# One Score, Two Decisions: Selective Prediction on the Rare-Disease Tail

arXiv: https://arxiv.org/abs/2608.14683 · HTML: https://mlsanity.com/paper/2608.14683/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 40 of its 91 references that are in the map.
- First posted on arXiv 5 Aug 2026.

## Builds on (most cited first, 30 of 40)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [On Calibration of Modern Neural Networks](https://mlsanity.com/paper/1706.04599/) (2017)
- [Qwen3-VL Technical Report](https://mlsanity.com/paper/2511.21631/) (2025)
- [A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks](https://mlsanity.com/paper/1610.02136/) (2016)
- [Large Language Models Encode Clinical Knowledge](https://mlsanity.com/paper/2212.13138/) (2022)
- [Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift](https://mlsanity.com/paper/1906.02530/) (2019)
- [When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories](https://mlsanity.com/paper/2212.10511/) (2022)
- [DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence](https://mlsanity.com/paper/2606.19348/) (2026)
- [Selective Classification for Deep Neural Networks](https://mlsanity.com/paper/1705.08500/) (2017)
- [Fact or Fiction: Verifying Scientific Claims](https://mlsanity.com/paper/2004.14974/) (2020)
- [ColBERTv2: Effective and Efficient Retrieval via Lightweight Late Interaction](https://mlsanity.com/paper/2112.01488/) (2021)
- [Adaptive-RAG: Learning to Adapt Retrieval-Augmented Large Language Models through Question Complexity](https://mlsanity.com/paper/2403.14403/) (2024)
- [BioMistral: A Collection of Open-Source Pretrained Large Language Models for Medical Domains](https://mlsanity.com/paper/2402.10373/) (2024)
- [Consistent Estimators for Learning to Defer to an Expert](https://mlsanity.com/paper/2006.01862/) (2020)
- [HuatuoGPT-o1, Towards Medical Complex Reasoning with LLMs](https://mlsanity.com/paper/2412.18925/) (2024)
- [Selective Question Answering under Domain Shift](https://mlsanity.com/paper/2006.09462/) (2020)
- [Self-Alignment Pretraining for Biomedical Entity Representations](https://mlsanity.com/paper/2010.11784/) (2020)
- [To Trust Or Not To Trust A Classifier](https://mlsanity.com/paper/1805.11783/) (2018)
- [Don't Hallucinate, Abstain: Identifying LLM Knowledge Gaps via Multi-LLM Collaboration](https://mlsanity.com/paper/2402.00367/) (2024)
- [Head-to-Tail: How Knowledgeable are Large Language Models (LLMs)? A.K.A. Will LLMs Replace Knowledge Graphs?](https://mlsanity.com/paper/2308.10168/) (2023)
- [Reliable Visual Question Answering: Abstain Rather Than Answer Incorrectly](https://mlsanity.com/paper/2204.13631/) (2022)
- [MedMentions: A Large Biomedical Corpus Annotated with UMLS Concepts](https://mlsanity.com/paper/1902.09476/) (2019)
- [Cross Modal Retrieval with Querybank Normalisation](https://mlsanity.com/paper/2112.12777/) (2021)
- [Investigating Selective Prediction Approaches Across Several Tasks in IID, OOD, and Adversarial Settings](https://mlsanity.com/paper/2203.00211/) (2022)
- [When Does Confidence-Based Cascade Deferral Suffice?](https://mlsanity.com/paper/2307.02764/) (2023)
- [When Do LLMs Need Retrieval Augmentation? Mitigating LLMs' Overconfidence Helps Retrieval Augmentation](https://mlsanity.com/paper/2402.11457/) (2024)
- [Baichuan-M2: Scaling Medical Capability with Large Verifier System](https://mlsanity.com/paper/2509.02208/) (2025)
- [An Agentic System for Rare Disease Diagnosis with Traceable Reasoning](https://mlsanity.com/paper/2506.20430/) (2025)
- [RareBench: Can LLMs Serve as Rare Diseases Specialists?](https://mlsanity.com/paper/2402.06341/) (2024)

## Abstract (arXiv)

Given a patient's clinical findings, a diagnostic system ranks possible diseases and must decide when to endorse its first prediction or defer it for review. This decision is usually made by thresholding the top score. Selective prediction over ranked outputs begins with two checks. First, the ranker must produce enough correct top-ranked predictions to make the target feasible. Across 2,000 patient records stratified by disease prevalence, eight small open-weight LLMs achieve at most 4.6% Recall@1 on ultra-rare diseases. At 10% coverage, even a perfect confidence ranking of their existing predictions therefore cannot reach 50% selective accuracy. More accurate models pass the same check, showing that the limit is regime-specific. Second, the confidence signal must match the decision being made. For fixed-candidate rankers, the top-two margin cancels components shared across candidates. On phenotype-only Exomiser, it selects 10% of cases at 29.0% accuracy, compared with 13.3% overall, while the top score provides no reliable gate. Yet that cancellation can remove information needed to detect whether the candidate list contains an answer. SciFact retrieval and biomedical entity linking confirm this distinction. Finally, we prove that unlabelled scores alone cannot determine whether switching to the margin will help.