Virginia Commonwealth University receives $76.8K Aging Research grant from the National Institutes of Health

A synergistic artificial intelligence (ai) approach: integrating structured and unstructured data for enhanced fall risk prediction among older adult cancer survivors - project summary there is growing recognition that treatment-related toxicities such as neuropathy, sarcopenia, psychological distress, and gait disturbance interact with age-related vulnerability to make falls one of the most consequential and preventable outcomes in cancer survivorship. Falls drive loss of independence, disability, and health care costs, and their burden is rising as the population of cancer survivors in the united states exceeds 18 million, with two-thirds of those survivors over the age of 60. However, in contrast to the complex and multifactorial drivers of survivorship outcomes, widely used risk scores rely solely on structured electronic health record (ehr) data. These tools miss critical signals of decline documented in clinical notes, including neuropathy symptoms, psychosocial stressors, and functional impairments, and therefore offer limited clinical actionability. To date, no oncology-specific fall-risk prediction model exists that integrates structured and unstructured ehr data into an explainable tool for clinical use. The proposed F32 award will unite survivorship science with artificial intelligence by developing the first oncology-specific, explainable fall-risk prediction model that leverages both structured ehr variables and elements extracted from clinical notes using large language models. During the F32 fellowship, dr. Namoos will undertake structured training in (1) machine learning and model validation, (2) clinical natural language processing, (3) ai-implementation science, and (4) professional development and grant writing. This integrated training will directly support research aims to develop and validate an explainable fall-risk prediction model in postmenopausal breast cancer survivors and to evaluate its generalizability across common cancers such as colorectal, lung, and prostate malignancies. Aims 1 and 2 (n=~3000) will focus on applied research to develop and validate a fall-risk prediction model in postmenopausal breast cancer survivors, and aim 3 (n=~15000), will test its generalizability across colorectal, lung, and prostate cancers. Future R01 research will embed the validated model within the epic ehr at Virginia Commonwealth University (vcu) health system, prospectively evaluate predictive performance across geriatric oncology patients, and assess provider usability and workflow integration. Combined, the training that dr. Namoos will receive and the research she will accomplish in this proposal will position her as a leader in artificial intelligence-enabled cancer survivorship science in the geriatric population, with a strong basis for an independent line of NIH-funded research at the intersection of survivorship, data science, and implementation. Importantly, this proposal is supported by an experienced and well-funded team of mentors and advisors, and by the collaborative research environment of the injury and violence prevention program, the massey comprehensive cancer center, and the ctsa-funded wright regional center for clinical and translational science at vcu.