Beyond Survival: AI’s Promise for the Future of Children with Cancer

Every year, approximately 400,000 children and adolescents develop a form of pediatric cancer globally. In other words, over a thousand families every day have their lives ripped apart due to a devastating diagnosis – one that will rewrite their future permanently. Childhood cancer remains the leading cause of disease-related death in children, and part of this is because of late/delayed diagnoses (6). As time passes, the aggressive progression of the disease requires harsher treatments that kids are not always able to withstand (2). 

This is where artificial intelligence (AI) can be promising. In the world of AI, machine learning (ML), which is where computers learn from data, and deep learning (DL), where the computer analyzes complex neural networks to uncover patterns in genetic sequences/clinical records, are working to increase not only the survival rate but also improve the quality of life for pediatric cancer patients (1). These tools, which have previously been tested solely in experimental settings, are being rapidly integrated into real clinics and hospitals, bringing a renewed sense of hope to many families.   

Medical imaging is one of the earliest yet powerful applications of AI in healthcare, allowing doctors to view and analyze tumors closely. However, because the developing brain can vary significantly from child to child, interpreting pediatric neuroimaging is challenging, as it can be difficult to distinguish normal developmental variations from abnormalities. To combat this, deep learning models that automatically segment tumors with precise margins have been developed, distinguishing tumor tissue from surrounding healthy tissue in medical images. This process provides clinicians with a clear map of the tumor’s size, shape, and location, which is critical for creating a treatment plan. This accelerates treatment planning and ensures therapies are directed precisely at the tumor, preserving as much healthy tissue as possible to minimize risk for the child’s “normal” development. For example, ML has been used to develop risk assessments, which have been used to guide treatment decisions, such as whether a child requires intensive chemotherapy, can benefit from surgery alone, or may qualify for reduced-toxicity therapies, allowing the treatment choice to be matched with the child’s individual risk profile (4).

While many assume remission is the end of the battle, few know the risks children face for the rest of their lives, even after successfully beating cancer. Nearly ⅔ of survivors develop at least one chronic health condition, such as arrhythmias, hypertension, endocrine disorders, or asthma, whether this be from relapse or treatment-related effects, and around ¼  experience severe late effects, such as heart failure, infertility, or secondary cancers, with a combined mortality rate of 23% (5) AI is specifically working to mitigate these risks through training ML models to analyze electrocardiograms, specifically to predict chemotherapy-induced cardiomyopathy years before clinical onset (3). This enables doctors to utilize AI to personalize long-term follow-up schedules based on each patient’s estimated risk of developing late-onset, potentially life-threatening chronic effects.

Although artificial intelligence will never be able to alleviate the excruciating pain of a cancer diagnosis, it is changing the meaning of a cure. By enabling earlier diagnoses and more precise treatment/long-term monitoring plans, AI is not only improving chances of survival but also enhancing the quality of life after treatment. For the hundreds of thousands of children diagnosed each year, these advances are the reason why they can grow and thrive in their future unafraid.         

References:

1. Bhinder, B., Gilvary, C., Madhukar, N. S., & Elemento, O. (2021, April 11). Artificial Intelligence in Cancer Research and Precision Medicine. PubMed Central. https://pmc.ncbi.nlm.nih.gov/articles/PMC8034385/ 

2. Dang-Tan, T., & Franco, E. (2007, August 15). Diagnosis Delays in Childhood Cancer: A Review. PubMed. https://pubmed.ncbi.nlm.nih.gov/17620277/ 

3. Güntürkün, F., Akbilgic, O., Davis, R. L., Armstrong, G. T., Howell, R. M., Jefferies, J. L., Ness, K. K., Karabayir, I., Lucas, J. T., Jr, Srivastava, D. K., Hudson, M. M., Robison, L. L., Soliman, E. Z., & Mulrooney, D. A. (2021). Artificial Intelligence-Assisted Prediction of Late-Onset Cardiomyopathy Among Childhood Cancer Survivors. JCO clinical cancer informatics, 5, 459–468. https://doi.org/10.1200/CCI.20.00176

4. Jahangiri, L. (2024, January 6). Predicting Neuroblastoma Patient Risk Groups, Outcomes, and Treatment Response Using Machine Learning Methods: A Review. PubMed Central. https://pmc.ncbi.nlm.nih.gov/articles/PMC10801560/ 

5. PDQ® Pediatric Treatment Editorial Board. (2025, April 14). Late Effects of Treatment for Childhood Cancer (PDQ®) - Health Professional Version. Late Effects of Treatment for Childhood Cancer (PDQ®) - NCI. https://www.cancer.gov/types/childhood-cancers/late-effects-hp-pdq 

6. World Health Organization. (2025, February 4). Childhood Cancer. World Health Organization. https://www.who.int/news-room/fact-sheets/detail/cancer-in-children