Searcharxiv⌕ Search

arXiv subjects

Hannes Ill

Publications and source records attributed to Hannes Ill.

3 recordsLinked to original sources

SLICES: A Benchmark and Evaluation Framework for Self-Supervised Learning on Sparse ICU Time Series

Self-supervised learning (SSL) results for intensive care unit (ICU) time series are hard to compare when data processing, encoders, and adaptation vary together. We present SLICES, a benchmark and evaluation framework for comparing and extending SSL methods under a shared clinical setup. It fixes 24-hour value and observation-mask inputs, a Transformer encoder, patient-level splits, four endpoints, two adaptation protocols, and five matched seeds across MIMIC-IV, eICU-CRD, and their pooled union. Masked autoencoding (MAE), joint-embedding prediction (JEPA), and contrastive learning provide reference implementations. Among these objectives, MAE leads mean AUPRC in all nine frozen classification comparisons, also under converged probes, and in seven of nine fine-tuning comparisons. Fine-tuned MAE leads hospital mortality at full labels and scores higher than a supervised Transformer of the same architecture at every label fraction, and SSL fine-tuning gives the lowest length-of-stay error. Supervised baselines remain strong elsewhere: GRU-D leads 24-hour mortality and XGBoost leads acute kidney injury in all three regimes. Rankings depend on the adaptation protocol, the comparator set, and whether the probe loss matches the metric. A two-seed MIMIC-IV rerun with the maintained code stays within two across-seed SDs of the reference in all 48 downstream runs. The code repository supports new objectives, shared downstream evaluation, and regeneration of the reference results, so new SSL methods can be compared with these references under the same conditions. Generalization to unseen hospitals is not tested here and is the natural next extension.

cs.LG↗

M4Bench: Evaluating Procedural Specification for Clinical EHR Derivation Agents

Clinical research over electronic health records depends on derived variables whose definitions sit outside the schema, in item codes, time windows, missingness rules, and output-grain choices that change cohorts and downstream conclusions. We introduce M4Bench, 28 clinical derivation tasks across 15 task families on MIMIC-IV and eICU, to test whether AI coding agents can reproduce such derivations and how clinician-reviewed agent skills change their results. Its controls decompose the value of skills into specification completeness, delivery format, and schema familiarity. In a Docker-isolated Codex campaign (GPT 5.4 Mini and GPT 5.5), skills raise reward by 0.136 over the no-skill baseline, with 26 of 28 tasks improving. Specification completeness accounts for most of this gain: on an eight-task control subset, an operational spec alone raises reward by 0.162, and skills add only 0.024 on top, indistinguishable from zero. On a ten-task subset, decoy skills score like no skill and raw reference SQL nearly matches targeted skills, arguing against generic context or the skill format as the explanation. Schema obfuscation and restructuring preserve 86% and 70% of the skill gain (GPT 5.4 Mini, 26 MIMIC-IV tasks), a descriptive robustness check rather than a bound on memorization. A descriptive Claude Code replication on all 28 tasks (Opus 4.7 and Sonnet 4.6, 560 runs) is directionally consistent (+0.120, 21 of 28 tasks improving). These gains are partial agreement with the reference, not completed tasks: only 25 of 280 with-skill and 13 of 280 no-skill Codex runs pass every strict check, and skills barely change which rows agents return (key F1 +0.003). Skills are released as a versioned, clinician-reviewable markdown delivery format, with a family-level reference-validity audit. M4Bench releases tasks, skills, evaluation code, and sanitized per-run evidence as an open-source benchmark.

cs.AI↗

Numerical Literals in Link Prediction: A Critical Examination of Models and Datasets

Link Prediction(LP) is an essential task over Knowledge Graphs(KGs), traditionally focussed on using and predicting the relations between entities. Textual entity descriptions have already been shown to be valuable, but models that incorporate numerical literals have shown minor improvements on existing benchmark datasets. It is unclear whether a model is actually better in using numerical literals, or better capable of utilizing the graph structure. This raises doubts about the effectiveness of these methods and about the suitability of the existing benchmark datasets. We propose a methodology to evaluate LP models that incorporate numerical literals. We propose i) a new synthetic dataset to better understand how well these models use numerical literals and ii) dataset ablations strategies to investigate potential difficulties with the existing datasets. We identify a prevalent trend: many models underutilize literal information and potentially rely on additional parameters for performance gains. Our investigation highlights the need for more extensive evaluations when releasing new models and datasets.

cs.LG↗