metadata.
3

Select a Surgery

Click on a specific simulation to view detailed results.

​
Trajectory API Endpoint

Programmatically retrieve simulation results: 
GET /api/v1/surgeries/{surgeryId}/trajectory
Authorization: Bearer {your-jwt-token}
Response: 
{
  "surgeryId": "550e8400-e29b-41d4-a716-446655440000",
  "startTime": "2026-03-05T10:30:00",
  "endTime": "2026-03-05T10:42:15",
  "movements": [
    {
      "coordinates": [2.34, 1.56, 0.89],
      "event": "START",
      "timestamp": 1737849000000
    },
    {
      "coordinates": [2.41, 1.60, 0.92],
      "event": "TUMOR_TOUCH",
      "timestamp": 1737849015000
    },
    {
      "coordinates": [3.12, 1.45, 1.10],
      "event": "HEMORRHAGE",
      "timestamp": 1737849045000
    },
    {
      "coordinates": [2.89, 1.58, 0.95],
      "event": "FINISH",
      "timestamp": 1737849735000
    }
  ],
  "score": 72.5,
  "feedback": "Good tumor removal technique. However, accidental artery contact reduced overall score. Focus on precision near vascular structures."
}

​
Backend Service Logic

The trajectory retrieval includes permission validation: 
public TrajectoryDTO getSurgeryTrajectory(
    UUID surgeryId, 
    UUID authenticatedSurgeonId, 
    String role
) {
    // Find surgery or throw 404
    SurgerySession session = surgeryRepository.findById(surgeryId)
        .orElseThrow(() -> new SurgeryNotFoundException(
            "La cirugía con id " + surgeryId + " no existe."
        ));

    // Validate permissions (surgeon can only see own data, AI sees all)
    if (!session.getSurgeonId().equals(authenticatedSurgeonId) 
        && !"ROLE_AI".equals(role)) {
        throw new ForbiddenActionException(
            "No tienes permiso para acceder a esta cirugía."
        );
    }

    // Map to DTO
    return new TrajectoryDTO(
        session.getId(),
        session.getStartTime(),
        session.getEndTime(),
        session.getTrajectory(),
        session.getScore(),
        session.getFeedback()
    );
}

​
Understanding Your Score

​
Score Components

Your AI-generated score is based on multiple factors:

Tumor Removal

Completeness and efficiency of tumor fragment removal. More complete removal = higher score.

Precision

Movement smoothness and accuracy. Erratic movements reduce the score.

Safety

Avoidance of critical structures (arteries). Hemorrhage events significantly reduce score.

Time Efficiency

Procedure duration relative to optimal time. Faster (without sacrificing safety) is better.

​
Score Ranges

Score RangePerformance LevelDescription
90-100ExcellentOutstanding performance with minimal errors
75-89GoodSolid technique with minor improvements needed
60-74SatisfactoryAcceptable but requires practice in specific areas
40-59Needs ImprovementSignificant errors; review fundamentals
0-39PoorCritical mistakes; extensive training recommended

​
AI Feedback Analysis

The AI system provides textual feedback addressing:
  1. Strengths: What you did well
  2. Weaknesses: Areas requiring improvement
  3. Critical Events: Hemorrhages or other major issues
  4. Recommendations: Specific training suggestions
Example Feedback: 
Good tumor removal technique with 92% completion rate. 
However, accidental artery contact at timestamp 1737849045000 
reduced overall score. Focus on improving precision near 
vascular structures. Consider reducing instrument speed when 
approaching critical anatomy.

​
Data Structure

​
SurgerySession Model

The complete surgery data structure: 
public class SurgerySession {
    private UUID id;
    private UUID surgeonId;
    private List<Movement> trajectory;
    private LocalDateTime startTime;
    private LocalDateTime endTime;
    private Long durationInSeconds;
    private Double score;           // AI-assigned score (0-100)
    private String feedback;        // AI-generated feedback

    public void updateAnalysis(Double score, String feedback) {
        this.score = score;
        this.feedback = feedback;
    }
}

​
Movement Record

public record Movement(
    double[] coordinates,  // [x, y, z]
    SurgeryEvent event,    // START, TUMOR_TOUCH, HEMORRHAGE, FINISH
    long timestamp         // Unix timestamp
) {}

​
How AI Submits Analysis

​
AI Analysis Endpoint

The AI system uses this endpoint to submit results: 
POST /api/v1/surgeries/{surgeryId}/analysis
Authorization: Bearer {ai-system-token}
Content-Type: application/json

{
  "score": 72.5,
  "feedback": "Good tumor removal technique. However, accidental artery contact reduced overall score."
}

​
Backend Processing

public void saveAiAnalysis(UUID surgeryId, AnalysisDTO analysis) {
    // Find surgery session
    SurgerySession session = surgeryRepository.findById(surgeryId)
        .orElseThrow(() -> new SurgeryNotFoundException(
            "La cirugía con id " + surgeryId + " no existe."
        ));

    // Update with AI results
    session.updateAnalysis(analysis.score(), analysis.feedback());

    // Persist to database
    surgeryRepository.save(session);
}

​
Analytics Module

For aggregate performance tracking:
1

Navigate to Analytics

Click the “Análisis de Datos” card on the dashboard.
{
  id: "analytics",
  title: "Análisis de Datos",
  description: "Métricas y estadísticas de rendimiento de los robots",
  icon: BarChart3,
  color: "purple",
  available: true
}
2

View Performance Trends

Access visualizations of:
  • Average scores over time
  • Improvement trajectories
  • Common error patterns
  • Procedure duration trends

​
Troubleshooting

​
Results Not Appearing

Issue: The results panel doesn’t show after finishing surgery. Solutions:
  • Wait up to 30 seconds for AI analysis to complete
  • Check browser console for WebSocket errors
  • Verify your internet connection is stable
  • Refresh the page and check Reports module

​
Permission Denied Error

Issue: “No tienes permiso para acceder a esta cirugía” Cause: You’re trying to access another surgeon’s simulation data. Solution: Only access simulations you personally completed. AI-role accounts can access all data.

​
Missing Score or Feedback

Issue: Trajectory returned but score and feedback are null. Cause: AI analysis not yet completed. Solution: The frontend will continue polling. If it persists beyond 2 minutes, contact support.

​
Best Practices

  • Review results immediately after each simulation while the experience is fresh
  • Track your progress by comparing scores across multiple sessions
  • Focus on specific weaknesses highlighted in AI feedback
  • Practice regularly to see measurable improvement
  • Analyze trajectories to understand movement patterns

​
Next Steps

Dashboard Overview

Learn about all dashboard modules

Run New Simulation

Practice with another surgical procedure

Build docs developers (and LLMs) love

Get started for freeTalk to us