Comparing Artificial Intelligence Guided Image Assessment to Current Methods of Burn Assessment
Bibliographic record
Abstract
Appropriate identification of burn depth and size is paramount. Despite the development of burn depth assessment aids [eg, laser Doppler imaging (LDI)], clinical assessment, which assesses partial-thickness burn depth with 67% accuracy, currently remains the most consistent standard of practice. We sought to develop an image-based artificial intelligence system that predicts burn severity and wound margins for use as a triaging tool in thermal injury management. Modified EfficientNet architecture trained by 1684 mobile-device-captured images of different burn depths was previously used to create a convoluted neural network (CNN). The CNN was modified to a novel boundary attention mapping (BAM) algorithm using elements of saliency mapping, which was used to recognize the boundaries of burns. For validation, 144 patient charts that included clinical assessment, burn location, total body surface area, and LDI assessment were retrieved for a retrospective study. The clinical images underwent CNN-BAM assessment and were directly compared with the LDI assessment. CNN using a 4-level burn severity classification achieved an accuracy of 85% (micro/macro-averaged receiver operating characteristic scores). The CNN-BAM system can successfully highlight burns from surrounding tissue with high confidence. CNN-BAM burn area segmentations attained a 91.6% accuracy, 78.2% sensitivity, and 93.4% specificity, when compared to LDI methodology. Results comparing the CNN-BAM outputs to clinical and LDI assessments have shown a high degree of correlation between the CNN-BAM burn severity predictions to those extrapolated from LDI healing potential (66% agreement). CNN-BAM algorithm gives equivalent burn-depth detection accuracy as LDI with a more economical and accessible application when embedded in a mobile device.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.008 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.002 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".