Using Machine Learning Algorithms to Predict Patient Portal Use Among Emergency Department Patients With Diabetes Mellitus.

Background: Different machine learning (ML) technologies have been applied in healthcare systems with diverse applications. We aimed to determine the model feasibility and accuracy of predicting patient portal use among diabetic patients by using six different ML algorithms. In addition, we also compared model performance accuracy with the use of only essential variables. Methods: This was a single-center retrospective observational study. From March 1, 2019 to February 28, 2020, we included all diabetic patients from the study emergency department (ED). The primary outcome was the status of patient portal use. A total of 18 variables consisting of patient sociodemographic characteristics, ED and clinic information, and patient medical conditions were included to predict patient portal use. Six ML algorithms (logistic regression, random forest (RF), deep forest, decision tree, multilayer perception, and support vector machine) were used for such predictions. During the initial step, ML predictions were performed with all variables. Then, the essential variables were chosen via feature selection. Patient portal use predictions were repeated with only essential variables. The performance accuracies (overall accuracy, sensitivity, specificity, and area under receiver operating characteristic curve (AUC)) of patient portal predictions were compared. Results: A total of 77,977 unique patients were placed in our final analysis. Among them, 23.4% (18,223) patients were diabetic mellitus (DM). Patient portal use was found in 26.9% of DM patients. Overall, the accuracy of predicting patient portal use was above 80% among five out of six ML algorithms. The RF outperformed the others when all variables were used for patient portal predictions (accuracy 0.9876, sensitivity 0.9454, specificity 0.9969, and AUC 0.9712). When only eight essential variables were chosen, RF still outperformed the others (accuracy 0.9876, sensitivity 0.9374, specificity 0.9932, and AUC 0.9769). Conclusion: It is possible to predict patient portal use outcomes when different ML algorithms are used with fair performance accuracy. However, with similar prediction accuracies, the use of feature selection techniques can improve the interpretability of the model by addressing the most relevant features.

Patient Portal Use Among Diabetic Patients With Different Races and Ethnicities.

Background: Patient portal (PP) use varies among different patient populations, specifically among those with diabetes mellitus (DM). In addition, it is still uncertain whether PP use could be linked to improved clinical outcomes. Therefore, the aim of this paper was to determine PP use status for patients, recognize factors promoting PP use, and further identify the association between PP use and clinical outcome among diabetic patients of different races and ethnicities. Methods: This was a single-center cross-section study. Patients were divided into non-Hispanic white (NHW), non-Hispanic black (NHB), and Hispanic/Latino groups. PP use was compared among these three groups. Multivariate logistic regressions were used to determine factors associated with PP use, serum glycemic control, and emergency department (ED) hospitalizations. Results: A total of 77,977 patients were analyzed. The rate of PP use among patients of NHW (24%) was higher than those of NHB (19%) and Hispanic/Latinos (18%, P < 0.0001). The adjusted odds ratio (AOR) of insurance coverage associated with PP use was 2.12 (2.02 - 2.23, P < 0.0001), and having a primary care physician (PCP) associated with PP use was 3.89 (3.71 - 4.07, P < 0.0001). In terms of clinical outcomes, the AOR of PP use associated with serum glycemic control was 0.98 (0.90 - 1.05, P = 0.547) and ED hospitalization was 0.79 (0.73 - 0.86, P < 0.0001). Conclusion: PP use disparity occurred among NHB and Hispanic/Latino patients in the ED. Having insurance coverage and PCPs seem to correlate with PP use. PP use did not seem to associate with serum glycemic control among DM patients present in the ED but could possibly reduce patient hospitalizations.

Coronavirus disease 2019 pandemic associated with anxiety and depression among Non-Hispanic whites with chronic conditions in the US.

OBJECTIVES: During the coronavirus 2019 (COVID-19) pandemic, increased anxiety and depression were reported, with mixed findings among individuals of different races/ethnicities. This study examines whether anxiety and depression increased during the COVID-19 pandemic compared to the pre-COVD-19 period among different racial/ethnic groups in the US. METHODS: The Health Information National Trend Surveys 5 (HINTS 5) Cycle 4 data was analyzed. We used the time when the survey was administered as the pre-COVID-19 period (before March 11, 2020, weighted N = 77,501,549) and during the COVID-19 period (on and after March 11, 2020, weighted N = 37,222,019). The Patient Health Questionnaire (PHQ) was used to measure anxiety/depression and further compared before and during COVID-19. Separate multivariable logistic regression analyses were used to determine the association of the COVID-19 pandemic with anxiety/depression after adjusting for age, sex, insurance, income, and education. RESULT: A higher percentage of Non-Hispanic whites (NHW) with chronic conditions reported anxiety (24.3% vs. 11.5%, p = 0.0021) and depression (20.7% vs. 9.3%, p = 0.0034) during COVID-19 than pre-COVID-19. The adjusted odds ratio (AOR) of anxiety and depression for NHWs with chronic conditions during the COVID-19 pandemic was 2.02 (95% confidence interval of 1.10-3.73, p = 0.025) and 2.33 (1.17-4.65, p = 0.018) compared to NHWs who participated in the survey before the COVID-19. LIMITATIONS: Limited to the NHW US population. PHQ can only be used as the initial screening tool. CONCLUSION: The COVID-19 pandemic was associated with an increased prevalence of anxiety and depression among NHW adults with chronic conditions, but not among people of color.