Living Systematic Review

The Challenge

Living Systematic Reviews are designed to keep medical evidence continuously up to date, yet the workflows that support them remain slow, fragmented, and cognitively demanding. Reviewers must manually extract and verify data across thousands of PDFs using disconnected tools, leading to fatigue, delays, and increased risk of error. While AI has the potential to accelerate this process, existing solutions lack the transparency and traceability required for clinical trust. These challenges are not merely operational inefficiencies, they directly impact how quickly validated evidence reaches clinical practice.

Why this matters?: In healthcare and research settings, speed without trust is unusable. If reviewers cannot clearly verify where data comes from or how it was generated, AI-assisted workflows fail to gain adoption, regardless of efficiency gains.

What I Set Out To Learn

Rather than jumping to interface solutions, we framed our work around three key questions:

1. "Where does cognitive load peak during the extraction workflow?"

2. "What prevents reviewers from trusting AI-assisted tools?"

3. "How can Human–AI collaboration remain flexible without disrupting existing mental models?"