Joint modeling strategy for using electronic medical records data to build machine learning models: an example of intracerebral hemorrhage.

BACKGROUND: Outliers and class imbalance in medical data could affect the accuracy of machine learning models. For physicians who want to apply predictive models, how to use the data at hand to build a model and what model to choose are very thorny problems. Therefore, it is necessary to consider outliers, imbalanced data, model selection, and parameter tuning when modeling. METHODS: This study used a joint modeling strategy consisting of: outlier detection and removal, data balancing, model fitting and prediction, performance evaluation. We collected medical record data for all ICH patients with admissions in 2017-2019 from Sichuan Province. Clinical and radiological variables were used to construct models to predict mortality outcomes 90 days after discharge. We used stacking ensemble learning to combine logistic regression (LR), random forest (RF), artificial neural network (ANN), support vector machine (SVM), and k-nearest neighbors (KNN) models. Accuracy, sensitivity, specificity, AUC, precision, and F1 score were used to evaluate model performance. Finally, we compared all 84 combinations of the joint modeling strategy, including training set with and without cross-validated committees filter (CVCF), five resampling techniques (random under-sampling (RUS), random over-sampling (ROS), adaptive synthetic sampling (ADASYN), Borderline synthetic minority oversampling technique (Borderline SMOTE), synthetic minority oversampling technique and edited nearest neighbor (SMOTEENN)) and no resampling, seven models (LR, RF, ANN, SVM, KNN, Stacking, AdaBoost). RESULTS: Among 4207 patients with ICH, 2909 (69.15%) survived 90 days after discharge, and 1298 (30.85%) died within 90 days after discharge. The performance of all models improved with removing outliers by CVCF except sensitivity. For data balancing processing, the performance of training set without resampling was better than that of training set with resampling in terms of accuracy, specificity, and precision. And the AUC of ROS was the best. For seven models, the average accuracy, specificity, AUC, and precision of RF were the highest. Stacking performed best in F1 score. Among all 84 combinations of joint modeling strategy, eight combinations performed best in terms of accuracy (0.816). For sensitivity, the best performance was SMOTEENN + Stacking (0.662). For specificity, the best performance was CVCF + KNN (0.987). Stacking and AdaBoost had the best performances in AUC (0.756) and F1 score (0.602), respectively. For precision, the best performance was CVCF + SVM (0.938). CONCLUSION: This study proposed a joint modeling strategy including outlier detection and removal, data balancing, model fitting and prediction, performance evaluation, in order to provide a reference for physicians and researchers who want to build their own models. This study illustrated the importance of outlier detection and removal for machine learning and showed that ensemble learning might be a good modeling strategy. Due to the low imbalanced ratio (IR, the ratio of majority class and minority class) in this study, we did not find any improvement in models with resampling in terms of accuracy, specificity, and precision, while ROS performed best on AUC.

Application of machine learning missing data imputation techniques in clinical decision making: taking the discharge assessment of patients with spontaneous supratentorial intracerebral hemorrhage as an example.

BACKGROUND: There are often many missing values in medical data, which directly affect the accuracy of clinical decision making. Discharge assessment is an important part of clinical decision making. Taking the discharge assessment of patients with spontaneous supratentorial intracerebral hemorrhage as an example, this study adopted the missing data processing evaluation criteria more suitable for clinical decision making, aiming at systematically exploring the performance and applicability of single machine learning algorithms and ensemble learning (EL) under different data missing scenarios, as well as whether they had more advantages than traditional methods, so as to provide basis and reference for the selection of suitable missing data processing method in practical clinical decision making. METHODS: The whole process consisted of four main steps: (1) Based on the original complete data set, missing data was generated by simulation under different missing scenarios (missing mechanisms, missing proportions and ratios of missing proportions of each group). (2) Machine learning and traditional methods (eight methods in total) were applied to impute missing values. (3) The performances of imputation techniques were evaluated and compared by estimating the sensitivity, AUC and Kappa values of prediction models. (4) Statistical tests were used to evaluate whether the observed performance differences were statistically significant. RESULTS: The performances of missing data processing methods were different to a certain extent in different missing scenarios. On the whole, machine learning had better imputation performance than traditional methods, especially in scenarios with high missing proportions. Compared with single machine learning algorithms, the performance of EL was more prominent, followed by neural networks. Meanwhile, EL was most suitable for missing imputation under MAR (the ratio of missing proportion 2:1) mechanism, and its average sensitivity, AUC and Kappa values reached 0.908, 0.924 and 0.596 respectively. CONCLUSIONS: In clinical decision making, the characteristics of missing data should be actively explored before formulating missing data processing strategies. The outstanding imputation performance of machine learning methods, especially EL, shed light on the development of missing data processing technology, and provided methodological support for clinical decision making in presence of incomplete data.

SARS-CoV-2 Seroprevalence and Profiles Among Convalescents in Sichuan Province, China.

Objectives: To explore and understand the SARS-CoV-2 seroprevalence of convalescents, the association between antibody levels and demographic factors, and the seroepidemiology of convalescents of COVID-19 till March 2021. Methods: We recruited 517 voluntary COVID-19 convalescents in Sichuan Province and collected 1,707 serum samples till March 2021. Then we reported the seroprevalence and analyzed the associated factors. Results: Recent travel history was associated with IgM levels. Convalescents who had recent travel history were less likely to be IgM antibody negative [OR = 0.232, 95% CI: (0.128, 0.420)]. Asymptomatic cases had, approximately, twice the odds of being IgM antibody negative compared with symptomatic cases [OR = 2.583, 95% CI: (1.554, 4.293)]. Participants without symptoms were less likely to be IgG seronegative than those with symptoms [OR = 0.511, 95% CI: (0.293, 0.891)]. Convalescents aged 40-59 were less likely to be IgG seronegative than those aged below 20 [OR = 0.364, 95% CI: (0.138, 0.959)]. The duration of positive IgM antibodies persisted 365 days while the IgG persisted more than 399 days. Conclusions: Our findings suggested that recent travel history might be associated with the antibody levels of IgM, while age could be associated with the antibody levels of IgG. Infection type could be associated with both antibody levels of IgM and IgG that declined quicker in asymptomatic cases.

A Breakdown Enhanced AlGaN/GaN Schottky Barrier Diode with the T-Anode Position Deep into the Bottom Buffer Layer.

In this paper, an AlGaN/GaN Schottky barrier diode (SBD) with the T-anode located deep into the bottom buffer layer in combination with field plates (TAI-BBF FPs SBD) is proposed. The electrical characteristics of the proposed structure and the conventional AlGaN/GaN SBD with gated edge termination (GET SBD) were simulated and compared using a Technology Computer Aided Design (TCAD) tool. The results proved that the breakdown voltage (VBK) in the proposed structure was tremendously improved when compared to the GET SBD. This enhancement is attributed to the suppression of the anode tunneling current by the T-anode and the redistribution of the electric field in the anode⁻cathode region induced by the field plates (FPs). Moreover, the T-anode had a negligible effect on the two-dimensional electron gas (2DEG) in the channel layer, so there is no deterioration in the forward characteristics. After being optimized, the proposed structure exhibited a low turn-on voltage (VT) of 0.53 V and a specific on-resistance (RON,sp) of 0.32 mΩ·cm², which was similar to the GET SBD. Meanwhile, the TAI-BBF FP SBD with an anode-cathode spacing of 5 µm achieved a VBK of 1252 V, which was enhanced almost six times compared to the GET SBD with a VBK of 213 V.

Step-Double-Zone-JTE for SiC Devices with Increased Tolerance to JTE Dose and Surface Charges.

In this paper, an edge termination structure, referred to as step-double-zone junction termination extension (Step-DZ-JTE), is proposed. Step-DZ-JTE further improves the distribution of the electric field (EF) by its own step shape. Step-DZ-JTE and other termination structures are investigated for comparison using numerical simulations. Step-DZ-JTE greatly reduces the sensitivity of breakdown voltage (BV) and surface charges (SC). For a 30-µm thick epi-layer, the optimized Step-DZ-JTE shows 90% of the theoretical BV with a wide tolerance of 12.2 × 1012 cm-2 to the JTE dose and 85% of the theoretical BV with an improved tolerance of 3.7 × 1012 cm-2 to the positive SC are obtained. Furthermore, when combined with the field plate technique, the performance of the Step-DZ-JTE is further improved.