Dermatology AI Standard of Care: Skin Cancer Detection, Melanoma Screening, and Liability

AI Enters the Skin Cancer Screening Revolution

#

Skin cancer is the most common cancer in the United States, yet approximately 25% of cases are misdiagnosed. In January 2024, the FDA authorized DermaSensor, the first AI-enabled dermatologic device cleared for use by non-specialists, opening a new frontier for skin cancer detection in primary care settings.

This breakthrough promises earlier detection and improved outcomes, but it also raises complex liability questions. When AI in a primary care office misses melanoma, who bears responsibility? The primary care physician? The dermatologist who wasn’t consulted? The device manufacturer? The healthcare system that deployed the technology?

This guide examines the standard of care for AI use in dermatology, the regulatory landscape for skin cancer detection devices, and the emerging liability framework for AI-assisted skin assessment.

FDA-Cleared Dermatology AI Devices

#

DermaSensor: First AI for Primary Care

#

On January 17, 2024, the FDA authorized DermaSensor, the first AI-enabled medical device for skin cancer detection intended for use in primary care settings.

How It Works:

• Uses elastic scattering spectroscopy (ESS)
• Emits light pulses over isolated areas of lesion
• Backscattered optical reflectance reveals cellular architecture
• AI algorithm analyzes spectral data
• Trained on 4,500+ lesions with histological confirmation

Clinical Performance:

Pivotal Trial (Mayo Clinic):

• 1,000+ patients
• 224 confirmed skin cancer cases
• 96% sensitivity across all skin cancers

Primary Care Impact Study:

• 108 primary care physicians
• DermaSensor decreased missed skin cancers by 50% (18% → 9%)

Regulatory Significance

#

First for Non-Specialists: DermaSensor establishes a new regulatory precedent as the first AI-enabled dermatologic device indicated for use by non-specialists. This represents a fundamental shift from the traditional dermatologist-centered model.

International Approvals: The device was already authorized in the European Union and Australia before U.S. clearance.

Emerging AI Research

#

Multimodal AI (2025): Researchers from Incheon National University and collaborators developed a deep learning model combining dermoscopic images with clinical patient data:

• 33,000+ dermoscopic images (SIIM-ISIC melanoma dataset)
• Multi-layer CNN for image analysis
• Clinical metadata integration
• More informed, precise melanoma predictions

Clinical Applications and Risk Areas

#

Melanoma Detection

#

The Stakes: Melanoma is the deadliest form of skin cancer:

• Early detection dramatically improves survival
• Late-stage diagnosis significantly reduces outcomes
• Visual inspection alone misses substantial percentage

AI Role:

• Augment visual examination
• Flag suspicious lesions for biopsy
• Reduce diagnostic uncertainty in primary care
• Enable earlier specialist referral

Liability Concerns:

• False negatives leading to delayed diagnosis
• False positives causing unnecessary biopsies
• Performance variations across skin tones
• Reliance on AI without clinical correlation

Non-Melanoma Skin Cancers

#

Basal Cell Carcinoma (BCC):

• Most common skin cancer
• AI detection supports early intervention
• Generally non-metastatic but locally destructive

Squamous Cell Carcinoma (SCC):

• Second most common skin cancer
• Can metastasize if untreated
• AI assists in early identification

Pediatric Skin Cancer

#

Special Concerns: Treatment of melanoma in children is frequently delayed due to misdiagnosis of pigmented lesions, occurring up to 40% of the time.

AI Potential: Earlier, more accurate screening could reduce diagnostic delays in pediatric populations.

Teledermatology Integration

#

Store-and-Forward AI: AI increasingly integrates with teledermatology platforms:

• Image analysis before dermatologist review
• Triage and prioritization
• Quality assurance for image adequacy

Real-Time AI Assistance:

• Live dermoscopy analysis
• Point-of-care decision support
• Remote expert consultation augmentation

The Liability Framework

#

Primary Care Deployment Challenges

#

DermaSensor’s clearance for primary care creates novel liability questions:

Who Bears Responsibility?

• Primary care physician ordering/using device
• Device manufacturer (DermaSensor)
• Healthcare system deploying technology
• Dermatologist (if referral not made)

The Standard of Care Shift: With AI available to non-specialists, expectations may change for what constitutes reasonable screening.

Skin Cancer Misdiagnosis Claims

#

Traditional Claim Elements: To establish liability in skin cancer misdiagnosis:

1. Duty of care under doctor-patient relationship
2. Breach of standard of care
3. Injuries directly resulted from breach

Common Diagnostic Errors:

• Missed diagnosis (failure to identify potentially cancerous spot)
• Delayed diagnosis (significant delay in identification)
• Misdiagnosis (incorrect cancer type identification)

Settlement Values

#

Medical Malpractice Statistics: The average value of a medical malpractice case involving failure to diagnose cancer is between $400,000 and $700,000.

Multiple Party Liability

#

Potential Defendants:

• Primary care provider or dermatologist
• Laboratory (if specimen analysis inadequate)
• Hospital or cancer center
• AI device manufacturer
• Healthcare system deploying AI

AI-Specific Liability Considerations

#

Performance Across Skin Tones: DermaSensor showed:

• 96% sensitivity in FST I-III (lighter skin)
• 92% sensitivity in FST IV-VI (darker skin)

This 4% gap could create liability exposure if darker-skinned patients experience higher miss rates.

The “Black Box” Problem: AI algorithms cannot fully explain their classifications, creating challenges for:

• Determining causation
• Apportioning fault
• Expert witness testimony

Standard of Care for Dermatology AI

#

What Reasonable Use Looks Like

#

For Primary Care Using AI Screening:

• Use only FDA-cleared devices within approved indications
• Understand device limitations and appropriate patient populations
• Do not use AI as substitute for clinical judgment
• Refer suspicious lesions regardless of AI output
• Document AI use and clinical reasoning
• Follow manufacturer protocols

For Dermatologists:

• Maintain expertise in visual and dermoscopic examination
• Use AI as adjunct, not replacement for training
• Apply clinical judgment to all AI-assisted findings
• Document independent assessment
• Stay current on AI device performance data

For Healthcare Systems:

• Deploy only FDA-cleared AI devices
• Train all users on capabilities and limitations
• Establish quality monitoring programs
• Track outcomes and false negative/positive rates
• Report adverse events to FDA MAUDE

What Falls Below Standard

#

Implementation Failures:

• Using non-FDA-cleared AI for diagnosis
• Deploying without adequate staff training
• No quality assurance program
• Failure to track outcomes

Clinical Failures:

• Treating AI output as definitive diagnosis
• Failing to refer suspicious lesions despite AI “clear”
• Ignoring clinical suspicion because AI is negative
• Using AI outside approved indications

Systemic Failures:

• No adverse event reporting
• Ignoring performance disparities across populations
• Failure to update for known issues

AAD Professional Guidelines

#

Teledermatology Standards (2024)

#

The American Academy of Dermatology’s Teledermatology Standards provide guidance on technology-assisted dermatology:

Key Recommendations:

Board-Certified Care: The AAD recommends that all teledermatology platforms offer access to care directed by a board-certified dermatologist.

Patient Choice: Platforms should offer patients the option to access in-person dermatology services when necessary or preferred.

Technical Standards:

• Minimum 800 x 600 pixel resolution for diagnostic images
• Monitor matched to camera resolution
• Minimum 384 kbps internet speed

Standard of Care Clarification: These Teledermatology Standards are intended to offer physicians guiding principles regarding the practice of dermatology. These standards are not intended to establish a legal standard of care.

AAD Position on AI (2019)

#

The AAD Position Statement on AI emphasizes:

• Technology should be collaboratively developed
• Minimize risk of disruptive effects
• Avoid unintended consequences
• Maintain dermatologist oversight

The Non-Specialist Question

#

DermaSensor’s approval for primary care use raises questions about:

• When is specialist referral required?
• Does AI screening change referral thresholds?
• What training is adequate for AI use?

The AAD has not yet issued specific guidance on AI device use by non-dermatologists.

Emerging Developments

#

UK Teledermatology AI Study (2024)

#

A study published in Frontiers in Medicine evaluated AI accuracy in a UK skin cancer teledermatology service:

• Real-world clinical setting
• Integration with existing referral pathways
• Assessment of AI performance vs. specialist review

Multimodal AI Integration

#

The future of dermatology AI points toward:

• Combined image and clinical data analysis
• Patient history integration
• Risk factor modeling
• Longitudinal tracking

Primary Care Expansion

#

DermaSensor’s clearance may accelerate:

• AI deployment in non-dermatology settings
• Earlier screening during routine visits
• Rural and underserved area access

Frequently Asked Questions

#

Related Resources

#

AI Liability Framework

#

• AI Misdiagnosis Case Tracker, Diagnostic failure documentation
• AI Product Liability, Strict liability for AI systems
• Radiology AI Standard of Care, Diagnostic imaging AI

Healthcare AI

#

• Healthcare AI Standard of Care, Overview of medical AI standards
• AI Medical Device Adverse Events, FDA MAUDE analysis
• Ophthalmology AI Standard of Care, Eye screening AI

Emerging Litigation

#

• AI Litigation Landscape 2025, Overview of AI lawsuits