Tailoring treatments to individual patients (precision medicine) has the potential to greatly improve patient outcomes. Oftentimes there are tradeoffs between more and less aggressive treatments. There is a need for new statistical methods that can be used to learn about what strategies work best from existing clinical data sets.

This study will evaluate the development of new machine learning methods that can inform how often to monitor clinical markers of disease and how to alter patients’ treatment based on these markers. The research will also create novel methods to interrogate causal mechanisms in the context of adaptive treatment and monitoring strategies as potential targets for further optimization. In addition, evidence will be generated to inform the ideal chemotherapy roadmaps and echocardiographic monitoring schedules for patients treated for pediatric acute myeloid leukemia (AML).

An advisory panel of key stakeholders including patients, physicians, clinical researchers, and methodologists will be formed, who will inform the design and conduct of each aim of the proposed research, as well as the dissemination of the methodological advances. The multidisciplinary membership of the advisory panel will ensure the methods developed have broad applicability and straightforward implementation. The study team will specifically engage the Children’s Hospital of Philadelphia Family Advisory Council, COG Patient Advocacy Committee, patients treated for childhood AML, and pediatric oncologists and cardiologists to be sure their perspectives on the risk tradeoffs related to chemotherapy and monitoring for associated cardiotoxicity are reflected in the design of the applied research.