Clinical troubleshooting

Introduction

Healthcare providers may encounter various clinical scenarios when using the Legit.Health Plus device that require troubleshooting guidance. This section addresses common clinical challenges, interpretation questions, and performance concerns to help healthcare professionals optimize their use of the device in clinical practice. The device is designed to support clinical decision-making by providing diagnostic support through ICD-11 class recognition and quantifying clinical signs, but proper understanding of its outputs and limitations is essential for effective clinical use.

Common clinical issues

This section describes the most frequently encountered clinical challenges when using the device and provides practical guidance for resolution. If you continue to experience issues after following these recommendations, please refer to the support section.

Unexpected or inconsistent diagnostic suggestions

Symptoms

• Device outputs differ significantly from clinical expectation
• Top-5 diagnostic suggestions don't include the suspected condition
• Inconsistent results when analyzing similar lesions
• Low confidence scores across all diagnostic suggestions

Possible causes

• Poor image quality affecting algorithm performance
• Multiple skin conditions present in the same image
• Skin condition outside the device's trained scope
• Inadequate lighting or focus during image capture
• The lesion is not the main element in the image

Solutions

• Review the How to take pictures guidelines to ensure optimal image quality
• Ensure the lesion occupies the majority of the image frame and is well-focused
• Use flash when lighting conditions are suboptimal
• Take multiple images of the same lesion from different angles
• Check image quality metadata provided by the device output
• Remember the device provides preliminary findings - always consider the complete clinical context
• For localized lesions, ensure the camera is close to the lesion to minimize background noise

Poor image quality affecting results

Symptoms

• Device reports low image quality scores
• Diagnostic confidence is consistently low
• "Non-specific lesion" appears as top result
• Unclear or blurry explainability media output

Possible causes

• Insufficient lighting during image capture
• Camera out of focus on the region of interest
• Image taken from too far away
• Movement blur during capture
• Compressed or resized images

Solutions

• Use the flash: This is the most effective solution for lighting issues
• Tap to focus: Ensure the camera focuses specifically on the lesion
• Get closer: Position the camera as close as possible to the lesion
• Steady hands: Hold the device steady during capture or use a tripod
• Use original, uncompressed images (avoid WhatsApp, email compression)
• Ensure minimum image resolution of 2000 pixels in width or height
• Review technical specifications for compatible camera requirements (minimum 12 MP)

Interpreting severity measurements and clinical signs

Symptoms

• Uncertainty about quantified clinical sign values
• Difficulty correlating device scores with clinical assessment
• Questions about score ranges and clinical significance
• Inconsistent severity scoring between similar cases

Possible causes

• Incomplete understanding of the device's scoring methodology
• Expectations of traditional clinical scoring systems
• Variation in image capture conditions
• Multiple clinical signs present simultaneously

Solutions

• Remember the device outputs
• Review the Clinical benefits section for performance evidence
• Scores range from 1-100 for improved granularity and minimum observable change
• Consider the device output as one element of overall clinical assessment
• Use explainability media to understand which areas the device identified
• Multiple images of the same lesion can improve measurement accuracy

Malignancy risk assessment concerns

Symptoms

• Unexpectedly high or low malignancy suspicion scores
• Disagreement between clinical suspicion and device output
• Questions about prioritization recommendations
• Uncertainty about when to refer patients

Possible causes

• Benign lesions with suspicious visual features
• Early-stage malignancies with subtle visual cues
• Image quality affecting feature detection
• Multiple lesions in the same image
• Conditions outside typical malignancy categories

Solutions

• The malignancy suspicion is a risk stratification tool, not a diagnostic confirmation
• Use the malignancy index for prioritization rather than definitive diagnosis
• Always combine device output with clinical history and examination findings
• Consider the AUC performance metric (0.96) when interpreting results
• Low suspicion scores don't rule out clinical concern - follow clinical judgment
• High suspicion scores warrant further investigation regardless of clinical appearance
• Review Key performance characteristics for validation data

Difficulty matching device output with clinical diagnosis

Symptoms

• Device suggests conditions different from clinical impression
• ICD-11 classifications don't match local terminology
• Top-5 results seem unrelated to presenting condition
• Confusion about condition names and classifications

Possible causes

• Different diagnostic nomenclature between device and local practice
• Device uses ICD-11 standard classifications
• Rare conditions not well-represented in training data
• Image shows features of multiple conditions

Solutions

• Familiarize yourself with ICD-11 classification system used by the device
• Review Testing the performance for condition matching guidance
• Use the Top-5 accuracy approach - consider all five suggestions, not just the top result
• Remember that some conditions can evolve (e.g., actinic keratosis to squamous cell carcinoma)
• Device provides interpretative distribution representation of possible ICD classes
• Consider that the device outputs preliminary findings to support, not replace, clinical judgment

Performance validation and testing concerns

Symptoms

• Need to validate device performance in local setting
• Questions about accuracy metrics
• Desire to conduct internal performance testing
• Uncertainty about appropriate benchmarks

Possible causes

• Local validation requirements
• Quality assurance protocols
• Research or audit purposes
• Performance monitoring needs

Solutions

• Use provided performance testing templates
• Ensure test conditions match real-world usage:
  • Use images directly from smartphones/digital cameras/dermatoscopes
  • Avoid compressed or internet-downloaded images
  • Maintain proper image resolution (minimum 2000 pixels)
• Match diagnostic terminology between gold standard and device output
• Select appropriate metrics for your use case:
  • Top-5 accuracy for clinical decision support
  • Malignancy suspicion AUC for prioritization
  • RMAE for clinical signs quantification
• Review published research and clinical validation studies for benchmarks

Device limitations and appropriate use

Understanding device scope

The device is designed for:

• Visible skin structures only - cannot assess non-visible conditions
• Preliminary diagnostic support - not for diagnostic confirmation
• Clinical decision support - to be used by healthcare professionals
• Quantification of visible clinical signs - intensity, count, and extent

When NOT to rely solely on device output

• Emergency situations requiring immediate clinical judgment
• Rare conditions not well-represented in training data
• Non-dermatological conditions or systemic diseases
• Confirming clinical diagnosis - device provides support, not confirmation
• Images of poor quality or non-dermatological content

Best practices for clinical integration

1. Always combine device output with clinical history and examination
2. Review image quality indicators provided by the device
3. Use explainability media to understand device reasoning
4. Consider the complete clinical context including patient history and symptoms
5. Follow institutional protocols for device integration and validation
6. Maintain clinical oversight as the device is an adjunctive tool

Need further assistance?

If these clinical guidelines do not resolve your concerns or if you need additional clinical support, please refer to the Need support? section. When contacting support, please include:

• Description of the clinical scenario
• Image quality assessment results
• Device output including diagnostic suggestions and confidence scores
• Clinical context and specific concerns

You may also find it helpful to review:

• Clinical benefits for evidence-based performance data
• Key performance characteristics for detailed metrics
• Precautions for identified risks and mitigation strategies