A comparison of approaches to improve worst-case predictive model performance over patient subpopulations

Pathophysiological Features in Electronic Medical Records Sustain Model Performance under Temporal Dataset Shift

Access to real-world data streams like electronic medical records (EMRs) has accelerated the development of supervised machine learning (ML) models for clinical applications. However, few studies investigate the differential impact of particular features in the EMR on model performance under temporal dataset shift. To explain how features in the EMR impact models over time, this study aggregates features into feature groups by their source (e.g. medication orders, diagnosis codes and lab results) and feature categories based on their reflection of patient pathophysiology or healthcare processes. We adapt Shapley values to explain feature groups' and feature categories' marginal contribution to initial and sustained model performance. We investigate three standard clinical prediction tasks and find that while feature contributions to initial performance differ across tasks, pathophysiological features help mitigate temporal discrimination deterioration. These results provide interpretable insights on how specific feature groups contribute to model performance and robustness to temporal dataset shift.

The path toward equal performance in medical machine learning

To ensure equitable quality of care, differences in machine learning model performance between patient groups must be addressed. Here, we argue that two separate mechanisms can cause performance differences between groups. First, model performance may be worse than theoretically achievable in a given group. This can occur due to a combination of group underrepresentation, modeling choices, and the characteristics of the prediction task at hand. We examine scenarios in which underrepresentation leads to underperformance, scenarios in which it does not, and the differences between them. Second, the optimal achievable performance may also differ between groups due to differences in the intrinsic difficulty of the prediction task. We discuss several possible causes of such differences in task difficulty. In addition, challenges such as label biases and selection biases may confound both learning and performance evaluation. We highlight consequences for the path toward equal performance, and we emphasize that leveling up model performance may require gathering not only more data from underperforming groups but also better data. Throughout, we ground our discussion in real-world medical phenomena and case studies while also referencing relevant statistical theory.

Characterizing subgroup performance of probabilistic phenotype algorithms within older adults: a case study for dementia, mild cognitive impairment, and Alzheimer's and Parkinson's diseases

Objective

Biases within probabilistic electronic phenotyping algorithms are largely unexplored. In this work, we characterize differences in subgroup performance of phenotyping algorithms for Alzheimer's disease and related dementias (ADRD) in older adults.

Materials and methods

We created an experimental framework to characterize the performance of probabilistic phenotyping algorithms under different racial distributions allowing us to identify which algorithms may have differential performance, by how much, and under what conditions. We relied on rule-based phenotype definitions as reference to evaluate probabilistic phenotype algorithms created using the Automated PHenotype Routine for Observational Definition, Identification, Training and Evaluation framework.

Results

We demonstrate that some algorithms have performance variations anywhere from 3% to 30% for different populations, even when not using race as an input variable. We show that while performance differences in subgroups are not present for all phenotypes, they do affect some phenotypes and groups more disproportionately than others.

Discussion

Our analysis establishes the need for a robust evaluation framework for subgroup differences. The underlying patient populations for the algorithms showing subgroup performance differences have great variance between model features when compared with the phenotypes with little to no differences.

Conclusion

We have created a framework to identify systematic differences in the performance of probabilistic phenotyping algorithms specifically in the context of ADRD as a use case. Differences in subgroup performance of probabilistic phenotyping algorithms are not widespread nor do they occur consistently. This highlights the great need for careful ongoing monitoring to evaluate, measure, and try to mitigate such differences.

Generalizability challenges of mortality risk prediction models: A retrospective analysis on a multi-center database

Modern predictive models require large amounts of data for training and evaluation, absence of which may result in models that are specific to certain locations, populations in them and clinical practices. Yet, best practices for clinical risk prediction models have not yet considered such challenges to generalizability. Here we ask whether population- and group-level performance of mortality prediction models vary significantly when applied to hospitals or geographies different from the ones in which they are developed. Further, what characteristics of the datasets explain the performance variation? In this multi-center cross-sectional study, we analyzed electronic health records from 179 hospitals across the US with 70,126 hospitalizations from 2014 to 2015. Generalization gap, defined as difference between model performance metrics across hospitals, is computed for area under the receiver operating characteristic curve (AUC) and calibration slope. To assess model performance by the race variable, we report differences in false negative rates across groups. Data were also analyzed using a causal discovery algorithm "Fast Causal Inference" that infers paths of causal influence while identifying potential influences associated with unmeasured variables. When transferring models across hospitals, AUC at the test hospital ranged from 0.777 to 0.832 (1st-3rd quartile or IQR; median 0.801); calibration slope from 0.725 to 0.983 (IQR; median 0.853); and disparity in false negative rates from 0.046 to 0.168 (IQR; median 0.092). Distribution of all variable types (demography, vitals, and labs) differed significantly across hospitals and regions. The race variable also mediated differences in the relationship between clinical variables and mortality, by hospital/region. In conclusion, group-level performance should be assessed during generalizability checks to identify potential harms to the groups. Moreover, for developing methods to improve model performance in new environments, a better understanding and documentation of provenance of data and health processes are needed to identify and mitigate sources of variation.