Growing Startup Has Roots in ROI Grant
An Update on Oncospace from Todd McNutt, PhD
The Oncospace research program began in 2007, seeking to transform clinical practice by capturing relevant data during the routine clinical workflow. Oncospace is an analytics platform that enables quality control, clinical decision support, and hypothesis generation for research. Essentially, it is a localized learning health system for radiation oncology. During the early stages of the program, the Oncospace team at Johns Hopkins University (JHU) focused on radiation therapy treatment plan automation and quality control. As the program matured and more clinical assessments of patients’ outcomes became available, our research initiatives moved into applying big data analytics to treatment related toxicities. This way, we could reveal the details of the radiation distribution that have the most impact on specific toxicities.
In 2017, the Oncospace team was awarded funding from the Radiation Oncology Institute (ROI) to examine how the spatial patterns of dose affected three common side effects in head and neck cancer patients receiving radiation therapy: xerostomia, dysphagia, and weight loss. The project provided us with a more in-depth look at how the dose is distributed in the patient’s anatomy through the generation of several spatially dependent radio-morphologic features of the 3D dose, taking us far beyond the traditional dose-volume histogram-based metrics. This work allowed us to see precisely where inside different critical organs the dose had the greatest impact on each toxicity. We applied this in-depth analysis to the major salivary glands and esophagus. We also looked into radiomic features of the same anatomy using CT and MRI to include features that may represent a patient’s predisposition to radiation-induced toxicity.
Through these studies, we have learned the value of big data in advancing clinical practice, and the cost-reduction benefits of data collection within routine clinical workflow. The technology development performed over the years to support the Oncospace program has shed light on how to improve routine data capture while also improving clinical communications. In addition, it has shown us that good management of clinical data provides a platform for increased quality control in which new cases can be evaluated in the backdrop of the patients treated before them. This technological framework can help to identify outliers that pose risks and to inform when a particular patient’s treatment plan is of the highest quality. Beyond quality, the research supported by the ROI helped make tools for clinical decision support, such as identifying patients at risk for severe toxicities, feasible in a clinical setting.
In order to expand the concepts of the Oncospace program, a commercialization pathway was considered most appropriate. Oncospace Inc. was formed and started operations on April 1, 2019. This new startup offers the opportunity to expand the learning health system more broadly. The newly formed company was awarded an SBIR Phase I grant for automated peer review from the National Science Foundation (NSF) and an SBIR Phase I contract for AI-driven radiation treatment planning for prostate cancer from the National Cancer Institute (NCI). The NSF is considering Phase II SBIR funding to advance the concept of peer review to include predicted treatment plan quality from the data for the most common treatment sites in radiation oncology. The model seeks to provide physicians with the ability to evaluate the plan quality based on the prior history of like patients. After the successful completion of the SBIR Phase I contract on AI-driven treatment planning for prostate cancer, the company was invited by the NCI to submit a Phase II proposal. Oncospace, Inc. has also completed its target seed round funding goals.
The company’s technology is implemented on the Microsoft Azure cloud and has been designed based on the institutional sharing principles and the analytic environment produced during the development of the database at JHU. The system seeks to capture full treatment planning data in the clinical workflow while providing valuable clinical tools for peer review, quality control and automated planning. Oncospace Inc. aims to develop what may turn out to be the best model for disseminating advanced prediction capabilities with a complement of outcome data in the future.