The Sports industry is undergoing a dramatic transformation driven by two technology trends: Artificial Intelligence and software-defined solutions are redefining the medical imaging workflow.
Artificial Intelligence, specifically deep learning, demonstrates great potential within sports for specific player action detection, localization, and classification. It has already shown it can augment humans by increasing their efficiency and effectiveness by prioritizing the most severe or time-sensitive moves. These efficiencies save teams time and money in training, focusing valuable resources on getting better in the sport.
Deep learning research in the field of sports imaging is booming, however, most of this research today happens in isolation and with limited datasets. This leads to overly simplified and unscalable models that only have high accuracy for a certain demographic and set of imaging devices. At the same time, smaller organizations lack the opportunity to provide a higher quality of training for their local population due to a lack of technical expertise, resources, and the sheer data volumes required for deep learning.
Even when creating effective deep learning algorithms, the challenges of integration and scalability in deploying these intelligent algorithms into clinical workflows seem daunting.
The Clara platform addresses these challenges by lowering the bar to adopting AI in sports training and game workflows. The Clara Train SDK provides transfer learning and AI-assisted annotation capabilities, enabling faster data annotation and adaptation of a neural network from the source domain to a target domain. Once a quality-assured neural network becomes available, the Clara Deploy SDK covered in this post provides the framework and tools required to develop an application workflow capable of integrating into existing hospital infrastructure.
The Clara Deploy SDK provides an industry-standard, container-based development, and deployment framework for building AI-accelerated medical imaging workflows. The SDK uses Kubernetes under the hood, providing developers and data scientists with the ability to define a multi-staged container-based pipeline. This modular architecture, shown in figure 1, allows developers to use the offerings of the platform out-of-the-box with minimal customization or create new workflows with bring-your-own algorithms.