How Good Are Large Language Models at Supporting Frontline Healthcare Workers in Low-Resource Settings – A Benchmarking Study & Dataset
Bibliographic record
Abstract
Abstract Large language models (LLMs) have demonstrated strong performance in medical contexts; however, existing benchmarks often fail to reflect the real-world complexity of low-resource health systems accurately. This study developed a dataset of 5,609 clinical questions contributed by 101 community health workers (CHWs) across four Rwandan districts and compared responses generated by five large language models (LLMs) (Gemini-2, GPT-4o, o3 mini, Deepseek R1, and Meditron-70B) with those from local clinicians. A subset of 524 question-answer pairs was evaluated using a rubric of 11 expert-rated metrics, scored on a five-point Likert scale. Gemini-2 and GPT-4o were the best performers (achieving mean scores of 4.49 and 4.48 out of 5, respectively, across all 11 metrics). All LLMs significantly outperformed local clinicians (ps < 0.001) across all metrics, with Gemini-2, for example, surpassing local GPs by an average of 0.83 points on every metric (range: 0.38 – 1.10). While performance degraded slightly when LLMs communicated in Kinyarwanda, the LLMs remained superior to clinicians and were over 500 times cheaper per response. These findings support the potential of LLMs to strengthen frontline care quality in low-resource, multilingual health systems.
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 | 0.012 | 0.030 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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".