Inter hospital external validation of interpretable machine learning based triage score for the emergency department using common data model.

Emergency departments (ED) are complex, triage is a main task in the ED to prioritize patient with limited medical resources who need them most. Machine learning (ML) based ED triage tool, Score for Emergency Risk Prediction (SERP), was previously developed using an interpretable ML framework with single center. We aimed to develop SERP with 3 Korean multicenter cohorts based on common data model (CDM) without data sharing and compare performance with inter-hospital validation design. This retrospective cohort study included all adult emergency visit patients of 3 hospitals in Korea from 2016 to 2017. We adopted CDM for the standardized multicenter research. The outcome of interest was 2-day mortality after the patients' ED visit. We developed each hospital SERP using interpretable ML framework and validated inter-hospital wisely. We accessed the performance of each hospital's score based on some metrics considering data imbalance strategy. The study population for each hospital included 87,670, 83,363 and 54,423 ED visits from 2016 to 2017. The 2-day mortality rate were 0.51%, 0.56% and 0.65%. Validation results showed accurate for inter hospital validation which has at least AUROC of 0.899 (0.858-0.940). We developed multicenter based Interpretable ML model using CDM for 2-day mortality prediction and executed Inter-hospital external validation which showed enough high accuracy.

Incidence of mental health diagnoses during the COVID-19 pandemic: a multinational network study.

AIMS: Population-wide restrictions during the COVID-19 pandemic may create barriers to mental health diagnosis. This study aims to examine changes in the number of incident cases and the incidence rates of mental health diagnoses during the COVID-19 pandemic. METHODS: By using electronic health records from France, Germany, Italy, South Korea and the UK and claims data from the US, this study conducted interrupted time-series analyses to compare the monthly incident cases and the incidence of depressive disorders, anxiety disorders, alcohol misuse or dependence, substance misuse or dependence, bipolar disorders, personality disorders and psychoses diagnoses before (January 2017 to February 2020) and after (April 2020 to the latest available date of each database [up to November 2021]) the introduction of COVID-related restrictions. RESULTS: A total of 629,712,954 individuals were enrolled across nine databases. Following the introduction of restrictions, an immediate decline was observed in the number of incident cases of all mental health diagnoses in the US (rate ratios (RRs) ranged from 0.005 to 0.677) and in the incidence of all conditions in France, Germany, Italy and the US (RRs ranged from 0.002 to 0.422). In the UK, significant reductions were only observed in common mental illnesses. The number of incident cases and the incidence began to return to or exceed pre-pandemic levels in most countries from mid-2020 through 2021. CONCLUSIONS: Healthcare providers should be prepared to deliver service adaptations to mitigate burdens directly or indirectly caused by delays in the diagnosis and treatment of mental health conditions.

Explainable multimodal prediction of treatment-resistance in patients with depression leveraging brain morphometry and natural language processing.

Although 20 % of patients with depression receiving treatment do not achieve remission, predicting treatment-resistant depression (TRD) remains challenging. In this study, we aimed to develop an explainable multimodal prediction model for TRD using structured electronic medical record data, brain morphometry, and natural language processing. In total, 247 patients with a new depressive episode were included. TRD-predictive models were developed based on the combination of following parameters: selected tabular dataset features, independent components-map weightings from brain T1-weighted magnetic resonance imaging (MRI), and topic probabilities from clinical notes. All models applied the extreme gradient boosting (XGBoost) algorithm via five-fold cross-validation. The model using all data sources showed the highest area under the receiver operating characteristic of 0.794, followed by models that used combined brain MRI and structured data, brain MRI and clinical notes, clinical notes and structured data, brain MRI only, structured data only, and clinical notes only (0.770, 0.762, 0.728, 0.703, 0.684, and 0.569, respectively). Classifications of TRD were driven by several predictors, such as previous exposure to antidepressants and antihypertensive medications, sensorimotor network, default mode network, and somatic symptoms. Our findings suggest that a combination of clinical data with neuroimaging and natural language processing variables improves the prediction of TRD.

Towards global model generalizability: independent cross-site feature evaluation for patient-level risk prediction models using the OHDSI network.

BACKGROUND: Predictive models show promise in healthcare, but their successful deployment is challenging due to limited generalizability. Current external validation often focuses on model performance with restricted feature use from the original training data, lacking insights into their suitability at external sites. Our study introduces an innovative methodology for evaluating features during both the development phase and the validation, focusing on creating and validating predictive models for post-surgery patient outcomes with improved generalizability. METHODS: Electronic health records (EHRs) from 4 countries (United States, United Kingdom, Finland, and Korea) were mapped to the OMOP Common Data Model (CDM), 2008-2019. Machine learning (ML) models were developed to predict post-surgery prolonged opioid use (POU) risks using data collected 6 months before surgery. Both local and cross-site feature selection methods were applied in the development and external validation datasets. Models were developed using Observational Health Data Sciences and Informatics (OHDSI) tools and validated on separate patient cohorts. RESULTS: Model development included 41 929 patients, 14.6% with POU. The external validation included 31 932 (UK), 23 100 (US), 7295 (Korea), and 3934 (Finland) patients with POU of 44.2%, 22.0%, 15.8%, and 21.8%, respectively. The top-performing model, Lasso logistic regression, achieved an area under the receiver operating characteristic curve (AUROC) of 0.75 during local validation and 0.69 (SD = 0.02) (averaged) in external validation. Models trained with cross-site feature selection significantly outperformed those using only features from the development site through external validation (P < .05). CONCLUSIONS: Using EHRs across four countries mapped to the OMOP CDM, we developed generalizable predictive models for POU. Our approach demonstrates the significant impact of cross-site feature selection in improving model performance, underscoring the importance of incorporating diverse feature sets from various clinical settings to enhance the generalizability and utility of predictive healthcare models.

Comparative estimation of the effects of antihypertensive medications on schizophrenia occurrence: a multinational observational cohort study.

BACKGROUND: The association between antihypertensive medication and schizophrenia has received increasing attention; however, evidence of the impact of antihypertensive medication on subsequent schizophrenia based on large-scale observational studies is limited. We aimed to compare the schizophrenia risk in large claims-based US and Korea cohort of patients with hypertension using angiotensin-converting enzyme (ACE) inhibitors versus those using angiotensin receptor blockers (ARBs) or thiazide diuretics. METHODS: Adults aged 18 years who were newly diagnosed with hypertension and received ACE inhibitors, ARBs, or thiazide diuretics as first-line antihypertensive medications were included. The study population was sub-grouped based on age (> 45 years). The comparison groups were matched using a large-scale propensity score (PS)-matching algorithm. The primary endpoint was incidence of schizophrenia. RESULTS: 5,907,522; 2,923,423; and 1,971,549 patients used ACE inhibitors, ARBs, and thiazide diuretics, respectively. After PS matching, the risk of schizophrenia was not significantly different among the groups (ACE inhibitor vs. ARB: summary hazard ratio [HR] 1.15 [95% confidence interval, CI, 0.99-1.33]; ACE inhibitor vs. thiazide diuretics: summary HR 0.91 [95% CI, 0.78-1.07]). In the older subgroup, there was no significant difference between ACE inhibitors and thiazide diuretics (summary HR, 0.91 [95% CI, 0.71-1.16]). The risk for schizophrenia was significantly higher in the ACE inhibitor group than in the ARB group (summary HR, 1.23 [95% CI, 1.05-1.43]). CONCLUSIONS: The risk of schizophrenia was not significantly different between the ACE inhibitor vs. ARB and ACE inhibitor vs. thiazide diuretic groups. Further investigations are needed to determine the risk of schizophrenia associated with antihypertensive drugs, especially in people aged > 45 years.

Real-world treatment trajectories of adults with newly diagnosed asthma or COPD.

BACKGROUND: There is a lack of knowledge on how patients with asthma or chronic obstructive pulmonary disease (COPD) are globally treated in the real world, especially with regard to the initial pharmacological treatment of newly diagnosed patients and the different treatment trajectories. This knowledge is important to monitor and improve clinical practice. METHODS: This retrospective cohort study aims to characterise treatments using data from four claims (drug dispensing) and four electronic health record (EHR; drug prescriptions) databases across six countries and three continents, encompassing 1.3 million patients with asthma or COPD. We analysed treatment trajectories at drug class level from first diagnosis and visualised these in sunburst plots. RESULTS: In four countries (USA, UK, Spain and the Netherlands), most adults with asthma initiate treatment with short-acting ß2 agonists monotherapy (20.8%-47.4% of first-line treatments). For COPD, the most frequent first-line treatment varies by country. The largest percentages of untreated patients (for asthma and COPD) were found in claims databases (14.5%-33.2% for asthma and 27.0%-52.2% for COPD) from the USA as compared with EHR databases (6.9%-15.2% for asthma and 4.4%-17.5% for COPD) from European countries. The treatment trajectories showed step-up as well as step-down in treatments. CONCLUSION: Real-world data from claims and EHRs indicate that first-line treatments of asthma and COPD vary widely across countries. We found evidence of a stepwise approach in the pharmacological treatment of asthma and COPD, suggesting that treatments may be tailored to patients' needs.

Prediction Models for Readmission Using Home Healthcare Notes and OMOP-CDM.

This study developed readmission prediction models using Home Healthcare (HHC) documents via natural language processing (NLP). An electronic health record of Ajou University Hospital was used to develop prediction models (A reference model using only structured data, and an NLP-enriched model with structured and unstructured data). Among 573 patients, 63 were readmitted to the hospital. Five topics were extracted from HHC documents and improved the model performance (AUROC 0.740).

Multi-Institutional Collaborative Research Using Ophthalmic Medical Image Data Standardized by Radiology Common Data Model (R-CDM).

Observational Medical Outcome Partners - Common Data Model (OMOP-CDM) is an international standard model for standardizing electronic medical record data. However, unstructured data such as medical image data which is beyond the scope of standardization by the current OMOP-CDM is difficult to be used in multi-institutional collaborative research. Therefore, we developed the Radiology-CDM (R-CDM) which standardizes medical imaging data. As a proof of concept, 737,500 Optical Coherence Tomography (OCT) data from two tertiary hospitals in South Korea is standardized in the form of R-CDM. The relationship between chronic disease and retinal thickness was analyzed by using the R-CDM. Central macular thickness and retinal nerve fiber layer (RNFL) thickness were significantly thinner in the patients with hypertension compared to the control cohort. It is meaningful in that multi-institutional collaborative research using medical image data and clinical data simultaneously can be conducted very efficiently.

A De-Identification Model for Korean Clinical Notes: Using Deep Learning Models.

To extract information from free-text in clinical records due to the patient's protected health information PHI in the records pre-processing of de-identification is required. Therefore we aimed to identify PHI list and fine-tune the deep learning BERT model for developing de-identification model. The result of fine-tuning the model is strict F1 score of 0.924. Due to the convinced score the model can be used for the development of a de-identification model.

RHEA: Real-World Observational Health Data Exploration Application.

We developed a standardized framework named RHEA to represent longitudinal status of patient with cancer. RHEA generates a dashboard to visualize patients' data in the Observational Medical Outcomes Partnership-Common Data Model format. The generated dashboard consists of three main parts for providing the macroscopic characteristics of the patient: 1) cohort-level visualization, 2) individual-level visualization and 3) cohort generation.

Perioperative adverse cardiac events and mortality after non-cardiac surgery: a multicenter study.

BACKGROUND: Perioperative adverse cardiac events (PACE), a composite of myocardial infarction, coronary revascularization, congestive heart failure, arrhythmic attack, acute pulmonary embolism, cardiac arrest, and stroke during 30-day postoperative period, is associated with long-term mortality, but with limited clinical evidence. We compared long-term mortality with PACE using data from nationwide multicenter electronic health records. METHODS: Data from 7 hospitals, converted to Observational Medical Outcomes Partnership Common Data Model, were used. We extracted records of 277,787 adult patients over 18 years old undergoing non-cardiac surgery for the first time at the hospital and had medical records for more than 180 days before surgery. We performed propensity score matching and then an aggregated meta­analysis. RESULTS: After 1:4 propensity score matching, 7,970 patients with PACE and 28,807 patients without PACE were matched. The meta­analysis showed that PACE was associated with higher one-year mortality risk (hazard ratio [HR]: 1.33, 95% CI [1.10, 1.60], P = 0.005) and higher three-year mortality (HR: 1.18, 95% CI [1.01, 1.38], P = 0.038). In subgroup analysis, the risk of one-year mortality by PACE became greater with higher-risk surgical procedures (HR: 1.20, 95% CI [1.04, 1.39], P = 0.020 for low-risk surgery; HR: 1.69, 95% CI [1.45, 1.96], P < 0.001 for intermediate-risk; and HR: 2.38, 95% CI [1.47, 3.86], P = 0.034 for high-risk). CONCLUSIONS: A nationwide multicenter study showed that PACE was significantly associated with increased one-year mortality. This association was stronger in high-risk surgery, older, male, and chronic kidney disease subgroups. Further studies to improve mortality associated with PACE are needed.

Real-World Effectiveness of Statin Therapy in Adult Asthma.

BACKGROUND: Blood lipids affect airway inflammation in asthma. Although several studies have suggested anti-inflammatory effects of statins on asthmatic airways, further studies are needed to clarify the long-term effectiveness of statins on asthma control and whether they are an effective treatment option. OBJECTIVE: To evaluate the long-term effectiveness of statins in the chronic management of adult asthma in real-world practice. METHODS: Electronic medical record data spanning 28 years, collected from the Ajou University Medical Center in Korea, were used to conduct a retrospective study. Clinical outcomes were compared between patients with asthma who had maintained statin use (the statin group) and those not taking statins, whose blood lipid tests were always normal (the non-statin group). We performed propensity score matching and calculated hazard ratios with 95% CIs using the Cox proportional hazards model. Severe asthma exacerbation was the primary outcome; asthma exacerbation, asthma-related hospitalization, and new-onset type 2 diabetes mellitus and hypertension were secondary outcomes. RESULTS: After 1:1 propensity score matching, the statin and non-statin groups each included 545 adult patients with asthma. The risk of severe asthma exacerbations and asthma exacerbations was significantly lower in the statin group than in the non-statin group (hazard ratios [95% CI] = 0.57 [0.35-0.90] and 0.71 [0.52-0.96], respectively). There were no significant differences in the risk of asthma-related hospitalization or new-onset type 2 diabetes mellitus or hypertension between groups (0.76 [0.53-1.09], 2.33 [0.94-6.59], and 1.71 [0.95-3.17], respectively). CONCLUSION: Statin use is associated with a lower risk of asthma exacerbation, with better clinical outcomes in adult asthma.

Assessing the potential of ChatGPT for psychodynamic formulations in psychiatry: An exploratory study.

Although there were several attempts to apply ChatGPT (Generative Pre-Trained Transformer) to medicine, little is known about therapeutic applications in psychiatry. In this exploratory study, we aimed to evaluate the characteristics and appropriateness of the psychodynamic formulations created by ChatGPT. Along with a case selected from the psychoanalytic literature, input prompts were designed to include different levels of background knowledge. These included naïve prompts, keywords created by ChatGPT, keywords created by psychiatrists, and psychodynamic concepts from the literature. The psychodynamic formulations generated from the different prompts were evaluated by five psychiatrists from different institutions. We next conducted further tests in which instructions on the use of different psychodynamic models were added to the input prompts. The models used were ego psychology, self-psychology, and object relations. The results from naïve prompts and psychodynamic concepts were rated as appropriate by most raters. The psychodynamic concept prompt output was rated the highest. Interrater agreement was statistically significant. The results from the tests using instructions in different psychoanalytic theories were also rated as appropriate by most raters. They included key elements of the psychodynamic formulation and suggested interpretations similar to the literature. These findings suggest potential of ChatGPT for use in psychiatry.

Safety outcomes of antipsychotics classes in drug-naïve patients with first-episode schizophrenia: A nationwide cohort study in South Korea.

INTRODUCTION: Given the similar efficacies across antipsychotic medications for schizophrenia, understanding their safety profiles, particularly concerning receptor-binding differences, is crucial for optimal drug selection, especially for patients with first episode schizophrenia. We aimed to compare the safety outcomes of second-generation antipsychotics. METHODS: We conducted a retrospective cohort study with new user active comparator design using a nationwide claims database in South Korea. Participants were drug-naïve adult patients with first-episode schizophrenia. Three representative drugs with different pharmacologic profiles were compared: risperidone, olanzapine, and aripiprazole. Propensity scores were used to match the study groups, and the Cox proportional hazard model was used to calculate hazard ratios. Sensitivity analyses were performed in various epidemiological settings. Seventeen safety outcomes, including neuropsychiatric, cardiometabolic and gastrointestinal events, were assessed, with upper-respiratory-tract infection as a negative control outcome. RESULTS: A total of 1044, 2078, and 3634 participants were matched for olanzapine vs. risperidone, olanzapine vs. aripiprazole, and risperidone vs. aripiprazole comparisons, respectively. For parkinsonism, there was a significant difference in outcomes between the risperidone and aripiprazole groups (HR 1.80 [95% CI 1.13-2.91]), with consistent sensitivity analysis results. There were no significant differences in other neuropsychiatry outcomes or in the risk of cardiometabolic and gastrointestinal outcomes between any of the comparative group pairs. CONCLUSIONS: The risk of drug-induced parkinsonism was significantly higher with risperidone than with aripiprazole. Although olanzapine is known for its metabolic risk, there were no significant differences in risk between the other pairs.

Effect of statin use on head and neck cancer prognosis in a multicenter study using a Common Data Model.

Few studies have found an association between statin use and head and neck cancer (HNC) outcomes. We examined the effect of statin use on HNC recurrence using the converted Observational Medical Outcome Partnership (OMOP) Common Data Model (CDM) in seven hospitals between 1986 and 2022. Among the 9,473,551 eligible patients, we identified 4669 patients with HNC, of whom 398 were included in the target cohort, and 4271 were included in the control cohort after propensity score matching. A Cox proportional regression model was used. Of the 4669 patients included, 398 (8.52%) previously received statin prescriptions. Statin use was associated with a reduced rate of 3- and 5-year HNC recurrence compared to propensity score-matched controls (risk ratio [RR], 0.79; 95% confidence interval [CI], 0.61-1.03; and RR 0.89; 95% CI 0.70-1.12, respectively). Nevertheless, the association between statin use and HNC recurrence was not statistically significant. A meta-analysis of recurrence based on subgroups, including age subgroups, showed similar trends. The results of this propensity-matched cohort study may not provide a statistically significant association between statin use and a lower risk of HNC recurrence. Further retrospective studies using nationwide claims data and prospective studies are warranted.

Scalable Infrastructure Supporting Reproducible Nationwide Healthcare Data Analysis toward FAIR Stewardship.

Transparent and FAIR disclosure of meta-information about healthcare data and infrastructure is essential but has not been well publicized. In this paper, we provide a transparent disclosure of the process of standardizing a common data model and developing a national data infrastructure using national claims data. We established an Observational Medical Outcome Partnership (OMOP) common data model database for national claims data of the Health Insurance Review and Assessment Service of South Korea. To introduce a data openness policy, we built a distributed data analysis environment and released metadata based on the FAIR principle. A total of 10,098,730,241 claims and 56,579,726 patients' data were converted as OMOP common data model. We also built an analytics environment for distributed research and made the metadata publicly available. Disclosure of this infrastructure to researchers will help to eliminate information inequality and contribute to the generation of high-quality medical evidence.

Impact of symptomatic menopausal transition on the occurrence of depression, anxiety, and sleep disorders: A real-world multi-site study.

BACKGROUND: The menopause transition is a vulnerable period that can be associated with changes in mood and cognition. The present study aimed to investigate whether a symptomatic menopausal transition increases the risks of depression, anxiety, and sleep disorders. METHODS: This population-based, retrospective cohort study analysed data from five electronic health record databases in South Korea. Women aged 45-64 years with and without symptomatic menopausal transition were matched 1:1 using propensity-score matching. Subgroup analyses were conducted according to age and use of hormone replacement therapy (HRT). A primary analysis of 5-year follow-up data was conducted, and an intention-to-treat analysis was performed to identify different risk windows over 5 or 10 years. The primary outcome was first-time diagnosis of depression, anxiety, and sleep disorder. We used Cox proportional hazard models and a meta-analysis to calculate the summary hazard ratio (HR) estimates across the databases. RESULTS: Propensity-score matching resulted in a sample of 17,098 women. Summary HRs for depression (2.10; 95% confidence interval [CI] 1.63-2.71), anxiety (1.64; 95% CI 1.01-2.66), and sleep disorders (1.47; 95% CI 1.16-1.88) were higher in the symptomatic menopausal transition group. In the subgroup analysis, the use of HRT was associated with an increased risk of depression (2.21; 95% CI 1.07-4.55) and sleep disorders (2.51; 95% CI 1.25-5.04) when compared with non-use of HRT. CONCLUSIONS: Our findings suggest that women with symptomatic menopausal transition exhibit an increased risk of developing depression, anxiety, and sleep disorders. Therefore, women experiencing a symptomatic menopausal transition should be monitored closely so that interventions can be applied early.

Use of eye tracking to improve the identification of attention-deficit/hyperactivity disorder in children.

Attention-deficit/hyperactivity disorder (ADHD) is the most common neurodevelopmental disorder of childhood. Although it requires timely detection and intervention, existing continuous performance tests (CPTs) have limited efficacy. Research suggests that eye movement could offer important diagnostic information for ADHD. This study aimed to compare the performance of eye-tracking with that of CPTs, both alone and in combination, and to evaluate the effect of medication on eye movement and CPT outcomes. We recruited participants into an ADHD group and a healthy control group between July 2021 and March 2022 from among children aged 6-10 years (n = 30 per group). The integration of eye-tracking with CPTs produced higher values for the area under the receiver operating characteristic (AUC, 0.889) compared with using CPTs only (AUC, 0.769) for identifying patients with ADHD. The use of eye-tracking alone showed higher performance compare with the use of CPTs alone (AUC of EYE: 0.856, AUC of CPT: 0.769, p = 0.029). Follow-up analysis revealed that most eye-tracking and CPT indicators improved significantly after taking an ADHD medication. The use of eye movement scales could be used to differentiate children with ADHD, with the possibility that integrating eye movement scales and CPTs could improve diagnostic precision.

Translation of Machine Learning-Based Prediction Algorithms to Personalised Empiric Antibiotic Selection: A Population-Based Cohort Study.

BACKGROUND: Prediction of antibiotic non-susceptibility based on patient characteristics and clinical status may support selection of empiric antibiotics for suspected hospital-acquired urinary tract infections (HA-UTIs). METHODS: Prediction models were developed to predict non-susceptible results of eight antibiotic susceptibility tests ordered for suspected HA-UTI. Eligible patients were those with urine culture and susceptibility test results after 48 hours of admission between 2010-2021. Patient demographics, diagnosis, prescriptions, exposure to multidrug-resistant organisms, transfer history, and a daily calculated antibiogram were used as predictors. Lasso logistic regression (LLR), extreme gradient boosting (XGB), random forest, and stacked ensemble methods were used for development. Parsimonious models were also developed for clinical utility. Discrimination was assessed using the area under the receiver operating characteristic curve (AUROC). RESULTS: In 10 474 suspected HA-UTI cases, the mean age was 62.1 ± 16.2 years and 48.1% were male. Non-susceptibility prediction for ampicillin/sulbactam, cefepime, ciprofloxacin, imipenem, piperacillin/tazobactam, and trimethoprim/sulfamethoxazole performed best using the stacked ensemble (AUROC 76.9, 76.1, 77.0, 80.6, 76.1, and 76.5, respectively). The model for ampicillin performed best with LLR (AUROC 73.4). Extreme gradient boosting only performed best for gentamicin (AUROC 66.9). In the parsimonious models, the LLR yielded the highest AUROC for ampicillin, ampicillin/sulbactam, cefepime, gentamicin, and trimethoprim/sulfamethoxazole (AUROC 70.6, 71.8, 73.0, 65.9, and 73.0, respectively). The model for ciprofloxacin performed best with XGB (AUROC 70.3), while the model for imipenem performed best in the stacked ensemble (AUROC 71.3). A personalised application using the parsimonious models was publicly released. CONCLUSIONS: Prediction models for antibiotic non-susceptibility were developed to support empiric antibiotic selection for HA-UTI.