Co‐designing a Large Language Model Benchmarking Dataset for Primary Care with Nurses in Kenya
Bibliographic record
Abstract
Abstract Large Language Models (LLMs) are increasingly applied in healthcare, yet their training and evaluation often lack grounding in frontline realities in low‐resource settings. By grounding content in nurses' everyday practice, this work contributes a localized benchmark for LLM training and evaluation and offers a replicable model for ethical, inclusive AI design responsive to care realities in resource‐constrained environments. It documents the participatory co‐design, curation, and descriptive characterization of a nurse‐generated dataset for LLM benchmarking in primary healthcare (PHC) in Kenya. Using human‐centred design methods, we trained 145 nurses across three counties to generate real‐world clinical scenarios and questions using an adapted SBAR (Situation, Background, Assessment, Recommendation) framework. Through workshops, audio recording and digital submissions, nurses contributed 7,606 scenarios. These scenarios captured decision‐making needs spanning clinical management, referral, communication/counselling, and constraints in diagnostics, equipment, and social context typical of PHC. This article details the co‐design process, data pipeline, and dataset descriptives; benchmarking methods and results using this dataset are reported separately.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.010 | 0.039 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".