Topics of Special Interest for the ICI Grants

Of interest are both research that leads to the creation and analysis of new data resources of genomic, imaging, microbiome, metabolic, clinical and other data from cancer and pre-cancer patients and the analysis of current data resources. Topics include data and analysis to increase understanding of the etiology and progression of cancer, the comparative effectiveness of approved treatments, single and combinations of targets for new treatments, and hypotheses about somatic or germline genetic markers, lifestyle factors and concomitant drugs predictive of recurrence and response to therapy. The focus should be on data resources and computational tools that are likely to have near or intermediate-term clinical impact. Of particular interest are the creation of AI-ready data resources and methods, including clinically relevant agent and world models.

Projects aiming to substantially enhance annotation of the functional and clinical implications of genetic, epigenetic, gene expression, and other changes available in large cancer genomics datasets. For example, computational tools that extract functional annotation from textual sources, functional genomics or large-scale human genetics datasets, data-rich clinical trajectories; or computational tools that facilitate crowd-sourcing of expert annotation by the biological and clinical research communities with built-in reliability metrics and quality control. The primary focus should be on clinically actionable functional annotation and prediction tools with the aim to support data-driven decision support in clinical practice.

Development of AI models, tools, and agents to improve the outcomes of cancer patients through advances in imaging informatics, including digital histopathology and radiology. Methods and models can include diagnostics, lesion detection, outcome prediction, and others.

Projects that advance the exploration, implementation, and assessment of potential clinical utility of liquid biopsies to track the temporal evolution of a patient's disease in support of more flexible adjustment of treatment options. Examples might include developing tools or computational methods to aggregate and harmonize data about a) circulating tumor cells, exosomes, cell-free tumor DNA (mutations, methylation, fragmentation), metabolites, tumor-specific plasma protein profiles, tumor-associated autoantibodies, b) clinical diagnoses, treatment history, and outcomes, and c) sample collection, preparation, and handling protocols.

Projects that focus on the development of software tools or data infrastructure that can improve data collection, data processing, data sharing, or facilitate decision-making in the treatment of cancer patients. Examples might include tools that can improve tumor board expertise, capture and annotate tumor board discussions and outcomes, share treatment and outcomes data, aid recruitment to focused clinical trials, or provide resources for physicians at non-cancer centers to access the expertise and informed treatment decisions of larger, specialized treatment centers.

Projects that focus on data analysis of electronic medical records or other health systems records (including genetic, molecular, and imaging profiles of early-stage cancerous and pre-cancerous tissue) to improve early detection, create personalized prevention programs, facilitate anti-cancer vaccines, or improve treatment. Examples might include the use of electronic health records to better target screening programs to the optimal patient population, identify obstacles to treatment compliance for patients, and identify early non-invasive biomarkers from easily accessible clinical data for early detection.

Projects that develop applications by which patients can record detailed health information and provide it to trusted partners, such as their physician, to improve their own treatment. For example, cancer patients would record how they are responding to treatment and information about their state of health on a smartphone on a daily or weekly basis and the encrypted information would be shared via a trusted repository. Patients would own their data, remain in complete control of its use, share their data with their oncologist, and, via an opt-in mechanism, provide it for aggregated use in cancer research using advanced data science and machine learning methods.