Phantom Dysplasia and Algorithmic Mirages: A Systematic Review Protocol

A 50.4 KB protocol outlines a systematic review to quantify false-positive rates of AI tools in Barrett's esophagus surveillance. The document, authored by Rashi Bilgaiyan and last updated in May 2026, aims to map the prevalence and triggers of AI-generated diagnostic hallucinations. It focuses on the CARB algorithm and related computer-aided detection platforms.

Use Cases

• Benchmarking AI diagnostic accuracy based on false-positive metrics described in the protocol.
• Identifying visual triggers for algorithmic misinterpretations based on mucosal characteristics like inflammation.
• Assessing clinical burdens such as unnecessary biopsy rates mentioned as a primary outcome.
• Informing consensus guidelines for differentiating true neoplasia from AI hallucinations.

Strengths

• Clear methodological framework defined using a PICO (Population, Intervention, Comparator, Outcomes) structure.
• Specific focus on a documented clinical anomaly ('phantom dysplasia') in AI-assisted diagnostics.
• Released under a permissive CC-BY-4.0 license for reuse.

Limitations

• The dataset is a 50.4 KB PDF protocol, not a primary data collection; actual study results are not included.
• Row count and column-level documentation are unknown, limiting suitability assessment for direct analysis.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

figshare

Collection Method

Systematic review protocol outlining a framework for literature synthesis.

Freshness

Last updated 2026-05-21 13:40:19; freshness should be verified.