Metraipolis › archive › dc_711089d08ef6

Towards a Deterministic Math Solver for Clinical Language Models — arXiv:2609.10728

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.10728
License (as archived)
CC0-1.0
Archived
2026-10-16T05:23:37.235Z
SHA-256 of the archived text
cc4287a15d70d6a3960f24823a36caa0c4b140f469d25b18477908bae798e2fc
Stable URL of this page
https://metraipolis.com/documento/dc_711089d08ef6

How to cite

Towards a Deterministic Math Solver for Clinical Language Models — arXiv:2609.10728 — arXiv (en), https://arxiv.org/abs/2609.10728 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_711089d08ef6 (Metraipolis archive dc_711089d08ef6, sha256 cc4287a15d70d6a3960f24823a36caa0c4b140f469d25b18477908bae798e2fc)
BibTeX
@misc{dc711089d08ef6,
  title = {Towards a Deterministic Math Solver for Clinical Language Models — arXiv:2609.10728},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.10728},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_711089d08ef6; sha256 cc4287a15d70d6a3960f24823a36caa0c4b140f469d25b18477908bae798e2fc},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.10728. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models are unreliable at arithmetic, which is a problem for clinical calculators where a single numerical error changes the recommendation. The standard response is to hardcode each calculator as a validated function, one at a time. We test an alternative: the model does not calculate. Instead, it writes case-specific Python that a restricted local executor runs as a deterministic solver, and the model's task reduces to deciding how to use it. We evaluate this Program-Solve interface on MedCalc-Bench Verified (1,100 cases, 55 calculators) against direct model arithmetic and a hand-written 22-calculator library, using Qwen2.5-7B and Qwen2.5-32B-AWQ, after auditing the benchmark's formulas against current clinical guidelines and flagging 16 of 55 with version, use or coefficient concerns. With formulas and gold variables supplied and both routes reading the whole note, handing off to the solver is not a reliable advantage at 7B (75.31% against 72.02%, a paired +3.29 points with a 95% calculator-cluster interval of [-3.49, 10.38]) but is one at 32B (90.53% against 83.47%, +7.05 [0.47, 14.60], clear of zero). The hand-written library is exact on its 440 supported cases but abstains elsewhere (40.0% overall). Adding an executor thus helps some open-weight models more than others even under matched formula, variable and note access, and is not a substitute for verified formulas or reliable variable extraction either way.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-11 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Software Engineering (cs.SE).
Pagina del lavoro: https://arxiv.org/abs/2609.10728 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T05:17:51.952Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).