# Feature Robustness in Non-stationary Health Records: Caveats to Deployable Model Performance in Common Clinical Machine Learning Tasks

arXiv: https://arxiv.org/abs/1908.00690 · HTML: https://mlsanity.com/paper/1908.00690/

## Key facts

- Cited by 39 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 2 Aug 2019.

## Built on by (most cited first, 30 of 39)

- [Meaningfully Debugging Model Mistakes using Conceptual Counterfactual Explanations](https://mlsanity.com/paper/2106.12723/) (2021)
- [Diagnosing failures of fairness transfer across distribution shift in real-world medical settings](https://mlsanity.com/paper/2202.01034/) (2022)
- [Contrastive Learning for Unsupervised Domain Adaptation of Time Series](https://mlsanity.com/paper/2206.06243/) (2022)
- [An Empirical Framework for Domain Generalization in Clinical Settings](https://mlsanity.com/paper/2103.11163/) (2021)
- [WOODS: Benchmarks for Out-of-Distribution Generalization in Time Series](https://mlsanity.com/paper/2203.09978/) (2022)
- [Healthsheet: Development of a Transparency Artifact for Health Datasets](https://mlsanity.com/paper/2202.13028/) (2022)
- ["Why did the Model Fail?": Attributing Model Performance Changes to Distribution Shifts](https://mlsanity.com/paper/2210.10769/) (2022)
- [Coarse race data conceals disparities in clinical risk score performance](https://mlsanity.com/paper/2304.09270/) (2023)
- [The Data Addition Dilemma](https://mlsanity.com/paper/2408.04154/) (2024)
- [A Stability Principle for Learning under Non-Stationarity](https://mlsanity.com/paper/2310.18304/) (2023)
- [Why Do Probabilistic Clinical Models Fail To Transport Between Sites?](https://mlsanity.com/paper/2311.04787/) (2023)
- [Quantifying disparities in intimate partner violence: a machine learning method to correct for underreporting](https://mlsanity.com/paper/2110.04133/) (2021)
- [Statistically Valid Post-Deployment Monitoring Should Be Standard for AI-Based Digital Health](https://mlsanity.com/paper/2506.05701/) (2025)
- [Minimax Regret Learning for Data with Heterogeneous Subgroups](https://mlsanity.com/paper/2405.01709/) (2024)
- [DeepJoint: Robust Survival Modelling Under Clinical Presence Shift](https://mlsanity.com/paper/2205.13481/) (2022)
- [BEDS-Bench: Behavior of EHR-models under Distributional Shift--A Benchmark](https://mlsanity.com/paper/2107.08189/) (2021)
- [Evaluating Model Performance in Medical Datasets Over Time](https://mlsanity.com/paper/2305.13426/) (2023)
- [Fast and Interpretable Mortality Risk Scores for Critical Care Patients](https://mlsanity.com/paper/2311.13015/) (2023)
- [Bayesian logistic regression for online recalibration and revision of risk prediction models with performance guarantees](https://mlsanity.com/paper/2110.06866/) (2021)
- [Disability prediction in multiple sclerosis using performance outcome measures and demographic data](https://mlsanity.com/paper/2204.03969/) (2022)
- [Looking for Out-of-Distribution Environments in Multi-center Critical Care Data](https://mlsanity.com/paper/2205.13398/) (2022)
- [Large-Scale Study of Temporal Shift in Health Insurance Claims](https://mlsanity.com/paper/2305.05087/) (2023)
- [Recent Advances, Applications, and Open Challenges in Machine Learning for Health: Reflections from Research Roundtables at ML4H 2023 Symposium](https://mlsanity.com/paper/2403.01628/) (2024)
- [Performance Deterioration of Deep Learning Models after Clinical Deployment: A Case Study with Auto-segmentation for Definitive Prostate Cancer Radiotherapy](https://mlsanity.com/paper/2210.05673/) (2022)
- [MLHOps: Machine Learning for Healthcare Operations](https://mlsanity.com/paper/2305.02474/) (2023)
- [Out-of-Distribution Detection and Data Drift Monitoring using Statistical Process Control](https://mlsanity.com/paper/2402.08088/) (2024)
- [Prediction of Survival Outcomes under Clinical Presence Shift: A Joint Neural Network Architecture](https://mlsanity.com/paper/2508.05472/) (2025)
- [SurvBench: A Standardised Preprocessing Pipeline for Multi-Modal Electronic Health Record Survival Analysis](https://mlsanity.com/paper/2511.11935/) (2025)
- [Deep Kernel Learning for Mortality Prediction in the Face of Temporal Shift](https://mlsanity.com/paper/2212.00557/) (2022)
- [The CRITICAL Records Integrated Standardization Pipeline (CRISP): End-to-End Processing of Large-scale Multi-institutional OMOP CDM Data](https://mlsanity.com/paper/2509.08247/) (2025)

## Abstract (arXiv)

When training clinical prediction models from electronic health records (EHRs), a key concern should be a model's ability to sustain performance over time when deployed, even as care practices, database systems, and population demographics evolve. Due to de-identification requirements, however, current experimental practices for public EHR benchmarks (such as the MIMIC-III critical care dataset) are time agnostic, assigning care records to train or test sets without regard for the actual dates of care. As a result, current benchmarks cannot assess how well models trained on one year generalise to another. In this work, we obtain a Limited Data Use Agreement to access year of care for each record in MIMIC and show that all tested state-of-the-art models decay in prediction quality when trained on historical data and tested on future data, particularly in response to a system-wide record-keeping change in 2008 (0.29 drop in AUROC for mortality prediction, 0.10 drop in AUROC for length-of-stay prediction with a random forest classifier). We further develop a simple yet effective mitigation strategy: by aggregating raw features into expert-defined clinical concepts, we see only a 0.06 drop in AUROC for mortality prediction and a 0.03 drop in AUROC for length-of-stay prediction. We demonstrate that this aggregation strategy outperforms other automatic feature preprocessing techniques aimed at increasing robustness to data drift. We release our aggregated representations and code to encourage more deployable clinical prediction models.