Background: In the era of artificial intelligence, event prediction models are abundant. However, considering the limitation of the electronic medical record–based model, including the temporally skewed prediction and the record itself, these models could be delayed or could yield errors. Objective: In this study, we aim to develop multiple event prediction models in intensive care units to overcome their temporal skewness and evaluate their robustness against delayed and erroneous input. Methods: A total of 21,738 patients were included in the development cohort. Three events—death, sepsis, and acute kidney injury—were predicted. To overcome the temporal skewness, we developed three models for each event, which predicted the events in advance of three prespecified timepoints. Additionally, to evaluate the robustness against input error and delays, we added simulated errors and delayed input and calculated changes in the area under the receiver operating characteristic curve (AUROC) values. Results: Most of the AUROC and area under the precision-recall curve values of each model were higher than those of the conventional scores, as well as other machine learning models previously used. In the error input experiment, except for our proposed model, an increase in the noise added to the model lowered the resulting AUROC value. However, the delayed input did not show the performance decreased in this experiment. Conclusions: For a prediction model that was applicable in the real world, we considered not only performance but also temporal skewness, delayed input, and input error.

Lee SH, et al. Yonsei Med J. 2021 Jan;62(1):50-58. https://doi.org/10.3349/ymj.2021.62.1.50

We investigated the effects of particulate matter (PM) factors on hospitalization rates for asthma and chronic obstructive pulmonary disease (COPD). We obtained data on pollutants—PM₁₀, PM_2.5—in Seoul, South Korea. We also investigated data for asthma and COPD exacerbation that required hospitalization from 2006 to 2016. We used a time-stratified case-crossover design and generalized additive models with log transformation to assess adjusted risk, and conditional logistic regression was performed to analyze these data. Our study showed that PM₁₀ and PM_2.5, on different best lag days, were associated with increased risks of COPD or asthma hospitalization. The odds ratios (ORs) for each per-unit increase in PM₁₀ and PM_2.5 were higher in patients with male asthma (PM₁₀: OR, 1.012; 95% confidence interval [CI], 1.008–1.016 and PM_2.5: OR, 1.015; 95% CI, 1008–1.023), preschool asthma (PM₁₀: OR, 1.015; 95% CI, 1.006–1.015 and PM_2.5: OR, 1.015; 95% CI, 1.009–1.024), male COPD (PM₁₀: OR, 1.012; 95% CI, 1.005–1.019 and PM_2.5: OR, 1.013; 95% CI, 1.000–1.026), and senior COPD (PM₁₀: OR, 1.016; 95% CI, 1.008–1.024 and PM_2.5: OR, 1.022; 95% CI, 1.007–1.036). Increasing PM levels increased hospitalizations for asthma and COPD. Additionally, the consequences may be different according to age and sex, and PM_2.5 may have a more significant effect on airway disease patients than PM₁₀.

Objectives: In the study, we explored whether sleep chronotypes are associated with asthma in adolescents. Methods: We analyzed 24,655 physician-diagnosed adolescent asthmatic patients and 253,775 non-asthmatic adolescent patients from the Korea Youth Risk Behavior Web-based Survey (KYRBWS). Socioeconomic factors, health behaviors factors, psychological factors, and sleep parameters were assessed using the Munich Chronotype Questionnaire (MCTQ). Logistic regression after adjusting for multiple confounders was used to explore the association between sleep chronotype and asthma. Results: The asthmatic adolescent group slept less (≤5 h: 24.3% vs. 23.2%) than the non-asthmatic adolescent group. Mean sleep duration (430.6 ± 95.6 vs. 433.5 ± 93.6 min), midpoint of sleep on school-free days (MSF; 255.9 ± 75.9 vs. 258.3 ± 73.6 min), midpoint of sleep on school days (MSW; 199.1 ± 49.1 vs. 200.1 ± 48.4 min), sleep duration on school days (SDW; 398.2 ± 98.1 vs. 400.2 ± 96.8 min), and sleep duration on school-free days (SDF; 511.8 ± 151.9 vs. 516.7 ± 147.2 min) were significantly lower, sleep satisfaction was significantly poorer (low sleep satisfaction: 41.3% vs. 37.5%), and late chronotype was significantly higher in the asthmatic adolescent (21.1% vs. 20.0%). After adjusting for multiple confounders, late chronotype was significantly associated with an increased frequency of adolescent asthma (OR 1.05; 95% CI 1.01–1.09) compared to intermediate chronotypes. Conclusions: Although our study shows a very modest association (OR of 1.05 in the fully adjusted model), we show that the late sleep chronotype is associated with asthma in adolescents in South Korea.