SpaceOmicsBench

A multi-omics AI benchmark for spaceflight biomedical data — 21 ML tasks across 9 modalities plus a 100-question LLM evaluation, using data from Inspiration4, the NASA Twins Study, and the JAXA CFE study.

ML Tasks

Modalities

100

LLM Questions

Baselines

LLM Models

Baseline Results

Performance across all tasks. The best score in each row is highlighted. E2 and E3 are supplementary tasks with extreme class imbalance.

Task	Name	Category	Tier	Metric	Random	Majority	LogReg	RF	MLP	XGB	LGBM

Performance Analysis

Normalized composite scores, category radar, and difficulty distribution.

Normalized Composite Score — (score − random) / (1 − random)

RF Category Performance — radar view

Difficulty Tier Distribution — 21 tasks

B1 Feature Ablation — AUPRC by feature set

Insight: Without effect-size features (fold changes), RF and MLP reach AUPRC values of 0.863 and 0.847, respectively, suggesting that performance is not driven solely by effect-size thresholding.

LLM Evaluation

A 100-question benchmark with five-dimensional scoring across nine models.

Canonical v2.1 Results

The dedicated interactive LLM leaderboard contains the current nine-model ranking, dimension breakdown, modality heatmap, and full scored table.