Quick Buzz Feed
AI-driven models using patient-reported outcomes and clinical data help predict risk in cancer survivors, enabling more personalized care.
Cancer survivorship often involves unpredictable challenges post-treatment, including physical symptoms, emotional distress, and unexpected medical needs. A new multidisciplinary study from Sylvester Comprehensive Cancer Center explores how AI technologies can analyze electronic health records and patient-reported data to anticipate these outcomes, moving towards more proactive and personalized care.
Patient-Reported Outcomes (PROs) capture crucial patient experiences like emotional well-being and functional limitations that traditional clinical data often misses. Dr. Akina Natori's study redefines PROs as prospective indicators, using them with clinical data to identify survivors at high risk for significant symptom burden or unplanned healthcare use, enabling earlier, targeted interventions.
The research team applied machine learning to data from over 25,000 cancer survivors to predict risks, finding that recent clinical activity strongly predicted acute events, while longer-term trends were better for symptom burden. Adding PROs significantly enhanced prediction accuracy. The models are designed to be interpretable, providing clinicians with actionable insights into patient risk factors for informed decision-making.
This project leveraged expertise from clinical oncology, psychosocial oncology, population sciences, and data science, including contributions from Dr. Vasileios Stathias. This collaborative effort emphasizes combining diverse data with advanced analytics to uncover hidden patterns, informing more proactive and effective survivorship care strategies by addressing the multifaceted nature of patient recovery.
The findings represent a significant step toward transforming cancer survivorship from reactive to proactive care, especially as survivor populations grow. By identifying patients likely to struggle through clinical and PRO-based predictive models, healthcare systems can more effectively allocate supportive resources and address unmet patient needs that might otherwise go unnoticed.
Future research will focus on refining and validating these AI models for broader survivor populations and integrating PRO data-driven risk stratification into standard care protocols. The long-term goal is to build an AI-powered data ecosystem that guides proactive and precision care, anticipating patient needs to enhance long-term outcomes and reduce the burden on both patients and health systems.
This section showcases additional AI-related research and initiatives at the Miller School of Medicine, highlighting diverse applications of artificial intelligence in medical fields such as surgical reporting, bone marrow analysis for myeloma treatment, and virtual care, underscoring the institution's commitment to leveraging AI for advancements in medicine.