AI Data Readiness Challenge: Revolutionizing Cancer Research with Enhanced Data

The National Cancer Institute (NCI) is dedicated to advancing cancer research through innovative data science initiatives. A major challenge is ensuring data are AI-ready, meaning they are well-structured, annotated, and accessible for machine learning (ML) and artificial intelligence (AI) applications. To address this, the NCI's Cancer Research Data Commons (CRDC) launched an AI Data Readiness Challenge, incentivizing researchers to develop solutions for improving the AI readiness of NCI CRDC data and components.

This article explores the significance of AI data readiness in cancer research, highlighting the winners of the NCI CRDC AI Data Readiness Challenge and their groundbreaking projects.

Why AI Data Readiness Matters

The success of AI and ML models hinges on the quality and accessibility of the data they are trained on. In cancer research, where data isvast and complex, ensuring data is AI-ready is crucial for:

• Accelerating discovery: AI-ready data enables faster and more efficient analysis, uncovering patterns and insights that might be missed by traditional methods.
• Improving accuracy: High-quality, well-annotated data leads to more accurate and reliable AI models, which can be used for diagnosis, treatment planning, and predicting patient outcomes.
• Enhancing collaboration: Standardized and accessible data promotes collaboration among researchers, facilitating the sharing of knowledge and resources.

NCI CRDC AI Data Readiness Challenge: A Catalyst for Innovation

The NCI CRDC AI Data Readiness Challenge was designed to encourage the development of innovative solutions for enhancing the AI readiness of cancer research data. Participants were tasked with defining AI data-readiness metrics and using them to preprocess data from one or more data commons for AI/ML model training. The challenge was divided into two tiers:

• Tier 1: Training an AI/ML model with single modal data.
• Tier 2: Training an AI/ML model with multi-modal data.

The winners of the challenge were announced at a special presentation, where they showcased their projects.

Tier 1 Winners: Single Modal Data Solutions

1st Place: Jennifer Blasé (Ruvos)

• Project: "Gene expression-based prediction of treatment response in ovarian cancer."
• Significance: This project focuses on using gene expression data to predict how ovarian cancer patients will respond to different treatments. By identifying predictive biomarkers, this solution has the potential to personalize treatment strategies and improve patient outcomes.

2nd Place: Agnes McFarlin

• Project: "Identifying cancerous lung nodules without the presence of annotated slides for reference."
• Significance: This project addresses the challenge of identifying lung nodules in the absence of annotated slides, which are often used as reference. This could streamline the diagnostic process and improve early detection of lung cancer.

Tier 2 Winners: Multi-Modal Data Solutions

1st Place: Abhishek Jha (Elucidata)

• Project: "Distinguishing primary tumor from normal solid tissue in lung squamous cell carcinoma."
• Significance: This project focuses on differentiating between primary tumors and normal tissue in lung squamous cell carcinoma using multi-modal data. By integrating different types of data, such as genomic, transcriptomic, and imaging data, this solution can improve the accuracy of diagnosis and treatment planning.

2nd Place: Jeff Van Oss (BAMF Health)

• Project: "Predicting Von Hippel-Lindau mutation in kidney tumors using radiomic features."
• Significance: This project explores the use of radiomic features to predict Von Hippel-Lindau (VHL) mutations in kidney tumors. Radiomics involves extracting quantitative features from medical images, which can provide valuable insights into tumor characteristics and genetics.

The Impact of the AI Data Readiness Challenge

The NCI CRDC AI Data Readiness Challenge has had a significant impact on the cancer research community, including:

• Raising awareness: The challenge has highlighted the importance of AI data readiness and encouraged researchers to prioritize data quality and accessibility.
• Fostering innovation: The challenge has spurred the development of new tools and techniques for improving the AI readiness of cancer research data.
• Driving collaboration: The challenge has brought together researchers from different disciplines to collaborate on solutions for advancing cancer research.

The results from the challenge will help the CRDC better meet the needs of AI-based research and accelerate the development of new cancer diagnostics and treatments. The NCI's commitment to data sharing and collaboration, as evidenced by initiatives like the Genomic Data Sharing (GDS) Policy, further supports these advancements.

The Future of AI in Cancer Research

The NCI CRDC AI Data Readiness Challenge is just one example of the many ways that AI is transforming cancer research. As AI and ML technologies continue to evolve, they have the potential to revolutionize cancer diagnosis, treatment, and prevention. By ensuring that cancer research data is AI-ready, we can unlock the full potential of these technologies and accelerate progress towards a future without cancer.

To stay updated on the latest advancements in cancer data science, consider subscribing to updates from the NCI and following the NCI Cancer Data Science community on LinkedIn.