Clinical characterization of treatment-related morbidity in pediatric brain tumor survivors with artificial intelligence (clarity-AI) - pediatric brain tumor (PBT) survivors experience chronic…

The National Cancer Institute awarded Brigham & Women's Hospital Inc. $815,772 on August 1, 2026, under the Cancer Cause and Prevention Research program (CFDA 93.393) to develop and validate artificial intelligence algorithms that diagnose sarcopenia and predict treatment-related morbidity in pediatric brain tumor survivors using routine clinical imaging.

The project, titled "Clinical Characterization of Treatment-Related Morbidity in Pediatric Brain Tumor Survivors with Artificial Intelligence (CLARITY-AI)," addresses a clinical need for cost-effective, practical diagnostic tools. Pediatric brain tumor survivors experience chronic morbidity including physiologic frailty, accelerated aging, and sarcopenia—muscle loss that reduces quality of life. Early interventions such as tailored exercise and nutrition regimens can mitigate downstream complications, but current clinical data collected for cancer survivors do not adequately characterize muscle status and associated morbidity. The research team will apply AI tools developed within their group to large multi-institutional cohorts, including one retrospective and two prospective cohorts comprising 2,649 patients with detailed clinical and outcomes data and approximately 20,000 longitudinal magnetic resonance images. The work will identify and characterize risk factors associated with AI-diagnosed sarcopenia and develop predictive models for morbidity in pediatric, adolescent, and young adult brain cancer survivors.

Performance of work occurs in Boston, Massachusetts. The period of performance runs from August 1, 2026, through July 31, 2031. This is a Project Grant, the standard assistance type for investigator-initiated research under the National Cancer Institute.

Clinical characterization of treatment-related morbidity in pediatric brain tumor survivors with artificial intelligence (clarity-AI) - pediatric brain tumor (PBT) survivors experience chronic morbidity from their cancers and associated treatments that have long-term negative impacts on quality of life, including physiologic frailty and accelerated aging, characterized by sarcopenia (i.e. muscle loss). Early interventions, such as tailored exercise and nutrition regimens can mitigate the downstream sequalae of sarcopenia, and pharmacologic therapies are under investigation, however, the clinical issues are nuanced, variable, and difficulty characterize from clinical data routinely collected for cancer survivors. Thus, there is an urgent need for cost-effective, practical tools to improve early diagnosis, characterization, and monitoring of muscle status, sarcopenia, and associated morbidity in PBT survivors. The long-term goal is to enable personalized management of PBT survivors by identifying those at risk of sarcopenia and its downstream complications and triaging them to the appropriate intervention. The objective of this proposal is to determine how artificial intelligence (AI) algorithms applied to routine, clinical imaging can diagnose sarcopenia and predict morbidity in PBT survivors. The central hypothesis is that AI innovations developed by our group can be successfully applied to children, adolescents, and young adult survivors of brain cancers to characterize and predict morbidity in survivorship. Our hypothesis makes specific predictions that we will test by applying AI tools developed within our group to large, multi-institutional cohorts, including one retrospective and two prospective cohorts of PBT survivors (2,649 patients with detailed clinical and outcomes data and ~20,000 longitudinal magnetic resonance images (mris). We will study the following aims: 1) identify and characterize risk factors associated with sarcopenia diagnosed from AI-based temporalis muscle thickness; 2) test the hypothesis that AI-based temporalis muscle thickness (itmt) can characterize and predict morbidity (physiologic frailty, neurocognitive and endocrine dysfunction) and mortality in PBT survivors; and 3) evaluate a 3-dimensional extracranial tissue assessment tool that includes temporalis muscle volume measurement and determine if volumetric muscle assessment improves outcome prediction in PBT survivors. The expected outcome of this work is the development of publicly available, cost-effective AI tools that effectively diagnosis sarcopenia and provide an early indicator of associated morbidity in pediatric cancer survivors. The results will have an immediate positive impact as they will establish a better understanding of sarcopenia and related morbidity in cancer survivorship and ultimately enable inexpensive, practical, automated diagnosis and tracking of sarcopenia across patients with cancer and other diseases.