Childhood Cancer Data Initiative Annual Symposium (Abstract Registration): Submission #81

Submission information
Submission Number: 81
Submission ID: 193694
Submission UUID: f01f39e8-35e9-4e0b-9cc3-4cb1cdb845ee

Created: Thu, 09/03/2026 - 11:42
Completed: Thu, 09/03/2026 - 11:48
Changed: Thu, 09/03/2026 - 11:48

Remote IP address: 10.208.28.41
Submitted by: Anonymous
Language: English

Is draft: No
Abstract Submission for Poster Presentation
AI-driven multimodal analysis integrating WSI, Methyl-Seq, and OncoKids Cancer Panel to improve diagnostic precision of pediatric tumors.
Previously, we added clinical and genomic data from more than 1,000 pediatric cancer patients from our racially and ethnically diverse patient population in Southern California to the CCDI (dbGaP: phs002518). The goal of our current P30 Supplement Project is to augment the CHLA dataset with new types of data from our unique patient population, and to develop tools for integrating the diverse datasets in CCDI to improve diagnosis and treatment of all children with cancer. To date, we have identified, screened, and selected the key whole slide images (WSI) from almost 700 CNS and non-CNS solid tumors in our cohort. We have developed a computational pipeline to access and organize metadata for each WSI. Additionally, our pipeline includes an analytical workflow to quantitatively assess WSI slides. In parallel, we have also generated whole-genome enzymatic methyl-seq data from 170 CNS tumors and developed a methyl-seq bioinformatics pipeline with a CNS tumor classifier compatible with methyl-seq data. We are actively developing tools to leverage these multimodal data for improved diagnosis and characterization of the tumors. Our pipeline is fully automated from raw bioinformatics processing to a classification report detailing predicted classification, classification score, UMAP clustering analysis, copy number profiling, and quality control metrics. Additionally, we have developed AI/ML tools to converge results from WSI analysis, methylation values, and OncoKids Cancer Panel results to enhance diagnostic decision-making. Our long-term goals are to incorporate additional data formats into our suite of multimodal analytic tools and to make these tools available to the CCDI community.
  1. First Name: Alexander
    Middle Initial: L.
    Last Name: Markowitz
    Degree(s): Ph.D.
    Organization: Children's Hospital Los Angeles, Keck School of Medicine, University of Southern California
  2. First Name: Jennifer
    Middle Initial: A.
    Last Name: Cotter
    Degree(s): M.D.
    Organization: Children's Hospital Los Angeles, Keck School of Medicine, University of Southern California
  3. First Name: David
    Middle Initial: N.
    Last Name: Buckley
    Degree(s): Ph.D.
    Organization: Children's Hospital Los Angeles
  4. First Name: Shengmei
    Last Name: Zhou
    Degree(s): M.D.
    Organization: Children's Hospital Los Angeles, Keck School of Medicine, University of Southern California
  5. First Name: William
    Last Name: Mango
    Organization: Children's Hospital Los Angeles
  6. First Name: Bruce
    Last Name: Pawel
    Degree(s): M.D.
    Organization: Children's Hospital Los Angeles, Keck School of Medicine, University of Southern California
  7. First Name: Fariba
    Last Name: Navid
    Degree(s): M.D.
    Organization: Children's Hospital Los Angeles, Keck School of Medicine, University of Southern California
  8. First Name: Jaclyn
    Last Name: Biegel
    Degree(s): Ph.D., FACMG
    Organization: Children's Hospital Los Angeles, Keck School of Medicine, University of Southern California
  9. First Name: James
    Middle Initial: F.
    Last Name: Amatruda
    Degree(s): M.D.,Ph.D.
    Organization: Children's Hospital Los Angeles, Keck School of Medicine, University of Southern California
Alexander L. Markowitz
Children's Hospital Los Angeles, Keck School of Medicine, University of Southern California