How AI is Transforming Pediatric IBD Care

Artificial intelligence (AI) is everywhere! We rely on it to ask Siri to perform the simplest task, such as checking the weather, turn to ChatGPT to assist with flashcard prompts for an upcoming quiz, or ask Google Gemini to create an image of a dog surfing at the beach. However, AI is much more than just a tool for our everyday convenience, it has the potential to transform how we learn, work and live. This shift is especially impactful in healthcare, including emerging applications that are reshaping how we diagnose, understand, and treat conditions such as pediatric inflammatory bowel disease (IBD).

There are several different types of AI models including generative AI, natural language processing (NLP), computer vision, and machine learning (ML). In medicine, recent progress has been driven by greater data availability and advances in ML and deep learning (DL) frameworks (Cabaço & Rodrigues, 2026). ML models learn patterns directly from data and improve their models as they receive more data, enabling accurate prediction of disease risk from an individual’s medical history and genetic information (Sadr et al., 2025). DL, a subset of ML, uses a multilayer neural network composed of “artificial neurons” to recognize unstructured or complex data, making them effective for tasks such as medical imaging (Sadr et al., 2025). By leveraging ML and DL models, healthcare professionals and researchers have been able to process large medical datasets more efficiently and reveal patterns that would be difficult and time consuming for humans to identify (Chakraborty et al, 2024). 

The integration of AI models like ML and DL, are particularly relevant in conditions such as pediatric IBD. Pediatric IBD is a difficult condition to diagnose due to its complex presentation, requiring clinical assessment supported by endoscopic and histological evaluation. Managing pediatric IBD is also challenging because the disease can present differently from one child to another and often follows a fluctuating course (Cabaço & Rodrigues, 2026). Additionally, compared to adults with IBD, pediatric patients frequently present with more extensive and aggressive inflammation and can experience long-term impact on growth development (Cabaço & Rodrigues, 2026). Ultimately, these factors complicate diagnosis and highlight the need for both image-based and multimodal assessment to accurately characterize disease activity.

In pediatric IBD, ML and DL have shown promising results for image-based assessment (Cabaço & Rodrigues, 2026). Image-based approaches focus on data derived from medical procedures such as endoscopy or imaging scans. Endoscopic evaluation is essential for IBD diagnosis, but is limited by interobserver variability and the time consuming interpretation of image results (Cabaço & Rodrigues, 2026). DL-based Convolutional Neural Network (CNN) models are able to overcome these limitations by improving both accuracy and efficiency. In one study, King et al, developed three two-dimensional CNN models to classify whole-slide histopathology images from endoscopic biopsy samples collected from a pediatric IBD cohort (King et al., 2026). These models accurately identified inflammation and chronic architectural changes across different intestinal tissue types (King et al., 2026). This approach enables whole‑slide images to be analyzed in a way that highlights the intestinal regions most relevant for diagnostic review, improving efficiency and facilitating the diagnosis process. 

Additionally, ML and DL have shown great potential for multimodal assessment. Multimodal approaches integrate imaging with clinical and laboratory data to provide a more comprehensive understanding of disease (Cabaço & Rodrigues, 2026). Disease characterization and activity assessment consistently improve when multimodal models combine radiomics with biochemical and clinical variables (Cabaço & Rodrigues, 2026). For instance, one study, performed by Guez et al., evaluated a multimodal ML framework for the non-invasive assessment of endoscopic activity in Crohn’s Disease (CD) using data from a pediatric cohort (Cabaço & Rodrigues, 2026). The model they used incorporated MRI, biochemical, and chemical data compared to the standard MRI-based regression model (Cabaço & Rodrigues, 2026). Their model demonstrated higher accuracy when predicting terminal ileal endoscopic activity (Cabaço & Rodrigues, 2026). Overall, multimodal ML approaches have the potential to enhance pediatric IBD care by reducing reliance on invasive procedures like ileocolonoscopies while still providing accurate, clinically meaningful evaluations.

As AI continues to advance, its impact on pediatric IBD care continues to provide advances in science and medical treatments. ML and DL frameworks have the ability to help clinicians interpret images more accurately, integrate complex data more efficiently, and explore non‑invasive ways to monitor disease activity. With that being said, it is imperative that ethics are taken into consideration and that clear guidelines are established to ensure that patient autonomy is protected. Looking forward, AI has the potential to significantly improve the quality of life and care for children living with IBD. 

‍

By: Hannah Marmor

‍

References: 

Chakraborty, C., Bhattacharya, M., Pal, S., & Lee, S. S. (2024). From machine learning to deep learning: Advances of the recent data-driven paradigm shift in medicine and healthcare. Current Research in Biotechnology, 7, 100164. https://doi.org/10.1016/j.crbiot.2023.100164

Dias Cabaço, G., Rodrigues, L., Dharmaraj, R., & Alkhouri, R. H. (2026). Artificial Intelligence in Pediatric Inflammatory Bowel Disease: Applications in Diagnosis, Monitoring, and Therapeutic Decision-Making. Children, 13(2). https://doi.org/10.3390/children13020260

Martin-King, C., Nael, A., Ehwerhemuepha, L., Calvo, B., Gates, Q., Janchoi, J., Ornelas, E., Perez, M., Venderby, A., Miklavcic, J., Chang, P., Sassoon, A., Rubio, B., Barragan, G., & Grant, K. (2026). Pediatric Inflammatory Bowel Disease Tissue Classification From Pathology Slide Images: Detecting Phenotypes Using Computer Vision. Gastro Hep Advances, 5(5). https://doi.org/10.1016/j.gastha.2026.100899

Sadr, H., Nazari, M., Khodaverdian, Z. et al. Unveiling the potential of artificial intelligence in revolutionizing disease diagnosis and prediction: a comprehensive review of machine learning and deep learning approaches. Eur J Med Res 30, 418 (2025). https://doi.org/10.1186/s40001-025-02680-7

‍