Improving colonoscopies to help prevent colorectal cancer

Published on 18 March 2026

Polyps can become cancerous over time. NUH’s new AI dashboard links colonoscopy and lab reports more efficiently, giving doctors faster feedback to remove polyps earlier.

At a glance

• Colorectal cancer often begins as small growths called polyps that can be removed before they turn cancerous.
• NUH has developed an AI-powered dashboard that links colonoscopy findings with lab reports, helping doctors review how colonoscopies are performed.
• By giving doctors faster and clearer feedback, the system helps improve consistency and strengthen colorectal cancer prevention.

During a colonoscopy, a doctor carefully guides a tiny camera through the colon, searching for small growths that most people will never feel or notice.

Some of these growths, known as polyps, are harmless. Others, however, may slowly develop into colorectal cancer over time.

Because the disease tends to develop gradually and follow a recognisable progression, doctors have a window of opportunity to detect and remove polyps early, often during a colonoscopy. 

“Colorectal cancer is usually quite a predictable disease,” said Dr Koo Chieh Sian, Consultant, National University Centre for Digestive Health (NUCD), National University Hospital (NUH). "But in most cases, the cancer is only detected at a fairly late stage, which limits our ability to treat it effectively."

Spotting problems early before cancer develops

Despite the opportunity for prevention, colorectal cancer remains one of the leading causes of cancer deaths in Singapore.

According to the Singapore Cancer Registry Annual Report 2023, the lifetime risk of developing colorectal cancer is 5.6 per cent, with about 2,600 new cases diagnosed each year.

To strengthen prevention efforts, a team at NUH, including Dr Koo, is now exploring how artificial intelligence (AI) can help improve the way colonoscopies are reviewed and monitored.

Currently, doctors already track several indicators to monitor how well colonoscopies are performed. However, gathering and analysing this information has traditionally been slow and labour-intensive.

“Data from colonoscopy reports and pathology findings are stored in different systems,” Dr Koo explained. “Pulling everything together often requires manual review and data entry, which can be time-consuming and sometimes prone to human error.”

To address this, the NUH team developed a Colonoscopy Quality Metrics Dashboard that automatically connects colonoscopy reports with laboratory findings.

The system uses large language models, a form of AI that can analyse written text, to review thousands of pathology reports and match them with the relevant colonoscopy records.

“We’re able to do this with about 97 per cent accuracy,” said Dr Koo.

By bringing this information together in one place, the dashboard provides a clearer picture of how colonoscopies are being carried out across the department.

“This frees up many hours of manual work and ensures that the data we use is consistent and reliable,” Dr Koo added. “It also helps department leaders identify early on where additional support may be needed, so they can maintain a high standard of patient care.”

Supporting both individual feedback and department oversight

The dashboard also helps both individual doctors and department leaders review how colonoscopies are being carried out.

Each doctor can now log in to review their own data, including current performance and historical trends. They can also compare their results with the departmental average, allowing them to see how their practice evolves over time.

As such, instead of relying only on occasional audits, doctors can receive more regular feedback that helps them refine their techniques and improve detection rates.

At the same time, department leaders have a broader overview of performance and workloads across the service. This makes it easier to identify high performers and also spot where additional training or support may be needed.

Ultimately, the goal is simple: to ensure colonoscopies are carried out as thoroughly and consistently as possible, so that potential problems can be identified and treated earlier.

In consultation with Dr Koo Chieh Sian, Consultant, NUCD, NUH.