Artificial Intelligence and Disparities in Colorectal Care
by Dr Elissa Dabaghi (Frontline Gastroenterology Global Taskforce 2025-6)
Colorectal cancer remains one of the most prevalent causes of cancer-related deaths in the United States, where geographic location can significantly affect whether appropriate colorectal cancer care is available. Individuals living in low socioeconomic status areas have about a 37% higher risk of colorectal cancer and a 24% higher risk of cancer-related death than those living in higher socioeconomic status areas. Given the rapid integration of artificial intelligence into medicine, we must assess its potential impact on health equity. As the use of artificial intelligence (AI) in medicine grows, will this technology bridge or deepen existing socioeconomic disparities in access to colorectal care?
Rural healthcare demonstrates the critical need to bridge the gap in healthcare disparities. The number of practicing general surgeons in rural communities has declined rapidly. Meanwhile, specialists, including colorectal surgeons, predominantly reside in urban communities. The general surgeon workforce in rural areas is projected to drop significantly in the next decade, while the workforce in urban and metropolitan communities is projected to almost double. Furthermore, there are substantial geographic barriers for patients living in rural areas. Some statistics show that 1 in 5 Americans residing in rural areas live more than 60 miles from a medical oncologist. These trends directly threaten colorectal cancer screening and treatment, as general surgeons perform over 50% of screening colonoscopies in these rural areas.
With the growing use of technology and the implementation of AI, many wonder whether AI can help bridge this gap in healthcare, particularly in colorectal care, with regard to screening for colorectal cancer and potentially surgical planning. New systems, such as GI Genius, can identify colonic polyps that can be easily missed by the human eye. Certain studies have shown that this technology can decrease the adenoma miss rate from 32.4% to 15.5% when AI is utilized, with improvement seemingly most apparent in non-expert endoscopists. This is critical for reducing colorectal cancer risk over 5 years and can have a significant impact on rural healthcare, which already has limited access to specifically expert endoscopists. These AI systems could therefore help to both train less-experienced endoscopists to achieve higher adenoma detection rates and, beyond that may serve as expert-level support during these screenings going forward. However, it is important to mention that even with the implementation of AI-assisted polyp detection during colonoscopy, endoscopists remain essential to ensuring these systems are utilized effectively and appropriately. Endoscopists must understand the limitations of AI, and avoid overreliance, but rather utilize this technology as an adjunct to clinical judgement and shared decision-making, as highlighted by Frontline Gastroenterology’s review (1).
Another potential utility of AI and machine learning algorithms in rural or resource-strapped settings is to triage patients who require urgent referral to specialist centers. Some AI-based prediction models have succeeded in stratifying colorectal cancer patients by one-year mortality risk, allowing the streamlining of care for more urgent cases as well as the tailoring of appropriate perioperative care to each patient.
Additionally, although patients in rural areas often already use telemedicine for remote consultations and office visits, it could be interesting to pair AI with diagnostics in this setting. Patients in these rural areas who require endoscopic screening could choose to undergo a capsule endoscopy that integrates AI and machine-based polyp detection at their local center. This would ultimately facilitate more convenient (and likely, by extension, timely) colorectal cancer screening for patients that may not have access to specialist centers.
Despite its promise, it is important to acknowledge the potential of AI to widen gaps in US healthcare, either through financial barriers or by creating additional limitations on access to resources. Additionally, implementing the use of AI in a rural healthcare system could be challenging due to the substantial investments and costs required to roll and maintain such systems (at all levels of care). Not only are these communities already financially stretched, but they may also lack the essential resources required to operate this technology (such as platform upgrades and even high-speed internet). Furthermore, many diagnostic AI systems are trained on data from large urban populations, which could lead to higher rates of incorrect diagnostic readings for rural populations with different demographics. As discussed in an article by Frontline Gastroenterology, the reliability of AI-driven predictions is dependent on high-quality data input (2). This article emphasizes that reducing irrelevant data input and systemic bias through precise and rigorous data selection to properly train the algorithm is essential to developing predictive tools that genuinely benefit patients of all demographics. Finally, given the relative ease of implementing these AI systems in large, urban healthcare facilities, as opposed to rural settings, there is a potential for this technology to drive patients away from local already struggling rural hospitals to these centers. This can place further strain on the finances of these institutions, and ultimately potentially worsen healthcare accessibility through their closure. Finally, as things currently stand, healthcare insurance systems and Medicare do not offer differential reimbursement rates for the use of AI-assisted technology, which could represent a challenge for rural hospitals in generating the initial financial outlay required to deploy this technology.
In conclusion, we must ensure that AI reduces and prevents the worsening of disparities in colorectal care, and healthcare in the US as a whole. This includes strengthening and training these algorithms on rural populations to prevent inequities in underserved areas, as well as ensuring adequate funding for AI-assisted care to be implemented in these rural hospitals, which already have limited resources. Overall, however, there is great potential for AI to augment, not replace, rural providers and enhance triaging patients, risk stratification, and access to expert and specialized support.
References
- Olabintan O, Fearnley L, Iniesta R, et al. Artificial intelligence in endoscopy: navigating risk, responsibility and ethical challenges. Frontline Gastroenterology Published Online First: 17 November 2025. doi: 10.1136/flgastro-2025-103107
https://fg.bmj.com/content/early/2025/11/17/flgastro-2025-103107
- Ashton JJ, Brooks-Warburton J, Allen PB On behalf of the British Society of Gastroenterology artificial intelligence in IBD special interest group, et al. The importance of high-quality ‘big data’ in the application of artificial intelligence in inflammatory bowel disease Frontline Gastroenterology 2023;14:258-262.
https://fg.bmj.com/content/14/3/258