Research map / Probabilistic and scientific modeling / Molecules and proteins
A Machine-Learned Comorbidity Index
Suleman Baloch, Kishlay Jha, Alberto Maria Segre, Philip M. Polgreen, Bijaya Adhikari
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| train score with multi-outcome nHSIC instead of fixed-weight clinical index | MLCI vs Elixhauser · MIMIC-IV MORT | distance correlation x10^2: 19.98 -> 54.80 | Improved |
| train score with multi-outcome nHSIC instead of fixed-weight clinical index | MLCI vs Elixhauser · MIMIC-III 30M | distance correlation x10^2: 24.70 -> 37.39 | Improved |
| nHSIC objective instead of BCE for the same DeepSets architecture | DeepSets encoder · MIMIC-IV MORT | distance correlation x10^2: 28.51 -> 54.80 | Improved |
| nHSIC objective instead of BCE for the same DeepSets architecture | DeepSets encoder · MIMIC-IV ICU | distance correlation x10^2: 51.92 -> 61.97 | Improved |
| nHSIC objective instead of BCE for the same DeepSets architecture | DeepSets encoder · MIMIC-IV LOS | distance correlation x10^2: 48.60 -> 51.15 | No clear effect |
| nHSIC objective instead of BCE for the same DeepSets architecture | DeepSets encoder · MIMIC-III ICU (late transfer) | distance correlation x10^2: 21.52 -> 18.55 | Hurt |
| nHSIC objective instead of BCE | DeepSets encoder · MIMIC-III MORT | mutual information x10^3: 73.69 -> 84.52 | Improved |
| nHSIC objective instead of BCE | DeepSets encoder · MIMIC-IV ICU | mutual information x10^2: 19.24 -> 19.01 | Mixed |
| use learned/ML score instead of CCI | Charlson · MIMIC-IV MORT | distance correlation x10^2: 12.59 -> 54.80 | Improved |
| add Elixhauser to Charlson as a summed scalar | CCI · MIMIC-IV MORT | distance correlation x10^2: 12.59 -> 17.46 | Improved |
| rank-one projection of stacked label structure as shared direction | MLCI theory diagnostic · MIMIC-IV | rank-one energy share: 0.49 | No clear effect |
| compare best full-objective threshold with rank-one threshold | MLCI · MIMIC-IV vs MIMIC-III | threshold split agreement: jJ 4863 = jv 4863 (IV); jJ 6707 vs jv 11128 (III) | Mixed |
Closest papers by meaning
- SAFER: A Calibrated Risk-Aware Multimodal Recommendation Model for Dynamic Treatment Regimes (ICML 2025 · similarity 0.82)
- Risk Horizons: Structured Hypothesis Spaces for Longitudinal Clinical Prediction (NeurIPS 2026 · similarity 0.81)
- Yet Another ICU Benchmark: A Flexible Multi-Center Framework for Clinical ML (ICLR 2024 · similarity 0.81)
- Aligning Evaluation with Clinical Priorities: Calibration, Label Shift, and Error Costs (NeurIPS 2025 · similarity 0.81)
- Contrastive Learning for Clinical Outcome Prediction with Partial Data Sources (ICML 2024 · similarity 0.81)
- Exploiting Negative Samples: A Catalyst for Cohort Discovery in Healthcare Analytics (ICML 2024 · similarity 0.81)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
