Automating Colon Polyp Classification in Digital Pathology by Evaluation of a “Machine Learning as a Service” AI Model: Algorithm Development and Validation Study
Bibliographic record
Abstract
BACKGROUND: Artificial intelligence (AI) models are increasingly being developed to improve the efficiency of pathological diagnoses. Rapid technological advancements are leading to more widespread availability of AI models that can be used by domain-specific experts (ie, pathologists and medical imaging professionals). This study presents an innovative AI model for the classification of colon polyps, developed using AutoML algorithms that are readily available from cloud-based machine learning platforms. Our aim was to explore if such AutoML algorithms could generate robust machine learning models that are directly applicable to the field of digital pathology. OBJECTIVE: The objective of this study was to evaluate the effectiveness of AutoML algorithms in generating robust machine learning models for the classification of colon polyps and to assess their potential applicability in digital pathology. METHODS: Whole-slide images from both public and institutional databases were used to develop a training set for 3 classifications of common entities found in colon polyps: hyperplastic polyps, tubular adenomas, and normal colon. The AI model was developed using an AutoML algorithm from Google's VertexAI platform. A test subset of the data was withheld to assess model accuracy, sensitivity, and specificity. RESULTS: The AI model displayed a high accuracy rate, identifying tubular adenoma and hyperplastic polyps with 100% success and normal colon with 97% success. Sensitivity and specificity error rates were very low. CONCLUSIONS: This study demonstrates how accessible AutoML algorithms can readily be used in digital pathology to develop diagnostic AI models using whole-slide images. Such models could be used by pathologists to improve diagnostic efficiency.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.010 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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 source (direct Gemma or distilled Codex), 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".