Deploying an artificial intelligence-based online search tool to increase patients' access to and understanding of solid tumor gastrointestinal clinical trials

Abstract

Background: The number and complexity of clinical trials is growing, making finding and accessing trials increasingly challenging. A key barrier to increasing recruitment efficiency is patients' low awareness of trials as an option. They rarely hear about trials from oncologists, unless these are based at a research center, which introduces trial population biases. Even if patients decide to search for trial information on their own, it is difficult for laypeople to understand. Trial search tools exist, but have multiple shortcomings: complex navigation, limited search functionality, data too complex for non-specialists, and next steps unclear.

Methods: Access to trials via a novel trial search tool using technology, including artificial intelligence (AI), to restructure trial information and match patients to trials was the focus of this study. We focused on 6 gastrointestinal cancers for evaluating the search tool's AI model in which their trial information was restructured based on key eligibility differentiators identified. Research participants were recruited with the support of patient associations leveraging social media. They were asked to complete a survey evaluating an established tool and the novel tool to find trials relevant for them. They assessed aspects of the tools' usability on 5-level Likert scales.

Results: Survey respondents had a base level of awareness of trials, with patient associations and the internet as leading information sources. The novel tool made it easier to find trials, led to an improvement in ease of understanding information presented and provided more clarity on what the next steps towards enrollment would be. Overall, this led to higher patient satisfaction.

Conclusions: Novel patient-focused tools can unlock trial access for all patients. Democratizing trial information will not only increase trial accrual, but also patient satisfaction while reducing disparities. This tool could also be of value for academic or community oncologists who face similar issues when navigating trials for patients.

Keywords: Artificial intelligence (AI); cancer disparities; clinical trials; health equity; patient access.

Figure 1

Wide variability of initial trial knowledge amongst participants. Participants were asked to rate “your current level of awareness of your clinical trial options on scale from 1 (not at all aware) to 5 (extremely aware)”.

Figure 2

Patient groups and internet as most important current sources for trial information. Participants were asked to select all applicable from a given set of options, including an “other option”.

Figure 3

Increased level of performance on a number of aspects of novel tool vs. currently available tool. Gastrointestinal cancer patients reviewed each tool in turn (sequence randomized) and then assessed these 4 aspects on 5-point Likert scale. Statistical significance was assessed by using a paired two-tailed Student’s t-test (*, P≤0.05; **, P≤0.01).

Figure 4

Faster identification of relevant trials with novel tool vs. currently available tool. Participants reviewed each tool in turn and then estimated the time in minutes it took to find relevant trials. Statistical significance was assessed by using a paired two-tailed Student’s t-test (*, P≤0.05).

Figure 5

Increased level of performance on a number of aspects of novel tool vs. currently available tool. Non-gastrointestinal cancer patients reviewed each tool in turn (sequence randomized) and then assessed these 4 aspects on 5-point Likert scale. Statistical significance was assessed by using a paired two-tailed Student’s t-test (ns, P>0.05; *, P≤0.05; **, P≤0.01; ***, P≤0.001).