ColoMAIA II raises ADR by 38% in a clinical trial.
ColoMAIA II was evaluated as a real-time assistive CADe tool in a 600-patient multicentre randomised clinical trial against standard colonoscopy. Physicians using ColoMAIA II detected more adenomas, with no serious adverse device effects reported.
Prevent more colorectal cancers. With evidence.
- +38% relative ADR in a 600-patient multicentre RCT (39.9% vs 29.0%, p = 0.0033).
- +61% relative APC — more adenomas detected per colonoscopy (0.976 vs 0.606, p = 0.0070).
- Strongest where it matters: +63% ADR in the proximal colon, +81% with lower bowel preparation quality.
- Zero serious adverse device effects across all 600 randomised patients.
- 2.1× shorter alert delay and 3.2× less false-alarm time versus CAD EYE on identical recordings.
Higher adenoma detection in a pivotal randomised trial
In the pivotal clinical investigation NCT07563816, physicians using ColoMAIA II detected adenomas in more patients and detected more adenomas per colonoscopy than physicians performing standard colonoscopy without AI assistance, under the investigated conditions of use.
39.9% vs 29.0% standard.
0.976 vs 0.606 standard.
Where AI helps most: hard-to-find lesions.
In subgroups where adenomas are easier to miss, the effect of CADe assistance is most pronounced.
29.9% vs 18.3% standard.
41.6% vs 23.0% standard.
More detection without a loss of clinical relevance
ColoMAIA II was built to fit naturally into the flow of colonoscopy and support physicians without unnecessary disruption. In the clinical trial, higher adenoma detection was accompanied by a comparable proportion of clinically significant excised lesions, supporting that the added detection activity remained clinically relevant.
Consistent direction of effect across the study setting
The observed benefit was not limited to a single site or operator. Exploratory analyses showed numerically higher detection metrics across all participating centres and across most clinically relevant subgroups, including indication, sex, bowel preparation quality, and most age groups.
Pronounced improvement in the proximal colon
In the proximal colon, where lesions can be more difficult to detect, ADR was 29.9% with ColoMAIA II versus 18.3% with standard colonoscopy — a +63% relative increase. The corresponding adenoma yield was also higher, with APC 0.639 versus 0.333.
Support with lower bowel preparation scores
Among patients with adequate but lower bowel preparation quality (BBPS 6–7), ADR was 41.6% with ColoMAIA II versus 23.0% with standard colonoscopy — a +81% relative increase. Adenoma yield was also higher in this subgroup, with APC 1.157 versus 0.541.
Adenoma detection rate (ADR)
Superiority demonstratedAdenomas per colonoscopy (APC)
Superiority demonstratedClinical relevance of excised lesions was preserved
The histology-based positive percent agreement for clinically significant excised lesions was comparable between arms: 77.6% with ColoMAIA II and 77.5% with standard colonoscopy. The pre-specified non-inferiority criterion was met; superiority was not demonstrated. This is best interpreted as preserved clinical relevance of excised lesions, not as improved lesion selectivity.
No serious adverse device effects were reported. No adverse device effects, serious adverse events, withdrawals due to adverse events, or deaths were reported in either arm during the study safety reporting period.
Compact study context
A compact overview of the population, comparator, study setting, and governance behind the pivotal clinical investigation.
- Trial registry
- NCT07563816
- Design
- Prospective · open-label · randomised · multicentre · 2-arm controlled
- Patients
- 600 randomised; 570 protocol-eligible patients in the Per-Protocol efficacy set
- Allocation
- 307 ColoMAIA II / 293 control in Safety/ITT; 291 / 279 in Per-Protocol
- Sites
- AGEL Nový Jičín · AGEL Ostrava-Vítkovice · NH Hospital Hořovice
- Endoscopists
- 10 experienced operators
- Population
- Adults 40–80 years indicated for screening, surveillance, or diagnostic colonoscopy
- Comparator
- Standard colonoscopy without AI assistance
- Enrolment
- 11 Nov 2025 – 11 Apr 2026
- Governance
- EU MDR · ISO 14155:2020 · Declaration of Helsinki · GDPR · ethics committees · SÚKL
2.1× shorter detection delay. 3.2× less false-alarm time.
In a retrospective single-centre analysis, ColoMAIA-2 and CAD EYE were evaluated on the same colonoscopy recordings. ColoMAIA-2 alerted faster on rapidly appearing polyps and spent less video time displaying false alarms during active polyp search.
Why this matters. False alarms cause alert fatigue and reduce physician trust in CADe systems — turning a clinical aid into a distraction.
Study design
The analysis used recorded data from 150 colorectal cancer screening colonoscopies performed by three experienced endoscopists. The dataset included 1,945 minutes of recordings and 189 polyps. CAD EYE operated in real time during the original procedures; ColoMAIA-2 was applied offline to the same videos.
| Measure | ColoMAIA-2 | CAD EYE | Interpretation |
|---|---|---|---|
| Sensitivity within 500 ms | 86.8% | 62.4% | Higher short-delay sensitivity for rapidly appearing polyps. |
| Mean detection delay | 382 ms | 814 ms | Shorter mean time from first polyp appearance to alert. |
| False alarms per minute | 3.09/min | 37.4/min | Lower number of individual false detections. |
| False-alarm time | 1.44% | 4.67% | Lower share of active-search video time displaying false alarms. |
Context for interpretation: CAD EYE ran during the original procedures, whereas ColoMAIA-2 was processed offline on the same recordings. The results describe alert-performance metrics on recorded procedures and should be read alongside, not as a substitute for, the clinical RCT results.
Evidence sources
Pivotal randomised clinical trial
Clinical Investigation of the ColoMaia II Software for Supporting Lesion Detection During Colonoscopy. ClinicalTrials.gov identifier: NCT07563816
CADe alert-performance study (DDW poster)
Král J, Knotková K, Hradiš M, Musil P, Svoboda P, Svět J. Retrospective evaluation of computer-aided polyp detection systems CAD EYE and ColoMAIA-2. Presented at Digestive Disease Week (DDW).
This source supports the detection delay, delay-specific sensitivity, false alarms per minute, and false-alarm time data on identical recordings.
Review the evidence behind ColoMAIA II.
Book a demo to discuss the clinical trial results, CADe alert-performance data, and the investigated conditions of use.
ColoMAIA II is an assistive computer-aided detection tool for qualified physicians; it supports but does not replace clinical judgement or physician responsibility for diagnostic and treatment decisions. The clinical-trial results summarised here pertain to the investigated software version and conditions of use. Additional alert-performance data are based on recorded procedures. This page is intended for healthcare professionals and is not medical advice for patients.