ＨＥ ＹＵＰＥＮＧ

Background and Objective: Excellent generalizability is the precondition for the widespread practical implementation of machine learning models. In our previous study, we developed the schizophrenia classification model (SZ classifier) to identify potential schizophrenia patients in the Japanese population. The SZ classifier has exhibited impressive performance during internal validation. However, ensuring the robustness and generalizability of the SZ classifier requires external validation across independent sample sets. In this study, we aimed to present an external validation of the SZ classifier using outpatient data. Methods: The SZ classifier was trained by using online survey data, which incorporate demographic, health-related, and social comorbidity features. External validation was conducted using an outpatient sample set which is independent from the sample set during the model development phase. The model performance was assessed based on the sensitivity and misclassification rates for schizophrenia, bipolar disorder, and major depression patients. Results: The SZ classifier demonstrated a sensitivity of 0.75 when applied to schizophrenia patients. The misclassification rates were 59% and 55% for bipolar disorder and major depression patients, respectively. Conclusions: The SZ classifier currently encounters challenges in accurately determining the presence or absence of schizophrenia at the individual level. Prior to widespread practical implementation, enhancements are necessary to bolster the accuracy and diminish the misclassification rates. Despite the current limitations of the model, such as poor specificity for certain psychiatric disorders, there is potential for improvement if including multiple types of psychiatric disorders during model development.

Abstract Objectives This study aimed to develop an approach to enhance the model precision by artificial images. Materials and Methods Given an epidemiological study designed to predict 1 response using f features with M samples, each feature was converted into a pixel with certain value. Permutated these pixels into F orders, resulting in F distinct artificial image sample sets. Based on the experience of image recognition techniques, appropriate training images results in higher precision model. In the preliminary experiment, a binary response was predicted by 76 features, the sample set included 223 patients and 1776 healthy controls. Results We randomly selected 10 000 artificial sample sets to train the model. Models’ performance (area under the receiver operating characteristic curve values) depicted a bell-shaped distribution. Conclusion The model construction strategy developed in the research has potential to capture feature order related information and enhance model predictability.

Background In Japan, challenges were reported in accurately estimating the prevalence of schizophrenia among the general population. Retrieving previous studies, we investigated that patients with schizophrenia were more likely to experience poor subjective well-being and various physical, psychiatric, and social comorbidities. These factors might have great potential for precisely classifying schizophrenia cases in order to estimate the prevalence. Machine learning has shown a positive impact on many fields, including epidemiology, due to its high-precision modeling capability. It has been applied in research on mental disorders. However, few studies have applied machine learning technology to the precise classification of schizophrenia cases by variables of demographic and health-related backgrounds, especially using large-scale web-based surveys. Objective The aim of the study is to construct an artificial neural network (ANN) model that can accurately classify schizophrenia cases from large-scale Japanese web-based survey data and to verify the generalizability of the model. Methods Data were obtained from a large Japanese internet research pooled panel (Rakuten Insight, Inc) in 2021. A total of 223 individuals, aged 20-75 years, having schizophrenia, and 1776 healthy controls were included. Answers to the questions in a web-based survey were formatted as 1 response variable (self-report diagnosed with schizophrenia) and multiple feature variables (demographic, health-related backgrounds, physical comorbidities, psychiatric comorbidities, and social comorbidities). An ANN was applied to construct a model for classifying schizophrenia cases. Logistic regression (LR) was used as a reference. The performances of the models and algorithms were then compared. Results The model trained by the ANN performed better than LR in terms of area under the receiver operating characteristic curve (0.86 vs 0.78), accuracy (0.93 vs 0.91), and specificity (0.96 vs 0.94), while the model trained by LR showed better sensitivity (0.63 vs 0.56). Comparing the performances of the ANN and LR, the ANN was better in terms of area under the receiver operating characteristic curve (bootstrapping: 0.847 vs 0.773 and cross-validation: 0.81 vs 0.72), while LR performed better in terms of accuracy (0.894 vs 0.856). Sleep medication use, age, household income, and employment type were the top 4 variables in terms of importance. Conclusions This study constructed an ANN model to classify schizophrenia cases using web-based survey data. Our model showed a high internal validity. The findings are expected to provide evidence for estimating the prevalence of schizophrenia in the Japanese population and informing future epidemiological studies.

The physical, psychiatric, and social comorbidities interfere with the everyday activities of community-dwelling individuals with schizophrenia and increase the risk of their readmission. However, these comorbidities have not been investigated comprehensively in Japan. We conducted a self-reported internet survey in February 2022 to identify individuals aged 20–75 years with and without schizophrenia using a prevalence case-control study. The survey compared physical comorbidities such as being overweight, hypertension, and diabetes; psychiatric comorbidities such as depressive symptoms and sleep disturbances; social comorbidities such as employment status, household income, and social support between participants with and without schizophrenia. A total of 223 participants with schizophrenia and 1776 participants without schizophrenia were identified. Participants with schizophrenia were more likely to be overweight and had a higher prevalence of hypertension, diabetes, and dyslipidemia than participants without schizophrenia. Additionally, depressive symptoms, unemployment, and non-regular employment were more prevalent in participants with schizophrenia than those without schizophrenia. These results highlight the necessity of comprehensive support and interventions addressing physical, psychiatric, and social comorbidities in individuals with schizophrenia in the community. In conclusion, effective interventions for managing comorbidities in individuals with schizophrenia are necessary to enable them to continue to live in the community.

Objectives: To examin whether public trust was associated with the utilization of COVID-19 Contact Confirming Application (COCOA) in those who self-reported a history of COVID-19. Study design: Cross-sectional study. Methods: Data were obtained from the Japan Society and New Tobacco Internet Survey, a nationwide online survey conducted from February to March 2021, which also assessed items related to COVID-19 and public trust. We included 453 participants with a history of COVID-19. Participants' reports of their general trust in the national government and the related policies, attitudes toward COVID-19 vaccination, and the adherence to the preventive measures against SARS-CoV-2 spread were compared between COCOA users and non-users controlling for age, sex, and socioeconomic statuses by analysis of covariance. Mediation analysis was conducted to examine whether public trust mediates the associations of certain participants' characteristics with COCOA utilization. Results: Seventy-six percent (344/453) reported the COCOA utilization. Compared to non-users, the users were younger, more likely to be men and had a tendency to have higher education. They were more willing to get COVID-19 vaccination, adherent to public health measures against the spread of the SARS-Cov-2, and more likely to express trust in government in general and policies related to COVID-19 independent of age, sex, and the socioeconomic status. Trust in government did not mediate the associations of age and education with COCOA utilization. Conclusions: The utilization of digital contact tracing technology for the health of public during pandemic was related to the degree of trust in the government in Japan.

Studies have reported that short-term blood pressure (BP) variability (BPV) is associated with type 2 diabetes mellitus (T2DM) incidence, but the association with long-term BPV remains unclear. The present study investigated the associations of long-term BPV as well as the time trend of BP changes over time with the incidence of T2DM. This study followed a cohort of 3017 Japanese individuals (2446 male, 571 female) aged 36-65 years from 2007 through March 31, 2019. The root-mean-square error (RMSE) and the slope of systolic BP (SBP) change regressed on year were calculated individually using SBP values obtained from 2003 to baseline (2007). A multivariable Cox proportional hazard model was applied to estimate hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) for tertiles of SBP RMSE and continuous SBP slopes adjusted for age, sex, smoking status, regular exercise, sodium intake, family history of diabetes, sleep disorder, body mass index (BMI), SBP, and fasting blood glucose (FBG) at baseline, and BMI slope from 2003 to 2007. The highest RMSE tertile compared to the lowest was associated with a significantly higher incidence of T2DM after adjusting for covariates (HR: 1.79, 95% CI: 1.15, 2.78). The slope was also significantly associated with T2DM incidence until baseline SBP and FBG were adjusted (HR: 1.03, 95% CI: 0.99, 1.07). In conclusion, long-term SBP variability was significantly associated with an increased incidence of T2DM independent of baseline age, sex, BMI, SBP, FBG, lifestyle factors and BMI slope from 2003 until baseline.

The aim of the present study was to derive dietary patterns to explain variation in a set of nutrient intakes or in the measurements of waist circumference (WC) and fasting blood glucose (FBG) using reduced rank regression (RRR) and to prospectively investigate these patterns in relation to the risk of developing metabolic syndrome (MetS) and its components during the follow-up. The study participants were comprised of 2944 government employees aged 30–59 years without MetS. RRR was applied with 38 food groups as predictors and with two sets of response variables. The first set included intake of putatively beneficial nutrients, and the first factor retained was named the Healthy Dietary Pattern (HDP). The second one included baseline WC and FBG, and the first factor was named the Unhealthy Dietary Pattern (UHDP). Multivariable Cox proportional hazard model was used to estimate hazard ratio and 95% confidence intervals with adjustments for age, sex, total energy consumption and other potential confounders. During the 5-year median follow-up, we ascertained 374 cases of MetS. The HDP score was inversely associated with the incidence of MetS (p-trend = 0.009) and hypertension (p-trend = 0.002) and marginally significantly associated with elevated triglyceride and decreased high-density lipoprotein cholesterol (p-trend = 0.08). The UHDP score was linearly positively associated with the incidence of MetS and all its components (all p-trend < 0.05). Both the HDP and UHDP predicted the development of MetS and its components.