Endoscopic mucosal resection (EMR) is an effective, minimally invasive technique for removing complex colorectal polyps, particularly those with sessile or flat morphology. However, procedure duration can vary widely depending on lesion complexity, impacting unit efficiency and patient scheduling.

This study aimed to develop a simple, objective, and clinically applicable predictive model — the SPEED-EMR Score (Simple PrEdictive Estimator of Duration for Endoscopic Mucosal Resection) — to estimate EMR duration based on lesion characteristics.

We retrospectively analyzed 1,077 colorectal EMRs performed between 2020 and 2025 at a tertiary endoscopy center.

EMR duration (minutes), measured from the first image of the lesion to the image of the post-resection mucosal defect, was the primary outcome.

Collected variables included clinicodemographic patient characteristics, lesion characterization data, and technical features of the endoscopic procedure.

Variables significantly associated with the outcome in univariate analysis were entered into a multiple linear regression model. Coefficients were converted into points (1 point ≈ 4 minutes of added duration) to create the SPEED-EMR Score.

Median patient age was 65 years (IQR 58–72); 57.9% were male. The median Charlson Comorbidity Index was 2 (IQR 1–4), and the median EMR duration was 13 minutes (IQR 7–24).

Longer procedures were independently associated with larger lesion size (p<0.001), flat or depressed morphology (Paris classification 0-IIb/0-IIc) (p<0.001), NICE type 2 pattern (p<0.001), difficult locations (ileocecal valve, peri-diverticular, peri-appendiceal, hepatic/splenic flexures, rectosigmoid junction, anal margin) (p<0.001), and recurrence on a scarred area (p=0.003).

These variables were used to create the SPEED-EMR Score (range 0–16 points): lesion sizes of 21–30 mm (+3 points), 31–40 mm (+6 points), >40 mm (+10 points); flat or depressed morphology (+2 points); NICE type 2 (+1 point); difficult location (+1 point); recurrence on a scarred area (+2 points).

The final SPEED-EMR Score calibration model showed good linearity and predictive accuracy (R=0.680, R²=0.463, p<0.001; standard error of the estimate = 12.2 min).Predicted time was calculated as: Predicted time (min) = 3.1 + 4.0 × Score value.

Rounded estimates were stratified into four complexity classes: Class I (0–4 pts): ≤20 min; Class II (5–8 pts): 25–35 min; Class III (9–12 pts): 40–50 min; Class IV (≥13 pts): ≥55 min.

The SPEED-EMR Score provides a practical tool for predicting EMR duration using readily available lesion characteristics. Its clinical application may improve scheduling efficiency, case allocation, and patient counseling in therapeutic colonoscopy.