AI-Driven Quality Improvement for Colonoscopy Surveillance: 97.5% Recall Resolution in Adenoma Patients Across 1.2 Million Procedures
DDW 2026 Research Summary · Latica & Gastro Health
Presented at DDW 2026 (Digestive Disease Week) · Community Gastroenterology Practice Group, Multi-State (Gastro Health) · AI Partner: Latica
The Problem
Missing documented recall intervals create gaps in colorectal cancer screening and prevention. At Gastro Health, a multi-state community gastroenterology practice, more than 250,000 of 1.2 million procedures (22.3%) lacked a documented recall plan. Latica and Gastro Health evaluated whether a data informed intervention process could systematically close this gap and align recall intervals with clinical guidelines with particular focus on higher-risk adenoma patients.
The AI Intervention
Latica’s algorithm automatically extracted adenoma findings from pathology reports for risk stratification and identified patients appropriately excluded from recall by analyzing physician notes. Recall resolution was defined as a documented recall date, a verified recall exclusion, or a future scheduled recall. The program, initiated in May 2024, combined two distinct patient populations. One analyzed retrospectively and one monitored prospectively:
802,446 historical procedures · 2017–2023
Historical procedures were reviewed using AI-prioritized lists stratified by adenoma burden and time-since-colonoscopy. Patient records were updated with guideline-based recall plans, and overdue patients were contacted and scheduled.
393,406 new procedures · January 2024–present
Continuous AI monitoring of new procedures, with reports generated approximately one month post-procedure to surface cases requiring recall entry — preventing new documentation gaps from forming going forward.
Results
Retrospective Remediation (2017–2023 procedures)
| Metric | Before | After | Change |
|---|---|---|---|
| Overall cohort recall resolution | 77.7% | 94% | +16.4% (p<0.001) |
| Adenoma subgroup (272,597 higher-risk patients) | 87% | 97.5% | +10.5% (p<0.001) |
The retrospective intervention resolved recall plans for 131,000 additional patients. Of the higher-risk overdue adenoma cohort, 3,000 prioritized patients achieved 100% scheduling completion in 2025, with intervention continuing for additional overdue patients.
Prospective Monitoring (Jan 2024–present procedures)
Continuous AI monitoring maintained a 91.5% recall resolution rate across new procedures overall, and 96.1% among new adenoma patients demonstrating that prospective surveillance sustains high resolution rates without requiring retrospective cleanup.
Combined Impact Across ~1.2 Million Colonoscopies
1,113,152 colonoscopies (93.2%) achieved a resolved recall plan across both cohorts combined. The remaining 6.8% (81,700 procedures) are still under ongoing intervention.
Why It Matters
This is, to Latica and Gastro Health’s knowledge, the first AI-enabled, EMR-integrated program to apply both systematic retrospective remediation and prospective monitoring across more than 1 million colonoscopies to improve recall documentation and scheduling. The result: near-complete recall resolution for adenoma patients and timely scheduling pathways for high-risk individuals in a real-world community practice setting. Large language models, now being implemented, will further enhance adenoma risk stratification and advance precision care navigation in community gastroenterology.
Interested in applying AI-driven risk stratification to your own recall or surveillance programs?
Schedule a Scientific Discovery Call with our team →
Frequently Asked Questions
What is the difference between the retrospective and prospective patient groups in this study?
The retrospective cohort was 802,446 historical procedures (2017–2023) reviewed after the fact to remediate missing recall plans. The prospective cohort was 393,406 new procedures (Jan 2024–present) continuously monitored by AI in near-real time to prevent gaps before they occur.
What recall resolution rate did the AI intervention achieve in adenoma patients?
97.5% in the retrospective adenoma subgroup (up from 87%, p<0.001), and 96.1% among adenoma patients under ongoing prospective monitoring.
How many total colonoscopy procedures were included?
Approximately 1.2 million, combining the 802,446-procedure retrospective cohort and the 393,406-procedure prospective cohort. 93.2% (1,113,152 procedures) achieved a resolved recall plan.
Who conducted this research?
Latica., in partnership with Gastro Health (a multi-state community gastroenterology practice). Findings were presented at DDW 2026.
Source: “AI-Driven Quality Improvement for Colonoscopy Surveillance: Achieving 97.5% Recall Resolution in Adenoma Patients Across 1.2 Million Procedures.” Presented at Digestive Disease Week (DDW) 2026. Latica.AI & GastroHealth.“At Forefront, every decision we make is guided by one goal: to elevate the experience of both our patients and our physicians.” said Betsy Wernli, President of Forefront Dermatology. “Partnering with Latica enables us to advance this mission even further – leveraging data to drive better insights, optimize care, and improve outcomes in conjunction with our physician-led, best-in-class clinical quality program Forefront IQ”