HTAi Annual Meeting 2026

June 6, 2026

Evidence Prime took part in the HTAi 2026 Annual Meeting, held in Istanbul, Türkiye, from 6 to 10 June. The theme, “HTA as a system shaper,” placed health technology assessment within a wider discussion about how health systems learn, set priorities and use evidence to guide decisions.

Transparent table extraction for HTA

On 8 June, Artur Nowak presented “A Two-Stage Human-In-The-Loop AI Approach For Transparent Table Extraction In HTA” in the Evidence Retrieval and Synthesis track.

The contribution addressed a practical difficulty in evidence review: complex publication tables contain relationships between study arms, outcomes, time points, units and uncertainty that need to survive extraction. A list of plausible values is not sufficient if reviewers cannot establish what those values describe or where they came from.

The approach separated reconstruction of the source table from entry into the extraction form. Vision-based models reconstructed the table structure, while reviewers mapped specific cells to the required fields and validated the data. Keeping those steps distinct made human review an explicit part of the workflow, with the source table available for checking.

From faster tasks to better evidence processes

Artur also participated in the Data & AI Interest Group. In his reflections after the meeting, he considered how AI might help teams redesign evidence processes, including more continuous retrieval, earlier handling of uncertainty and documentation that remains useful when work passes from researchers to decision-makers.

For Laser AI, these discussions connect methodological design with software design. Reviewers need ways to inspect and correct extracted evidence, understand unresolved cases and retain responsibility for decisions. The work presented in Istanbul explored how those requirements can be built into a specific extraction workflow; it does not establish that every review or document type will achieve the same results.

The table-extraction presentation was co-authored by Artur Nowak, Ewa Borowiack, Monika Opalek and Ewelina Sadowska.

HTAi 2026 Annual Meeting

‍

Blank white image with no visible content or details