Long Article

The Value of Decision Support Software for Cancer Patients

As caring for patients with cancer has become more complex, multidisciplinary management has become standard. The regular meeting of clinical care teams with multiple specialist physicians in different disciplines, so-called “tumor boards,” allow the physicians to collaboratively review individual cases and make decisions about their oncology care. Tumor boards often consist of an oncologist, pathologist, radiologist, surgeon, radiation oncologist, social worker, nurse, and possible other members of the care team. A holistic understanding of the individual patient’s clinical picture and objective data is important for the clinical care team to come to  the best decisions about each patient’s care.1

With the complexity of medical data about each patient that tumor boards must assemble, present, digest, and analyze, there is a growing need for faster, more accurate and more targeted decision-making tools in oncology care.Precision medicine within oncology demands this level of complexity due to its data-driven nature.

Enter decision support software: assistive technology to aid clinicians who are facing the cognitive challenges of making clinical decisions in this world of complex, multidisciplinary data. According to a study in JCO Clinical Cancer Informatics, decision support systems are “one of the greatest potential benefits of a digital health care ecosystem.”2

The sheer amount and complexity of data pushes the limits of the human cognitive capacity to make decisions, which is generally limited to managing a maximum of five variables at a time. Because precision medicine requires the use of biomarkers to monitor and manage the disease, such complexity is baked into it.2

A recent survey by the American Society of Clinical Oncology revealed that physicians cited a need need for improving the infrastructure of tumor boards, including the use of advanced systems to aid in documentation. This is because the preparation process and workflow for tumor boards is often onerous and time-consuming.3

The Potential of Decision Support Software

While human intelligence is far superior to artificial intelligence, decision support software and systems can serve as a scaffold to human decision-making, much as AI assists pilots. AI can offer powerful tools to help analyze and classify data, and it can streamline and support workflow as the clinical care team assembles healthcare information.2

As the care team prepares for tumor boards, they must synthesize information that typically resides in isolated databases in various hospital systems. The data must be pulled from each of these systems and compiled in a way that the tumor board can share and view the information with each other. According to a study in the Journal of Pathology Informatics, currently, each clinician assembles the data in isolation from other clinicians prior to the tumor board convening. “This creates challenges such as potential miscommunication, overlooked or duplicate information, or not using the most current information,” the authors write. Such inefficiencies can lead to slowdowns in decision making as well as a greater burden on the clinical care team.1

There is a growing body of research that strongly suggests that health information technology such as decision support software can alleviate these slowdowns and inefficiencies, streamlining the process of decision-making in oncology, and allowing tumor boards to function more efficiently and accurately.1

Rapid-learning health care, the reuse of health care data from clinical practice or clinical trials to support health care decision making, holds great potential for transforming the delivery of health care. Rapid-learning health care (RLHC) is made up of four steps: data, knowledge, application, and evaluation. These four steps happen consecutively and are repeated infinitely as models for decision support systems are developed and validated. Data includes the procuring and mining of data. The knowledge step utilizes AI to extract information from the data via machine-learning algorithms which can be used to predict clinical outcomes for various treatments on future patients. The application step is where the knowledge gleaned from the data is used for decision making. The final step, evaluation, is where the performance and outcome of decision making support software is measured.2

Any RLHC system must optimize the “four Vs”—veracity, velocity, variety and volume—of big data to realize its full potential. Veracity speaks to the quality of knowledge distilled from data. Velocity refers to how rapidly this distillation happens. Variety of data allows for the evaluation of a number of different treatment types. Similarly, volume of data is important in terms of the predictive power of the knowledge it yields.2

The Current Limits of Decision Support Software

Like any AI, decision support software has some limitations. Because human intuition does not play into AI decision-making, many view it as a black box and thus feel uneasy about AI-driven decisions. Newer AI tools often aim to mitigate this with transparency, but as decision support software is developed, it is important that clinicians and researchers who are developing the solutions collaborate and engage so that the software provides the most tailored and biggest benefit to decision-making.2

A recent study in the Journal of the National Comprehensive Cancer Network concluded that there is a lack of data evaluating the outcomes of utilizing decision support software in oncology care, though the available data does suggest that it can have a “positive impact on the quality of cancer care delivery,” write the authors. They concluded that future studies should systematically evaluate decision support systems to provide the data needed to continue to develop systems that lead to better patient outcomes.4

The Future of Decision Support Systems in Oncology Care

Decision support software has the potential to streamline decision-making and information-sharing for tumor boards in cancer care, leading to potentially better patient outcomes. As health care shifts toward a more patient-centered approach, decision support systems can also facilitate communication between patients and clinicians, allowing for shared decision making. As artificial intelligence matures, decision support systems will take on more and more powerful tasks. Current decision support systems have been shown to improve care process measures, but more data is needed to fully assess how current systems are working and to help identify the strongest needs for continuing to develop AI-based supports for the clinical care team.4

 

References

1. A New Software Platform to Improve Multidisciplinary Tumor Board Workflows and User Satisfaction: A Pilot Study. Journal of Pathology Informatics. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6106126/?report=reader Last Accessed November 21, 2019.

2. Decision Support Systems in Oncology. JCO Clinical Cancer Informatics.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6563614/ Last Accessed November 21, 2019.

3. Global Practice and Efficiency of Multidisciplinary Tumor Boards: Results of an American Society of Clinical Oncology International Survey. Journal of Global Oncology. https://www.ncbi.nlm.nih.gov/pubmed/28804774/ Last Accessed November 21, 2019.

4. A Systematic Review of Clinical Decision Support Systems for Clinical Oncology Practice. Journal of the National Comprehensive Cancer Network. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6563614/ Last Accessed November 21, 2019.