Why we invested in Docbot


Wally Wang


At Scale Asia Ventures, we are excited about how artificial intelligence is transforming healthcare delivery. With IDx-DR’s recent FDA-approval as the first AI diagnostic tool permitted for patient use in the US, we expect to see dozens of promising health AI companies garner attention soon.

At the forefront of this innovative space is Docbot, one of our latest health AI investments.


Colorectal cancer is the second leading cause of cancer-related death (1) and remains prevalent globally. While screening techniques like colonoscopies have helped reduce prevalence, colorectal cancer is still missed by colonoscopy 5% of the time (2), with up to 9% of patients developing cancer even when following national guidelines (3). Furthermore, the US is currently experiencing a colonoscopy specialist shortage (4), resulting in immense geographic health disparities.

Our question is simple — how can we increase colonoscopy accuracy, while simultaneously improving access?

Introducing Docbot

Docbot is developing an AI-driven tool for enhanced adenoma detection during colonoscopy screenings. Called the Ultivision AI imaging platform, Docbot’s system is capable of real-time, expert-level analysis of potential adenomas during colonoscopy screenings.

Importantly, Docbot’s system is capable of being used by both physician and non-physician providers. In their preliminary single center study, Ultivision improved adenoma detection rate by 37%. In areas with low concentrations of colonoscopy specialists, Docbot has the potential to increase colon cancer screening rates by orders of magnitude.

We are thrilled to support Docbot in their latest US clinical trial, a multicenter study currently enrolling 978 patients across 6 hospital sites. We hope this clinical trial will pave the way for Docbot’s upcoming FDA De Novo Classification application, with delivery soon to hospital systems across the US.

Our Investment Thesis

Among many competitors, Docbot has the right team with the right experience

It’s not uncommon to see health AI companies, composed solely of ex-FAANG data scientists and machine learning engineers, develop excellent AI models that never make it to clinical trials, let alone to market. Why? Because subject-specific knowledge is essential in medicine — not only in building accurate and generalizable models, but also in understanding how to deploy systems into clinical settings.

At Scale Asia Ventures, we believe in interdisciplinary collaboration. With a team composed of experienced data scientists, engineers, physicians, and medical device executives, Docbot’s team has the necessary people to develop unique insight in addressing one of the largest problems facing the US health system.

In particular, Docbot’s unique business model demonstrates their interdisciplinary roots in leveraging machine learning to promote accessible and cost-efficient medical care. Looking forward, we believe in the potential of a new machine learning-powered GI business model, where clinics of all settings can utilize both physician and non-physician providers to provide high quality colonoscopies and medical care.

AI diagnostics should address not only accuracy, but also accessibility

Mainstream media has a fascination with AI accuracy — A.I. Took a Test to Detect Lung Cancer. It Got an A. While accuracy is certainly important (and Docbot is capable of increasing adenoma detection rate by 37%), the bigger question on our minds recently is accessibility.

Conditions like colon cancer and hypertension are called “silent killers,” not because physicians are unable to diagnose them, but because many patients are unable to see specialist physicians frequently enough to catch early signs of disease. More important than Docbot’s Ultivision system’s accuracy, is its potential to improve accessibility of high-quality colonoscopies. By allowing non-physician providers to deliver accurate and safe colonoscopy examinations, millions of patients in low-resource settings can get the care they need without relocating to urban health centers.

Real-world performance > controlled settings

Google recently tested in Thailand their latest AI diagnostic tool, a 90% accurate diabetic retinopathy detection system. Their goal? To screen 60% of all people with diabetes in Thailand annually (5).

But despite excellent theoretical accuracy, their system proved impractical in real-world use. In Google’s Thailand experiment, we’ve learned a great deal about the impact of environment on AI, and how simple differences in lighting conditions, imaging technique, and camera model can vastly impact real-world AI performance (6).

Docbot takes a different approach, prioritizing real-time and real-world usage and testing. While many of their competitors have opted for pre-recorded videos or image frames from labeled colonoscopies, Docbot’s Ultivision system’s performance has been validated with real-time video feeds. For this reason, we are confident in Docbot’s capability to bring their product to market with end users in mind, and not just theoretical accuracy numbers.

Docbot can save lives

The global colorectal cancer market is massive, valued at $13.7 billion in 2018 with a CAGR of 6.1% between 2018 to 2023 (7). More important however than the monetary valuation of the colorectal cancer market is what it represents — an opportunity to address one of the leading causes of cancer death in the world. As colorectal cancer rates persist in the US, and even grow in other areas of the world, there is an immense need for new technology to enable early, accurate, and safe diagnosis of colorectal cancers.

We believe Docbot will be part of the solution.


1. https://www.cdc.gov/cancer/colorectal/statistics/

2. https://www.nejm.org/doi/10.1056/NEJMoa1309086?url_ver=Z39.88-2003

3. https://www.cghjournal.org/article/S1542-3565(13)00587-9/fulltext

4. https://bhw.hrsa.gov/sites/default/files/bhw/health-workforce-analysis/research/projections/internal-medicine-subspecialty-report.pdf

5. https://www.technologyreview.com/2020/04/27/1000658/google-medical-ai-accurate-lab-real-life-clinic-covid-diabetes-retina-disease/

6. https://www.blog.google/technology/health/healthcare-ai-systems-put-people-center/

7. https://www.bccresearch.com/market-research/pharmaceuticals/colorectal-cancer-therapeutics.html

This may interest you