SBIR Phase II:Building the digital twin of radiology operations

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) Phase II project is to improve the utilization of medical imaging equipment and potentially increase access to medical imaging for the population. Today, medical imaging facilities operate expensive equipment but lack access to operational data and modern tools to monitor and use them more efficiently. The company is building a digital twin of radiology operations to continuously capture imaging operations, monitor them, suggest optimizations, and optimally schedule patients. Beyond reporting and scheduling capabilities, the models will allow the prediction of interventions in software for the evaluation and comparison of different scenarios in-silico without real-world experimentation. The more efficient use of scanners is expected, in turn, to potentially benefit the patient population as it will reduce the wait time for imaging, increase patient access, shorten imaging protocols, reduce sedation duration, reduce and predict delays, and ultimately improve the patient experience. The data unified in the digital twin will also open new avenues of research for radiologists and researchers. _x000D_ _x000D_ The proposed project aims at developing and testing key technological innovations underpinning our digital twin vision. The company will develop a generic architecture to harmonize data across many sources, including from the scheduling system and the scanners themselves. The company will develop and test new artificial intelligence (AI) techniques to augment the data and unlock essential descriptors to manage operations. This solution will include AI to passively learn imaging protocols from the patients’ exams and automatically detect protocol deviations. The company will also use AI to automatically characterize the content of images and enable them to be queried. The company will evaluate federated learning techniques to allow learning “at the edge” on large datasets at scale, without sharing the data, alleviating data privacy challenges. The company will incorporate advances into a smart recommendation engine that continuously mines customer data to identify opportunities for improvement and proposes interventions. Finally, the company will develop and test key building blocks for implementing a smart scheduling assistant that uses retrospective data and digital simulations to optimally schedule exams while maximizing equipment utilization._x000D_ _x000D_ This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.