Leveraging AI in Radiation Oncology: What You Need to Know
The radiation oncology (RO) market continues to change rapidly. Efforts with the goals of improving the effectiveness of treatment, while reducing costs, and preventable treatment errors is driving the new capabilities into both delivery and treatment planning systems. In a change from recent years, a very large percentage of product value creation is coming from the development of sophisticated software capabilities, in yet another example of the inexorable expansion of digital health into the various domains of the healthcare industry.
These new software capabilities have the potential to meet all three of the above goals, but they may tax the design limits of the current generation of products. So, the question for OEMs is two-fold: can my system architecture expand to include this new functionality and, if so, what is the most minimally invasive approach to re-architecting my product to include the new technology.
The Unique Challenge
Treatment planning systems (TPS) are complex software systems fully integrated with other parts of the RO ecosystem, both diagnostic systems upstream and treatment delivery systems down. The rapidly evolving artificial intelligence (AI) capabilities are having a significant impact on the TPS sector and represent one of the most difficult classes of technology to incorporate: fast moving technology into a slower moving, highly regulated sector.
AI functionality is improving at an exponential rate, a rate that is difficult for even those dedicated to the technology to fully comprehend. The change in engine capability is measured in hundreds of percent per year – far outside the range of anything that has been integrated into healthcare systems in the past. TPS OEMs are faced with real-life challenges regarding how to effectively integrate these new AI capabilities into their existing products, when the new technology is evolving at such a rate that by the time they design in a new AI engine, it will most likely be obsolete.
TPS providers are faced with difficult decisions regarding evaluating not only the current state of different AI engines, but also predicting the rate of improvement of the various AI technologies. Betting on the wrong AI engine could have very negative impacts in competitive position in the very near future.
System Architecture Imperatives
This is a classic example of the need to approach a system architecture design in a manner that increases the ease with which apparently core capabilities can be excised and replaced, while minimizing the overall stress on the system design and reducing the ripple effects into other areas of the system.
Properly architected, a TPS OEM can put themselves into the position to be effectively “AI technology agnostic”, allowing the industry to continue to evolve at its own rate and periodically integrating “best in class” capabilities into their own product. In order to achieve this, the TPS provider must properly architect their software system now – but how?
Where to Begin
The first step is to define the new AI capabilities in a lexicon that aligns with their system design. The new functionality must fit into the existing system within the context of the current design. The AI capabilities will, most likely in a incremental manner, at first augment, then ultimately replace current analytical capabilities. By aligning the new functionality to the design concepts of the existing system it is possible to approach the redesign in a minimalist fashion.
The next step requires assessing the existing TPS software system from an architectural perspective. Using techniques such as Architecture Tradeoff Analysis Methodology (ATAM) or similar, a system can be easily decomposed to illuminate design tradeoffs that may be either supportive or inhibitive to inclusion of new functionality.
This assessment will effectively create a roadmap to potential approaches to enhance the system design to support integration of an AI module (or various AI modules) into the system in a more ‘loosely coupled’ manner.
Allowing for each of the major components of treatment planning – beam pathway definition, dose control and fields optimization, for example – to have its own more robustly defined inputs and outputs, it is possible to wrap AI engines in a “generic AI wrapper” that can inure the TPS from the vagaries of specific AI implementations. As new AI approaches evolve, this approach supports the process of swapping one engine out, and a new engine in.
This approach also supports the ever-difficult problem of interoperability. Since its creation in 2004, the IHE-RO initiative has made large strides in improving overall interoperability of oncology diagnostic, TPS and treatment delivery systems, but more work needs to be done from an OEM system design perspective to further reduce interoperability-induced errors.
Inclusion of rapidly evolving AI engines provides an enormous opportunity and challenge to the RO industry. Properly approached, the challenge can provide a significant market advantage, especially in the mid-to-long term.
Assessing the new technology, to better understand its alignment with the industry lexicon and addressing specific interoperability limitations is critical to ensure integration in an interoperable manner.
Assessing existing systems to understand how best to incorporate, with the least structural dislocation will accelerate the new capabilities to market, reduce system stress and improve reliability.
Designing more expansive functional components will allow for the rapid improvement in AI technology and support rapid improvement of the healthcare systems as well.