Pediatric cancer outcomes are one of the feel-good stories in modern medicine. For example, in the 1960s, a child diagnosed with leukemia had less than 10% chance of survival. Today, that’s flipped, with survival rates for acute lymphoblastic leukemia of greater than 90%. And most will go on to be long-term survivors.
But is survival enough?
Jessica Keim-Malpass, RN, NP, ACPNP, PhD, is a nurse scientist and pediatric cancer nurse practitioner. Her research focuses on how big data could improve patients’ lives. She also wants to see life quality, in addition to quantity, considered as we move to a model of personalized medicine and precision oncology.
The Unique Struggle of Pediatric Cancer
Because pediatric cancer patients go through treatment while their bodies are still physically and mentally developing, the long-term complications are often more widespread. The burden of disease and treatment leaves many young survivors with lifelong physical, emotional, and psychological effects.
These effects are difficult to incorporate into research. Patient-reported outcomes (PRO), especially in connection with clinical trials, could help to shape research in a way that centers patients. It could even help develop precision treatment options that balance effectiveness and quality of life more meaningfully.
But collecting PRO and quantifying it is a challenge. And to have enough data to be meaningful, it needs to be pooled.
Pediatric cancer also struggles from a lack of research funding. Only 4% of cancer research resources are allocated to pediatric cancer. That means researchers have to do more with less and pool their data together to have meaningful effects.
Building Into Existing Architecture
The Childhood Cancer Data Initiative (CCDI) exists to advance research by allowing data sharing. Already, national pediatric cancer data has been added, including clinical trial data from the Children’s Oncology Group, as well as genomic data, biospecimen data, and survivorship data, from organizations like St. Jude, the National Program of Cancer Registries, and the Childhood Cancer Survivorship Study.
Malpass’ project, dubbed The Pediatric Cancer Analytics Network Pathway to artificial Intelligence readiness (PeCan PI) is a pathway for adding more data from a wider network of sources. But that means helping smaller entities connect with the CCDI. Which sounds much easier than it actually is.
Are You Ready for AI?
Artificial intelligence has been transformative in many industries. But few have as much to gain as medicine, where big data allows for more meaningful results. What a human could assess in decades takes a machine minutes.
But participating in big data efforts isn’t as easy as it might first seem.
The first concern is patient privacy. When only 1 or 2 patients are diagnosed with a rare type of cancer, masking their data requires manual effort. But these are some of the most important data sets to contribute.
Masking data allows for contribution without making patients identifiable. But it takes time.
Another practical concern is data language compatibility. “Even if we’re using the same electronic management records as another hospital, we may not have the same build,” Keim-Malpass explains. Two hospitals running Epic, one of the most common EMRs, are still unlikely to have the same configuration.
There’s also the question of how to make PRO into data that has objective measurements that can be analyzed and compared.
One of PeCAN-PI’s most important contributions will be the set of tools and resources it makes publicly available. These include:
- A comprehensive, open-access database of patient-reported and clinical outcomes used in pediatric cancer trials.
- A standardized protocol for collecting PROs across institutions.
- A suite of Python-based AI tools for analyzing time-series data from real-world clinical settings.
- Published guidance to help other researchers and clinicians incorporate PROs and AI tools into their own data systems.
PeCAN- PI In Action
Having a successful case study that shows what this type of modeling can do is the easiest way to get buy-in. Something Malpass’ team is very familiar with. For their first proof-of-concept, the team is mathematically modeling neutrophil dynamics and treatment timing and cumulative chemotherapy. Specifically, they’re looking at acute lymphoblastic leukemia clinical trial data from the Children’s Oncology Group, a robust data set available through the CCDI.
Using this modeling and looking at disease recurrence, outcome metrics, and the impact of race and geography on treatment outcome, they can apply time series algorithms. These findings could help clarify the feasibility of less toxic methods of treatment.
By identifying the least toxic treatment that’s still effective, the team could help create treatment protocols that improve quality of life for patients.
A Team Built for Impact
PeCAN-PI brings together a multidisciplinary team, including UVA Health Children’s oncologists and specialists from data science, nursing, engineering, and education. Their combined expertise spans pediatric oncology, AI modeling, clinical trials, health equity, and data harmonization.
This type of research is something unique that academic medical centers can facilitate. They also benefit from institutional backing by the UVA Comprehensive Cancer Center and the Center for Advanced Medical Analytics, ensuring both scientific rigor and real-world relevance.
Looking Ahead
Pediatric cancer may be rare, but its impact is lifelong, and every data point matters. By building tools that center the patient experience, connecting local data to national efforts, and harnessing the power of AI, PeCAN-PI represents a transformative step forward for pediatric oncology.
Ultimately, this project is about giving every child facing cancer a better chance, not just at survival, but at a healthier, more hopeful future.