Mi-Young Oh

Stroke, a leading cause of death, requires precise predictions of life expectancy. Traditional scores focused on short-term outcomes and required labor-intensive data collection. Recent machine learning (ML)-based prediction models showed superior performance. However, they were predominantly geared towards classification tasks, omitting crucial temporal aspects. Thus, an unmet need arises for more convenient, time-aware predictive methodologies for long-term mortality in managing acute stroke patients. The developmental cohort consisted of 3,411 patients from two tertiary hospitals, while the external test cohort included 502 patients from a secondary cardiovascular center. We developed various ML- and deep learning (DL)-based models to predict post-ischemic stroke mortality, utilizing clinical data. Among these, we selected the best-performing model and compared it with traditional risk scores. Furthermore, we assessed key features based on their importance, determining their optimal number for practical application. The total number of deceased patients in the developmental cohort was 136, and 118 in the external test cohort. The Gradient Boosting Cox Proportional Hazards model demonstrated the best performance with C-index of 0.785 in internal validation set, and C-index of 0.845 in external test set. It surpassed the conventional risk score, PREMISE score with C-index of 0.783(p = 0.017) in the external dataset. The key features included age, National Institutes of Health Stroke Scale score, hemoglobin, and so on. The study developed and validated a machine learning-based model tailored for post-stroke survival prediction, outperforming existing scores. It could enhance the individual stroke patient’s survival predictions and optimize medical resource allocation.

Background: Brain age, a neurobiological construct derived from structural neuroimaging, reflects overall brain health and holds promise for various clinical applications. Brain resilience, defined as the difference between brain age and chronological age, represents an individual's ability to maintain brain health relative to their age. Lower brain resilience may indicate increased vulnerability to neurodegenerative conditions, while higher resilience suggests better brain health. However, the role of brain age and resilience in post‐stroke dementia has not been extensively studied. Method: Data were collected from 81 patients admitted to Bucheon Sejong Hospital between October 2016 and May 2022. Brain age was derived using FreeSurfer software, which segments T1‐weighted MRI to extract 153 features related to cortical thickness, surface area, and subcortical volume (available at https://github.com/freesurfer/freesurfer ). Of these, 77 key features were used to predict brain age using a ridge regression model trained on a healthy cohort aged 18–75 years (available at https://www.photon‐ai.com/enigma_brainage ). Brain resilience was defined as the difference between brain age and chronological age. Result: Of the 81 patients, 58 were diagnosed with post‐stroke dementia. Brain resilience was significantly associated with post‐stroke dementia (OR: 1.07, 95% CI: 1.01–1.13, p = 0.02). Additionally, higher national institutes of health stroke scale scores were associated with a higher risk of dementia (OR: 1.16, 95% CI: 1.01–1.13, p = 0.02). Conclusion: Our findings suggest that brain resilience, as measured by the difference between brain age and chronological age, may be a critical factor in the development of post‐stroke dementia. These results indicate that brain age and resilience could serve as valuable biomarkers for predicting post‐stroke dementia and guiding clinical interventions.

Background: Machine learning (ML) has the potential to enhance performance by capturing nonlinear interactions. However, ML-based models have some limitations in terms of interpretability. Objective: This study aimed to develop and validate a more comprehensible and efficient ML-based scoring system using SHapley Additive exPlanations (SHAP) values. Methods: We developed and validated the Explainable Automated nonlinear Computation scoring system for Health (EACH) framework score. We developed a CatBoost-based prediction model, identified key features, and automatically detected the top 5 steepest slope change points based on SHAP plots. Subsequently, we developed a scoring system (EACH) and normalized the score. Finally, the EACH score was used to predict perioperative stroke. We developed the EACH score using data from the Seoul National University Hospital cohort and validated it using data from the Boramae Medical Center, which was geographically and temporally different from the development set. Results: When applied for perioperative stroke prediction among 38,737 patients undergoing noncardiac surgery, the EACH score achieved an area under the curve (AUC) of 0.829 (95% CI 0.753-0.892). In the external validation, the EACH score demonstrated superior predictive performance with an AUC of 0.784 (95% CI 0.694-0.871) compared with a traditional score (AUC=0.528, 95% CI 0.457-0.619) and another ML-based scoring generator (AUC=0.564, 95% CI 0.516-0.612). Conclusions: The EACH score is a more precise, explainable ML-based risk tool, proven effective in real-world data. The EACH score outperformed traditional scoring system and other prediction models based on different ML techniques in predicting perioperative stroke.

Background: Perioperative stroke is a severe complication following surgery. To identify patients at risk for perioperative stroke, several prediction models based on the preoperative factors were suggested. Prediction models often focus on preoperative patient characteristics to assess stroke risk. However, most existing models primarily base their predictions on the patient's baseline characteristics before surgery. We aimed to develop a machine‐learning model incorporating both pre‐ and intraoperative variables to predict perioperative stroke. Methods and Results: This study included patients who underwent noncardiac surgery at 2 hospitals with the data of 15 752 patients from Seoul National University Hospital used for development and temporal internal validation, and the data of 449 patients from Boramae Medical Center used for external validation. Perioperative stroke was defined as a newly developed ischemic lesion on diffusion‐weighted imaging within 30 days of surgery. We developed a prediction model composed of pre‐ and intraoperative factors (integrated model) and compared it with a model consisting of preoperative features alone (preoperative model). Perioperative stroke developed in 109 (0.69%) patients in the Seoul National University Hospital group and 11 patients (2.45%) in the Boramae Medical Center group. The integrated model demonstrated superior predictive performance with area under the curve values of 0.824 (95% CI, 0.762–0.880) versus 0.584 (95% CI, 0.499–0.667; P P =0.018) in the external validation, compared to the preoperative model. Conclusions: We suggest that incorporating intraoperative factors into perioperative stroke prediction models can improve their accuracy.

BACKGROUND Machine learning (ML) has the potential to enhance performance by capturing nonlinear interactions. However, ML-based models have some limitations in terms of interpretability. OBJECTIVE This study aimed to develop and validate a more comprehensible and efficient ML-based scoring system using SHapley Additive exPlanations (SHAP) values. METHODS We developed and validated the Explainable Automated nonlinear Computation scoring system for Health (EACH) framework score. We developed a CatBoost-based prediction model, identified key features, and automatically detected the top 5 steepest slope change points based on SHAP plots. Subsequently, we developed a scoring system (EACH) and normalized the score. Finally, the EACH score was used to predict perioperative stroke. We developed the EACH score using data from the Seoul National University Hospital cohort and validated it using data from the Boramae Medical Center, which was geographically and temporally different from the development set. RESULTS When applied for perioperative stroke prediction among 38,737 patients undergoing noncardiac surgery, the EACH score achieved an area under the curve (AUC) of 0.829 (95% CI 0.753-0.892). In the external validation, the EACH score demonstrated superior predictive performance with an AUC of 0.784 (95% CI 0.694-0.871) compared with a traditional score (AUC=0.528, 95% CI 0.457-0.619) and another ML-based scoring generator (AUC=0.564, 95% CI 0.516-0.612). CONCLUSIONS The EACH score is a more precise, explainable ML-based risk tool, proven effective in real-world data. The EACH score outperformed traditional scoring system and other prediction models based on different ML techniques in predicting perioperative stroke.

This paper examines 730 examples of auxiliary verbs indicating the past tense ‘きki’ and ‘けりkeri’ and the perfective aspect ‘つtsu’, ‘ぬnu’, ‘たりtari’ and ‘りri’ in Mubatsu-bon Lunyu Jijie volume 1 (abb. Mubatsu-bon), which is a standard glossed reading in the Medieval Japan. There are 11 examples of the past auxiliary verb ‘ki’, regardless of whether there are any factors that indicate the past tense in the Chinese original text. Of these 11, there are only 3 examples of ‘ki’ found in Shibunkai Kunten Lunyu. Further, ‘keri’ is used in 1 example only in Mubatsu-bon, but semantically, it is the case where the continuous aspect is expected to be used. The past tense auxiliary verbs are seldom used in the glossed reading of Lunyu including Mubatsu-bon. This is because the ‘historical present tense’ is used even in the case that describes the past. This is to express vividly the relevant scene, but is also related to the characteristic of the literary style of Kanbun Kundoku whereby Chinese text is translated into Japanese simultaneously to assist with remembering and memorizing the original text. There are 21 examples of ‘tsu’ (11 examples are used as perfective aspect, and 10 examples are given as emphasis), and 35 examples of ‘nu’ (22 examples are used as perfective aspect, and 13 examples are given as emphasis), 24 examples of ‘tari’ (12 examples are used as perfective aspect, and 12 examples are given as continuous aspect), and 67 examples of ‘ri’ (13 examples are used as perfective aspect, and 54 examples are given as continuous aspect). It is clarified as a result of the analysis that the aspect-related factor, which was discussed in Edwin Pulleyblank (2010), does not always appear in all original Chinese texts. Rather there is no consistent tendency as shown in the fact that Japanese auxiliary verbs of continuous aspect are used in the Chinese particles that express the perfective aspect in some examples, whereas the opposite is also found in some other examples. In terms of a succession of Kundoku, ‘tari’ shows the highest frequency rate, followed by ‘ri’. The other auxiliaries ‘tsu’ and ‘nu’ are seldom inherited.

Background: Chronic renal insufficiency, diagnosed using creatinine based estimated glomerular filtration rate (GFR) or microalbumiuria, has been associated with the presence of cerebral microbleeds (CMBs). Cystatin C has been shown to be a more sensitive renal indicator than conventional renal markers. Under the assumption that similar pathologic mechanisms of the small vessel exist in the brain and kidney, we hypothesized that the levels of cystatin C may delineate the relationship between CMBs and renal insufficiency by detecting subclinical kidney dysfunction, which may be underestimated by other indicators, and thus reflect the severity of CMBs more accurately.Methods: Data was prospectively collected for 683 patients with ischemic stroke. The severity of CMBs was categorized by the number of lesions. Patients were divided into quartiles of cystatin C, estimated GFR and microalbumin/creatinine ratios. Ordinal logistic regression analysis was used to examine the association of each renal indicator with CMBs.Results: In models including both quartiles of cystatin C and estimated GFR, only cystatin C quartiles were significant (the highest vs. the lowest, adjusted OR, 1.88; 95% CI 1.05-3.38;p= 0.03) in contrast to estimated GFR (the highest vs. the lowest, adjusted OR, 1.28; 95% CI 0.38-4.36;p= 0.70). A model including both quartiles of cystatin C and microalbumin/creatinine ratio also showed that only cystatin C quartiles was associated with CMBs (the highest vs. the lowest, adjusted OR, 2.06; 95% CI 1.07-3.94;p= 0.03). These associations were also observed in the logistic models using log transformed-cystatin C, albumin/creatinine ratio and estimated GFR as continuous variables. Cystatin C was a significant indicator of deep or infratenorial CMBs, but not strictly lobar CMBs. In addition, cystatin C showed the greatest significance in c-statistics for the presence of CMBs (AUC = 0.73 ± 0.03; 95% CI 0.66-0.76;p= 0.02).Conclusion: Cystatin C may be the most sensitive indicator of CMB severity among the renal disease markers.