Status of MyHealthWay and Suggestions for Widespread Implementation, Emphasizing the Utilization and Practical Use of Personal Medical Data.

OBJECTIVES: In the Fourth Industrial Revolution, there is a focus on managing diverse medical data to improve healthcare and prevent disease. The challenges include tracking detailed medical records across multiple institutions and the necessity of linking domestic public medical entities for efficient data sharing. This study explores MyHealthWay, a Korean healthcare platform designed to facilitate the integration and transfer of medical data from various sources, examining its development, importance, and legal implications. METHODS: To evaluate the management status and utilization of MyHealthWay, we analyzed data types, security, legal issues, domestic versus international issues, and infrastructure. Additionally, we discussed challenges such as resource and infrastructure constraints, regulatory hurdles, and future considerations for data management. RESULTS: The secure sharing of medical information via MyHealthWay can reduce the distance between patients and healthcare facilities, fostering personalized care and self-management of health. However, this approach faces legal challenges, particularly relating to data standardization and access to personal health information. Legal challenges in data standardization and access, particularly for secondary uses such as research, necessitate improved regulations. There is a crucial need for detailed governmental guidelines and clear data ownership standards at institutional levels. CONCLUSIONS: This report highlights the role of Korea's MyHealthWay, which was launched in 2023, in transforming healthcare through systematic data integration. Challenges include data privacy and legal complexities, and there is a need for data standardization and individual empowerment in health data management within a systematic medical big data framework.

Early Prediction of Mortality for Septic Patients Visiting Emergency Room Based on Explainable Machine Learning: A Real-World Multicenter Study.

BACKGROUND: Worldwide, sepsis is the leading cause of death in hospitals. If mortality rates in patients with sepsis can be predicted early, medical resources can be allocated efficiently. We constructed machine learning (ML) models to predict the mortality of patients with sepsis in a hospital emergency department. METHODS: This study prospectively collected nationwide data from an ongoing multicenter cohort of patients with sepsis identified in the emergency department. Patients were enrolled from 19 hospitals between September 2019 and December 2020. For acquired data from 3,657 survivors and 1,455 deaths, six ML models (logistic regression, support vector machine, random forest, extreme gradient boosting [XGBoost], light gradient boosting machine, and categorical boosting [CatBoost]) were constructed using fivefold cross-validation to predict mortality. Through these models, 44 clinical variables measured on the day of admission were compared with six sequential organ failure assessment (SOFA) components (PaO2/FIO2 [PF], platelets (PLT), bilirubin, cardiovascular, Glasgow Coma Scale score, and creatinine). The confidence interval (CI) was obtained by performing 10,000 repeated measurements via random sampling of the test dataset. All results were explained and interpreted using Shapley's additive explanations (SHAP). RESULTS: Of the 5,112 participants, CatBoost exhibited the highest area under the curve (AUC) of 0.800 (95% CI, 0.756-0.840) using clinical variables. Using the SOFA components for the same patient, XGBoost exhibited the highest AUC of 0.678 (95% CI, 0.626-0.730). As interpreted by SHAP, albumin, lactate, blood urea nitrogen, and international normalization ratio were determined to significantly affect the results. Additionally, PF and PLTs in the SOFA component significantly influenced the prediction results. CONCLUSION: Newly established ML-based models achieved good prediction of mortality in patients with sepsis. Using several clinical variables acquired at the baseline can provide more accurate results for early predictions than using SOFA components. Additionally, the impact of each variable was identified.

Prediction of post-stroke cognitive impairment after acute ischemic stroke using machine learning.

BACKGROUND AND OBJECTIVES: Post-stroke cognitive impairment (PSCI) occurs in up to 50% of patients with acute ischemic stroke (AIS). Thus, the prediction of cognitive outcomes in AIS may be useful for treatment decisions. This PSCI cohort study aimed to determine the applicability of a machine learning approach for predicting PSCI after stroke. METHODS: This retrospective study used a prospective PSCI cohort of patients with AIS. Demographic features, clinical characteristics, and brain imaging variables previously known to be associated with PSCI were included in the analysis. The primary outcome was PSCI at 3-6 months, defined as an adjusted z-score of less than - 2.0 standard deviation in at least one of the four cognitive domains (memory, executive/frontal, visuospatial, and language), using the Korean version of the Vascular Cognitive Impairment Harmonization Standards-Neuropsychological Protocol (VCIHS-NP). We developed four machine learning models (logistic regression, support vector machine, extreme gradient boost, and artificial neural network) and compared their accuracies for outcome variables. RESULTS: A total of 951 patients (mean age 65.7 ± 11.9; male 61.5%) with AIS were included in this study. The area under the curve for the extreme gradient boost and the artificial neural network was the highest (0.7919 and 0.7365, respectively) among the four models for predicting PSCI according to the VCIHS-NP definition. The most important features for predicting PSCI include the presence of cortical infarcts, mesial temporal lobe atrophy, initial stroke severity, stroke history, and strategic lesion infarcts. CONCLUSION: Our findings indicate that machine-learning algorithms, particularly the extreme gradient boost and the artificial neural network models, can best predict cognitive outcomes after ischemic stroke.

Deep learning application for the classification of Alzheimer's disease using 18F-flortaucipir (AV-1451) tau positron emission tomography.

The positron emission tomography (PET) with 18F-flortaucipir can distinguish individuals with mild cognitive impairment (MCI) and Alzheimer's disease (AD) from cognitively unimpaired (CU) individuals. This study aimed to evaluate the utility of 18F-flortaucipir-PET images and multimodal data integration in the differentiation of CU from MCI or AD through DL. We used cross-sectional data (18F-flortaucipir-PET images, demographic and neuropsychological score) from the ADNI. All data for subjects (138 CU, 75 MCI, 63 AD) were acquired at baseline. The 2D convolutional neural network (CNN)-long short-term memory (LSTM) and 3D CNN were conducted. Multimodal learning was conducted by adding the clinical data with imaging data. Transfer learning was performed for classification between CU and MCI. The AUC for AD classification from CU was 0.964 and 0.947 in 2D CNN-LSTM and multimodal learning. The AUC of 3D CNN showed 0.947, and 0.976 in multimodal learning. The AUC for MCI classification from CU had 0.840 and 0.923 in 2D CNN-LSTM and multimodal learning. The AUC of 3D CNN showed 0.845, and 0.850 in multimodal learning. The 18F-flortaucipir PET is effective for the classification of AD stage. Furthermore, the effect of combination images with clinical data increased the performance of AD classification.

Machine learning application for classification of Alzheimer's disease stages using 18F-flortaucipir positron emission tomography.

BACKGROUND: The progression of Alzheimer's dementia (AD) can be classified into three stages: cognitive unimpairment (CU), mild cognitive impairment (MCI), and AD. The purpose of this study was to implement a machine learning (ML) framework for AD stage classification using the standard uptake value ratio (SUVR) extracted from 18F-flortaucipir positron emission tomography (PET) images. We demonstrate the utility of tau SUVR for AD stage classification. We used clinical variables (age, sex, education, mini-mental state examination scores) and SUVR extracted from PET images scanned at baseline. Four types of ML frameworks, such as logistic regression, support vector machine (SVM), extreme gradient boosting, and multilayer perceptron (MLP), were used and explained by Shapley Additive Explanations (SHAP) to classify the AD stage. RESULTS: Of a total of 199 participants, 74, 69, and 56 patients were in the CU, MCI, and AD groups, respectively; their mean age was 71.5 years, and 106 (53.3%) were men. In the classification between CU and AD, the effect of clinical and tau SUVR was high in all classification tasks and all models had a mean area under the receiver operating characteristic curve (AUC) > 0.96. In the classification between MCI and AD, the independent effect of tau SUVR in SVM had an AUC of 0.88 (p < 0.05), which was the highest compared to other models. In the classification between MCI and CU, the AUC of each classification model was higher with tau SUVR variables than with clinical variables independently, which yielded an AUC of 0.75(p < 0.05) in MLP, which was the highest. As an explanation by SHAP for the classification between MCI and CU, and AD and CU, the amygdala and entorhinal cortex greatly affected the classification results. In the classification between MCI and AD, the para-hippocampal and temporal cortex affected model performance. Especially entorhinal cortex and amygdala showed a higher effect on model performance than all clinical variables in the classification between MCI and CU. CONCLUSIONS: The independent effect of tau deposition indicates that it is an effective biomarker in classifying CU and MCI into clinical stages using MLP. It is also very effective in classifying AD stages using SVM with clinical information that can be easily obtained at clinical screening.

Indirect Volume Estimation for Acute Ischemic Stroke from Diffusion Weighted Image Using Slice Image Segmentation.

The accurate estimation of acute ischemic stroke (AIS) using diffusion-weighted imaging (DWI) is crucial for assessing patients and guiding treatment options. This study aimed to propose a method that estimates AIS volume in DWI objectively, quickly, and accurately. We used a dataset of DWI with AIS, including 2159 participants (1179 for internal validation and 980 for external validation) with various types of AIS. We constructed algorithms using 3D segmentation (direct estimation) and 2D segmentation (indirect estimation) and compared their performances with those annotated by neurologists. The proposed pretrained indirect model demonstrated higher segmentation performance than the direct model, with a sensitivity, specificity, F1-score, and Jaccard index of 75.0%, 77.9%, 76.0, and 62.1%, respectively, for internal validation, and 72.8%, 84.3%, 77.2, and 63.8%, respectively, for external validation. Volume estimation was more reliable for the indirect model, with 93.3% volume similarity (VS), 0.797 mean absolute error (MAE) for internal validation, VS of 89.2% and a MAE of 2.5% for external validation. These results suggest that the indirect model using 2D segmentation developed in this study can provide an accurate estimation of volume from DWI of AIS and may serve as a supporting tool to help physicians make crucial clinical decisions.