A hybrid adaptive approach for instance transfer learning with dynamic and imbalanced data.

Machine learning has demonstrated success in clinical risk prediction modeling with complex electronic health record data. However, the evolving nature of clinical practices can dynamically change the underlying data distribution over time, leading to model performance drift. Adopting an outdated model is potentially risky and may result in unintentional losses. In this paper, we propose a novel Hybrid Adaptive Boosting approach (HA-Boost) for transfer learning. HA-Boost is characterized by the domain similarity-based and class imbalance-based adaptation mechanisms, which simultaneously address two critical limitations of the classical TrAdaBoost algorithm. We validated HA-Boost in predicting hospital-acquired acute kidney injury using real-world longitudinal electronic health records data. The experiment results demonstrate that HA-Boost stably outperforms the competing baselines in terms of both AUROC and AUPRC across a 7-year time span. This study has confirmed the effectiveness of transfer learning as a superior model updating approach in dynamic environment.

Characterizing the temporal changes in association between modifiable risk factors and acute kidney injury with multi-view analysis.

BACKGROUND: Acute kidney injury (AKI) is a common life-threatening clinical syndrome in hospitalized patients. Advances in machine learning has demonstrated success in AKI risk prediction using electronic health records (EHRs). However, to prevent AKI, it is critical to identify clinically modifiable factors and understand their impact at different prevention windows. METHOD: We extracted 4129 clinical variables including demographics, social history, past diagnoses, procedures, labs, medications, vitals from EHRs for a cohort of 144,084 eligible inpatient encounters. We developed a multi-view learning framework for XGBoost (MV-XGB) to enhance algorithm attention on modifiable factors. To study effects of modifiable factors at different time points, we built AKI prediction models at 24-hours, 48-hours, 72-hours before AKI onset. To characterize the temporal changes in effect of modifiable factors on AKI, we derived two indicators, inter-class score-difference and exposed-score-difference, based on SHAP values to compare effects of modifiable factors in different windows. RESULT: MV-XGB effectively increased attention on modifiable factors (explained 92.4%-94.1% inter-class score-difference, i.e., predictive difference between AKI and non-AKI samples) while maintaining good predictive performance (AUROCs were 0.854, 0.798, 0.765 in models for 24-48-72 h AKI prediction respectively). We observed that 62% of predicted odds-ratio difference between AKI and non-AKI patients in 24 h can be explained by factors occurring between 24 and 72 h. Among the important modifiable factors, electrolyte balance explained 38.3% of the inter-class score difference increase between 24 h and 72 h, followed by high-risk medications (13.7%), care strategy (12.1%), blood pressure (10%), infection (7.8%), and anemia (5.4%). Effects of cardiac surgery or condition, respiratory ventilation, and anemia remained important longer than 72 h. CONCLUSION: Better understanding of the clinically modifiable factors is important to AKI prevention. The proposed multi-view learning approach improved the identification of modifiable factors of AKI and allowed characterization of the temporal dynamics of their potential benefit in intervention.

Which risk predictors are more likely to indicate severe AKI in hospitalized patients?

OBJECTIVES: Acute kidney injury (AKI) is a sudden episode of kidney failure or damage and the risk of AKI is determined by the complex interactions of patient factors. In this study, we aimed to find out which risk factors in hospitalized patients are more likely to indicate severe AKI. METHODS: We constructed a retrospective cohort of adult patients from all inpatient units of a tertiary care academic hospital between November 2007 and December 2016. AKI predictors included demographic information, admission and discharge dates, medications, laboratory values, past medical diagnoses and admission diagnosis. We developed a machine learning-based knowledge mining model and a screening framework to analyze which risk predictors are more likely to imply severe AKI in hospitalized populations. RESULTS: Among the final analysis cohort of 76,957 hospital admissions, AKI occurred in 7,259 (9.43 %) with 6,396 (8.31 %) at stage 1, 678 (0.88 %) at stage 2, and 185 (0.24 %) at stage 3. We compared the non-AKI (without AKI) vs any AKI (stages 1-3), and mild AKI (stage 1) vs severe AKI (stages 2 and 3), where the best cross-validated area under the receiver operator characteristic curve (AUC) were 0.81 (95 % CI, 0.79-0.82) and 0.66 (95 % CI, 0.62-0.71), respectively. Using the developed knowledge mining model and screening framework, we identified 33 risk predictors indicating that severe AKI may occur. CONCLUSIONS: This study screened out 33 risk predictors that are more likely to indicate severe AKI in hospitalized patients, which would help strengthen the early care and prevention of patients.