The Challenge: Moving ahead while maintaining control and credibility.
Clinical research is data-heavy by design and accuracy is non-negotiable. Each trial can produce
millions of data points, and every outlier and adverse event must be checked, validated, and documented. For years, Bioforum relied on a gold-standard process called “double programming” (i.e. two independent teams building identical analyses to ensure eachresult matches, and that the conclusion is correct).
But as workloads mounted, the time cost became difficult to justify. The team at Bioforum realised there was potential for AI to improve that capacity by half, thereby also halving the workload. In other words, AI would not replace human oversight but instead, act as a second set of eyes.
The Outcome: From concept to completion.
The first major outcome for Bioforum was “Brian” (An acronym for “Building Reports In AI Now”). This refers to Bioforum’s AI assistant, created to help clients explore clinical data on
demand.
Instead of waiting for custom dashboards, clients can now ask questions directly, e.g. “Show me the adverse events for Site 7.” The assistant generates a visualisation instantly, complete with
a short narrative summary. If they want to revisit it, they can pin it to their dashboard for future use. This saves everyone time. Bioforum’s developers no longer have to code and validate each new widget, because the AI assistant is already validated. Clients get answers on demand, and the Bioforum team can focus on higher-value analyses.
