AI is on Track – the GI Tract that is – to Take Over Colonoscopies

Maya Lapinski

HST 401 - Final Paper


AI is on Track – the GI Tract that is – to Take Over Colonoscopies


Artificial intelligence’s growing influence on everyday life is undeniable. From being intertwined with one’s education to becoming an integral part of scientific analysis, AI has become an essential component of a vast network of diverse fields. AI’s future is not a matter of if it will translate to a specific discipline but when it will inevitably affect it. As a result, it is no surprise that artificial intelligence is being incorporated into healthcare. Whether it is for the purpose of taking detailed notes for medical visits or helping medical professionals diagnose a problem, AI has increasing potential to benefit medicine. This potential is even spanning to AI paving the way for preventative healthcare. But should we pause before welcoming AI with open arms? 


To attempt an answer to this question, I will be focusing on AI’s entrance into the realm of gastrointestinal care– specifically for colonoscopies. Colonoscopies are one of the most promoted forms of preventative healthcare. According to the American Gastroenterological Association, researchers Shahnaz Sultan et al. note that over 15 million colonoscopies are performed yearly within the United States alone (2025). This procedure is praised as the industry standard of colorectal cancer prevention. Without proper preventative measures, colorectal cancer is a life-threatening disease. The American Cancer Society reports that colorectal cancer is the second leading cause of cancer mortality when combining statistics for women and men, with approximately 53,000 expected deaths for 2025 (2025).


Colorectal cancer is the result of uncontrolled development of abnormal tissue growths, or polyps. Colonoscopies screen for precancerous polyps by inserting a flexible tube into the rectum and colon, allowing for the necessary equipment to pass through and remove any areas of concern. This process can be visualized in Figure 1. Colonoscopy effectiveness depends upon the quality of the screening. Adenoma detection rate is the most important indicator of a high quality procedure. Adenomas are precursor polyps that are the most common type to develop into more aggressive forms, such as colorectal cancer. Despite the significance of this indicator, approximately 25% of adenomas are missed– ones failed to be detected or failed to be fully removed (Sultan et al, 2025). Missed adenomas can continue with uncontrolled growth, increasing the hazardous possibility of becoming malignant. The percentage of missed adenomas is evidence of the limitations of traditionally performed colonoscopies.


Figure 1: Illustration of the Colonoscopy Process to Remove Polyps 

(AGA GI Patient Center, 2024)


If colonoscopies are to remain an effective form of preventative healthcare, it is essential for procedures performed to detect all polyps of risk. Within the Internal Medicine Journal, researchers Liu et al. discuss the significant trend of colorectal polyps rising in young adult populations, even for individuals as young as 18 (2024). Already a prevalent issue within older populations, this trend in the rise of precancerous polyps is a strong indication that efforts should be made to advance the technology available to detect these growths. Traditional colonoscopies are not cheap either, with each procedure varying between $1000-3000. According to the Center for Surgery and Public Health at Brigham and Women’s Hospital, colonoscopies generate 14 to 42 billion dollars in revenue (2021). 


It is clear that colonoscopies are not going anywhere, but variation across practice and incomplete visualizations are insufficient in this technologically progressive era. This gap facilitates the entrance of artificial intelligence. 


AI models for colonoscopy assistance have already been developed and in some cases even deployed. While far from being the standard of care, early AI models have demonstrated the ability to assist endoscopists with detection and diagnosis during and throughout the colonoscopy process. Reporting for the National Cancer Institute, Sharon Reynolds highlights how this software is fed with millions of colonoscopy-derived images to later utilize image recognition to detect polyps and alert the attending physician with potential areas of concern (2023). One of the most developed examples of these AI models is the Wision AI EndoScreener. As modeled in Figure 2, Wision AI’s EndoScreener analyzes video stream from the camera of the colonoscope and overlays a blue box on the received images to relay possible polyps (2023). 


Figure 2: Wision AI’s EndoScreener in Action (Wision A.I.,2023)


Not only can AI help detect polyps, but it can also help with real-time diagnosis, such as the characterization of polyp type. This helps physicians discern between adenomatous and noncancerous polyps, helping make difficult decisions about removing areas of concern. (Sultan et al., 2025). The instantaneous analysis that AI is able to perform alongside a colonoscopy can help minimize the amount of missed polyps, simultaneously showing promise in removing more at-risk polyps. AI is able to detect polyps that are difficult to identify with the human eye, such as more flat lesions that do not project outward from the surrounding tissue as easily (Sultan et al., 2025). This subsequently increases ADR, enhancing the quality of the overall procedure. 


By catching the precancerous polyps that a physician may overlook and offering additional support throughout the entirety of the process, artificial intelligence reveals one of its most advantageous features: the ability to standardize care. With such prevalent variation between colonoscopies performed, the ability to standardize protocol and quality would increase accessibility to better treatment. In this way, all who undergo a colonoscopy would have the same standard of care no matter the location or the availability of specialists. As a result, screenings may become more accessible in areas where there are a shortage of specialists, such as rural ones. 


From these advantages alone, AI seems like the obvious direction of colonoscopies. And while it is no doubt that AI’s grasp over preventative medicine will grow, there is a level of hesitation that this software should meet before being readily adopted across boards. The benefits previously covered for AI are the ideal aspects that will come about with a pivot towards technological advancement. Unfortunately, AI is just not there yet. 


AI-assisted colonoscopy systems are not the standard of care just yet, and as a result, these units are expensive to implement. Before AI is adopted across healthcare, one must consider how the cost of implementation leads to the risk that only wealthier hospitals will afford this luxury. Without the development of proper regulation, AI-assisted systems may further cause healthcare disparities to grow instead of standardizing care. 


Not only is AI expensive to use, but the utilization of AI during colonoscopies comes with the risk of overdiagnosis. AI’s immense capabilities allows it to spot polyps that are less than 5 millimeters in size, but these polyps are not likely to become cancerous in the first place (Sultan et al., 2025). Is it better to be safe than sorry? The answer is complicated.


Overdiagnosis leads to a multitude of excessive intervention that may be of no help to the patient in question. When polyps are identified, the patient is meant to come back after a prolonged period to redo the colonoscopy process to see if there are any new polyps of concern. These follow-ups place the individual on more stringent observation, leading to increased levels of anxiety (Sultan et al., 2025). These levels of anxiety may be unwarranted if the AI identified polyps of low malignancy that the endoscopist may not have found themselves. With more polyps uncovered, more people will continue to be surveillanced, leading to more procedures done. A spiral of complications occur as more procedures lead to more money, risks, and resources used. Is this better than potentially missing a polyp that may turn cancerous? 


The truth remains that despite your opinion, AI is not going anywhere. Even despite the complications that may currently limit its reach within preventative healthcare, AI remains on track to redefine the industry. As AI models are increasingly tested and developed for colonoscopy care, they will continue to detect more than the human eye ever could and will most likely have increased ability to discern between noncancerous and precancerous polyps. Yet even with all these potential future improvements, it is important to emphasize that AI only serves as additional support for medical professionals. It is not their replacement, and it should not be treated that way. The physician should not solely rely on artificial intelligence to detect and diagnose polyps but rather should appreciate the extra pair of ‘eyes’ throughout the process.


Circling back to my initially proposed question, should we pause before welcoming AI with open arms? You can, but try not to pause for too long or AI’s eventual integral role in preventative healthcare will catch you off guard. With proper supervision and regulation, AI can provide better care for all those in need of a colonoscopy, which at the end of the day is the whole point of healthcare: equipping people with the support and resources needed to live healthier and happier lives. Colonoscopies are only the start. 


Works Cited

American Cancer Society. Key Statistics for Colorectal Cancer, American Cancer Society, 2025, 

https://www.cancer.org/cancer/types/colon-rectal-cancer/about/key-statistics.html. 

AGA GI Patient Center. Colonoscopy, American Gastroenterological Association, 2024, 

https://patient.gastro.org/colonoscopy/. 

Center for Surgery and Public Health, Estimating Annual Expenditures for Cancer Screening in 

the United States, Brigham and Women’s Hospital, 2021, https://csph.brighamandwomens .org/wp-content/uploads/2021/12/Estimating-Annual-Expenditures-for-Cancer-Screening-in-the-United-States.pdf. 

Liu, L. Nagel, R. Verma, S. Pinidiyapathirage, J. Colorectal Polyps in Young Adults: A 

Retrospective Review of Colonoscopy Data from Toowoomba and the Darling Downs, Internal Medicine Journal, Volume 54:9, pages 1471-1477, 2024, https://onlinelibrary.wiley.com/doi/10.1111/imj.16420. 

Reynolds, S. Is AI Ready to Play a Leading Role in Colorectal Cancer Screening, National 

Cancer Institute, 2023, https://www.cancer.gov/news-events/cancer-currents-blog /2023/colonoscopy-cad-artificial-intelligence. 

Sultan, S. Shung, D. Kolb, J. Foroutan, F. Hassan, C. Kahi, C. Liang, P. Levin, T. Siddique, S. 

Lebwohl, B. AGA Living Clinical Practice Guideline on Computer-Aided Detection–Assisted Colonoscopy, Gastroenterology, Volume 168:4, pages 691 - 700, 2025, https://www.gastrojournal.org/article/S0016-5085(25)00035-6/fulltext. 

Wision A.I. The Pioneer of AI in Evidence-based Medicine, Wision A.I., 2023, 

https://www.wision.com/#/endoScreener. 

Comments

Post a Comment

Popular posts from this blog

Scaling the Potential of Vertical Farming Going into 2025 and Beyond

Image
  Matthew P Kearney Scaling the Potential of Vertical Farming Going into 2025 and Beyond      Vertical farming may seem like something out of a futuristic movie, but they are very real in today’s agricultural industry. Vertical farming may promote a stronger relationship between urbanization and sustainable crop production. Based on recent innovations across agriculture and technology, vertical farming has the potential to revolutionize our current farming system. Through the development of vertical farming-based production lines, cities and otherwise unfarmable areas can be provided with fresh, local, and pesticide-free produce. Vertical farming and other adaptations of agriculture are not new ideas, but instead a historic and human process.      These technologies are ready to meaningfully address global food security through the technological and computational advancements in agriculture. Vertical farming has a promising future, which becomes a case...
Read more

Not Just Tired: The Darker Side of New Motherhood

  Vrinda Modi Not Just Tired: The Darker Side of New Motherhood “134, we got a suicidal,” the radio crackles. As my EMS crew heads to the apartment, I think back to my training, where there was little instruction on how to help those experiencing suicidal ideation. A middle-aged woman sits on the couch, visibly exhausted and cowering from the police officers surrounding her. Her husband, cradles an infant gently in his arms, his face a mix of concern and helplessness. I turn down my radio, sit beside her, and begin asking questions, careful not to sound interrogative. She confesses feeling suicidal since giving birth three weeks ago, her voice trembling with shame. She hasn’t slept properly in days. Although she has no plan to harm herself, her worsening thoughts drove her to call the suicide hotline. Her mother-in-law enters, coldly urging her to “wrap things up.”  The woman’s shame deepens when she refuses to go to the hospital, embarrassed by the stigma around postpartum de...
Read more

Molecules, Models, and Magic: The Exciting World of Computational Chemistry

Vrinda Modi I often spend time in a little cubicle in McLean in the freezing corporate cold wrapped in a blanket in the Hoboken summers sipping on free coffee from the office above. It almost always tastes like chalk, but the routine is so embedded in me. Hence, this is the perfect opportunity to talk a little more about my research in computational chemistry here at Stevens and the advances I expect in the next few years. This field excites me because it bridges my interest in chemistry with modern computing techniques, a powerful tool for advancing our understanding of chemical reactions and molecular interactions. This means we are a dry lab – running simulations on computers. No chemicals, no PPE, just models and algorithms. But why does computational chemistry exist in the first place, especially when chemistry is so rooted in physical matter? The answer is efficiency. Modeling molecules on a computer saves us from ordering chemicals, waiting for deliveries (a delay in chemistry t...
Read more