Closing Care Gaps in Precision Oncology Through Application of Language-Based Artificial Intelligence

Abstract

Artificial intelligence (AI) has the promise to close gaps in the delivery of precision cancer care. While general purpose large language models (LLMs) rely on manual prompts, they lack integration into clinical workflows, often producing inconsistent outputs. However, when optimized with clinical guidelines and integrated into existing workflows, LLMs offer potential through their ability to retrieve complex clinical data from medical records, apply guideline-based reasoning, and proactively generate actionable and explainable intelligence aligned with agreed workflows for oncologists and multidisciplinary teams (MDTs). Growing evidence indicates that this approach can help address care gaps arising from clinicians’ lack of familiarity with evolving guidelines, which has been shown to result in missed opportunities for genomic biomarker testing and treatments. AI offers a powerful tool for addressing these gaps across the patient journey, through earlier diagnosis, biomarker testing and guideline-directed therapies, checking eligibility for clinical trials, optimizing MDTs, and automating preauthorization checklists and potential to support health equity, which could enhance care while reducing clinician burden. However, realizing the full potential of AI in precision oncology, while ensuring financial sustainability and responsible use, will require deep operational and technical integration, rigorous quality assurance, and sustained interdisciplinary collaboration.