Establishment and evaluation of prediction model of recurrence after laparoscopic choledocholithotomy.

BACKGROUND: Choledocholithiasis is a common clinical bile duct disease, laparoscopic choledocholithotomy is the main clinical treatment method for choledocholithiasis. However, the recurrence of postoperative stones is a big challenge for patients and doctors. AIM: To explore the related risk factors of gallstone recurrence after laparoscopic choledocholithotomy, establish and evaluate a clinical prediction model. METHODS: A total of 254 patients who underwent laparoscopic choledocholithotomy in the First Affiliated Hospital of Ningbo University from December 2017 to December 2020 were selected as the research subjects. Clinical data of the patients were collected, and the recurrence of gallstones was recorded based on the postoperative follow-up. The results were analyzed and a clinical prediction model was established. RESULTS: Postoperative stone recurrence rate was 10.23% (26 patients). Multivariate Logistic regression analysis showed that cholangitis, the diameter of the common bile duct, the diameter of the stone, number of stones, lithotripsy, preoperative total bilirubin, and T tube were risk factors associated with postoperative recurrence (P < 0.05). The clinical prediction model was ln (p/1-p) = -6.853 + 1.347 × cholangitis + 1.535 × choledochal diameter + 2.176 × stone diameter + 1.784 × stone number + 2.242 × lithotripsy + 0.021 × preoperative total bilirubin + 2.185 × T tube. CONCLUSION: Cholangitis, the diameter of the common bile duct, the diameter of the stone, number of stones, lithotripsy, preoperative total bilirubin, and T tube are the associated risk factors for postoperative recurrence of gallstone. The prediction model in this study has a good prediction effect, which has a certain reference value for recurrence of gallstone after laparoscopic choledocholithotomy.

Scoring and validation of a simple model for predicting diabetic retinopathy in patients with type 2 diabetes based on a meta-analysis approach of 21 cohorts.

AIM: To develop and validate a model for predicting diabetic retinopathy (DR) in patients with type 2 diabetes. METHODS: All risk factors with statistical significance in the DR prediction model were scored by their weights. Model performance was evaluated by the area under the receiver operating characteristic (ROC) curve, Kaplan-Meier curve, calibration curve and decision curve analysis. The prediction model was externally validated using a validation cohort from a Chinese hospital. RESULTS: In this meta-analysis, 21 cohorts involving 184,737 patients with type 2 diabetes were examined. Sex, smoking, diabetes mellitus (DM) duration, albuminuria, glycated haemoglobin (HbA1c), systolic blood pressure (SBP) and TG were identified to be statistically significant. Thus, they were all included in the model and scored according to their weights (maximum score: 35.0). The model was validated using an external cohort with median follow-up time of 32 months. At a critical value of 16.0, the AUC value, sensitivity and specificity of the validation cohort are 0.772 ((95% confidence interval (95%CI): 0.740-0.803), p < .01), 0.715 and 0.775, respectively. The calibration curve lied close to the ideal diagonal line. Furthermore, the decision curve analysis demonstrated that the model had notably higher net benefits. The external validation results proved the reliability of the risk prediction model. CONCLUSIONS: The simple DR prediction model developed has good overall calibration and discrimination performance. It can be used as a simple tool to detect patients at high risk of DR.

Comprehensive comorbidity assessment for the ECOPD: a long-term multi-centre retrospective study.

BACKGROUND: Exacerbation of chronic obstructive pulmonary disease (ECOPD) results in severe adverse outcomes and mortality. It is often associated with increased local and systemic inflammation. However, individual susceptibility to exacerbations remains largely unknown. Our study aimed to investigate the association between comorbidities and exacerbation outcomes. METHODS: We included patients with the primary discharge diagnosis of exacerbation for more 10 years in China. Data on all comorbidities were collected and analysed to determine the impact of the comorbidities on 1-year exacerbation readmission, length of hospital stay, and hospital cost. Univariable and multivariable logistic regression analyses were performed, and predictive models were developed. RESULTS: This extensive investigation evaluated a total of 15,708 individuals from five prominent locations in China, revealing notable variations in the prevalence of comorbidities and healthcare expenses among different regions. The study shows that there is a high rate of readmission within one year, namely 15.8%. The most common conditions among readmitted patients are hypertension (38.6%), ischemic heart disease (16.9%), and diabetes mellitus (16.6%). An extensive multivariable study revealed that age, gender, and particular comorbidities such as malnutrition and hyperlipidemia are important factors that can significantly predict greater readmission rates, longer hospital stays or increased healthcare costs. The multivariable models show a moderate to good ability to predict patient outcomes, with concordance index ranging from 0.701 to 0.752. This suggests that targeted interventions in these areas could improve patient outcomes and make better use of healthcare resources. CONCLUSIONS: The results regarding the association between severe exacerbations and systemic disease status support the integration of systematic evaluation of comorbidities into the management of exacerbations and the intensification of treatment of important comorbidities as a appropriate measure for prevention of further exacerbations. Our models also provide a novel tool for clinicians to determine the risk of the 1-year recurrence of severe ECOPD in hospitalised patients.

Prediction begins with diagnosis: Estimating seizure recurrence risk in the First Seizure Clinic.

OBJECTIVES: To assess the feasibility of using a seizure recurrence prediction tool in a First Seizure Clinic, considering (1) the accuracy of initial clinical diagnoses and (2) performance of automated computational models in predicting seizure recurrence after first unprovoked seizure (FUS). METHODS: To assess diagnostic accuracy, we analysed all sustained and revised diagnoses in patients seen at a First Seizure Clinic over 5 years with 6+ months follow-up ('accuracy cohort', n = 487). To estimate prediction of 12-month seizure recurrence after FUS, we used a logistic regression of clinical factors on a multicentre FUS cohort ('prediction cohort', n = 181), and compared performance to a recently published seizure recurrence model. RESULTS: Initial diagnosis was sustained over 6+ months follow-up in 69% of patients in the 'accuracy cohort'. Misdiagnosis occurred in 5%, and determination of unclassified diagnosis in 9%. Progression to epilepsy occurred in 17%, either following FUS or initial acute symptomatic seizure. Within the 'prediction cohort' with FUS, 12-month seizure recurrence rate was 41% (95% CI [33.8%, 48.5%]). Nocturnal seizure, focal seizure semiology and developmental disability were predictive factors. Our model yielded an Area under the Receiver Operating Characteristic curve (AUC) of 0.60 (95% CI [0.59, 0.64]). CONCLUSIONS: High clinical accuracy can be achieved at the initial visit to a First Seizure Clinic. This shows that diagnosis will not limit the application of seizure recurrence prediction tools in this context. However, based on the modest performance of currently available seizure recurrence prediction tools using clinical factors, we conclude that data beyond clinical factors alone will be needed to improve predictive performance.

Assessment and comparison of model estimated and directly observed weather data for prediction of diarrhoea aetiology.

Recent advances in clinical prediction for diarrhoeal aetiology in low- and middle-income countries have revealed that the addition of weather data to clinical data improves predictive performance. However, the optimal source of weather data remains unclear. We aim to compare the use of model estimated satellite- and ground-based observational data with weather station directly observed data for the prediction of aetiology of diarrhoea. We used clinical and etiological data from a large multi-centre study of children with moderate to severe diarrhoea cases to compare their predictive performances. We show that the two sources of weather conditions perform similarly in most locations. We conclude that while model estimated data is a viable, scalable tool for public health interventions and disease prediction, given its ease of access, directly observed weather station data is likely adequate for the prediction of diarrhoeal aetiology in children in low- and middle-income countries.

Development and validation of a risk prediction model for linezolid-induced anemia in elderly patients: a retrospective cohort study.

Background: Linezolid-induced anemia (LI-AN) is a severe adverse reaction, but risk factors of the LI-AN for elderly patients have not been established. Objectives: The objective of this study was to develop a nomogram capable of predicting LI-AN in elderly patients. Design: This is a retrospective study to develop and validate a nomogram for anemia prediction in elderly patients treated with linezolid. Methods: We retrospectively screened elderly patients treated with linezolid at Inner Mongolia People's Hospital from January 2020 to December 2023 and validated our findings using the MIMIC-IV 2.2 database. Anemia was defined as hemoglobin reduction to 75% of baseline value. Univariate and multivariable logistic regression models were used to identify predictors and construct the nomogram, which was evaluated using receiver operating characteristic (ROC) curve analysis, calibration plot, and decision curve analysis. Results: A total of 231 patients were enrolled in this study. The training set comprised 151 individuals, and anemia occurred in 28 cases (18.54%). In the external validation set of 80 individuals, 26 (32.5%) were diagnosed with anemia. The predictors included duration of linezolid therapy, patient estimated glomerular filtration rate value, and sequential organ failure assessment score ⩾2. The ROC curve for the training set was 0.830 (95% CI: 0.750-0.910), while a similar ROC curve of 0.743 (95% CI: 0.621-0.865) was obtained for the validation set. The calibration curve demonstrated good correlation between predicted and observed results, indicating that this study effectively predicts risk factors associated with LI-AN in elderly patients. Conclusion: The developed prediction model can provide valuable guidance for clinicians to prevent anemia and facilitate rational linezolid use in elderly patients.

Study analyzing the clinical data of elderly patients using linezolid to better understand what factors may contribute to anemia in patients Why was the study done? This study aimed to develop a tool that predicts the risk of anemia in elderly patients treated with linezolid, a medication that can cause severe side effects like low hemoglobin levels. Identifying factors that contribute to this adverse reaction can help doctors prevent it and ensure safer use of linezolid. What did the researchers do? The researchers studied the medical records of elderly patients treated with linezolid at Inner Mongolia People's Hospital over a 4-year period. To better understand which factors are related to the occurrence of anemia, so we can find ways to predict the occurrence of problems. What did the researchers find? Factors that increase the risk of anemia after using linezolid include the duration of use of linezolid, kidney function, and SOFA score, that is, the longer the use of linezolid, the worse the kidney function, the higher the SOFA score, and the more likely the patient is to develop anemia. What do the findings mean? The researchers successfully created a tool, called a prediction model, which can help doctors predict the likelihood of anemia in elderly patients taking linezolid. This can guide clinicians in monitoring and managing patients more effectively, potentially reducing the occurrence of anemia and ensuring safer use of linezolid in elderly populations.

A predictive model for stunting among children under the age of three.

Background: In light of the global effort to eradicate stunting in childhood, the objective of this research endeavor was to assess the prevalence of stunting and associated factors, simultaneously construct and validate a risk prediction model for stunting among children under the age of three in Shenzhen, China. Methods: Using the stratified random sampling method, we selected 9,581 children under the age of three for research and analysis. The dataset underwent a random allocation into training and validation sets, adhering to a 8:2 split ratio. Within the training set, a combined approach of LASSO regression analysis and binary logistic regression analysis was implemented to identify and select the predictive variables for the model. Subsequently, model construction was conducted in the training set, encompassing model evaluation, visualization, and internal validation procedures. Finally, to assess the model's generalizability, external validation was performed using the validation set. Results: A total of 684 (7.14%) had phenotypes of stunt. Utilizing a combined approach of LASSO regression and logistic regression, key predictors of stunting among children under three years of age were identified, including sex, age in months, mother's education, father's age, birth order, feeding patterns, delivery mode, average daily parent-child reading time, average time spent in child-parent interactions, and average daily outdoor time. These variables were subsequently employed to develop a comprehensive prediction model for childhood stunting. A nomogram model was constructed based on these factors, demonstrating excellent consistency and accuracy. Calibration curves validated the agreement between the nomogram predictions and actual observations. Furthermore, ROC and DCA analyses indicated the strong predictive performance of the nomograms. Conclusions: The developed model for forecasting stunt risk, which integrates a spectrum of variables. This analytical framework presents actionable intelligence to medical professionals, laying down a foundational framework and a pivot for the conception and execution of preemptive strategies and therapeutic interventions.

Stakeholder-informed positivity thresholds for disease markers and risk scores: a methodological framework and an application in obstructive lung disease.

OBJECTIVES: A positivity threshold is often applied to markers or predicted risks to guide disease management. These thresholds are often decided exclusively by clinical experts despite being sensitive to the preferences of patients and general public as ultimate stakeholders. STUDY DESIGN AND SETTING: We propose an analytical framework for quantifying the net benefit (NB) of an evidence-based positivity threshold based on combining preference-sensitive (eg, how individuals weight benefits and harms of treatment) and preference-agnostic (eg, the magnitude of benefit and the risk of harm) parameters. We propose parsimonious choice experiments to elicit preference-sensitive parameters from stakeholders, and targeted evidence synthesis to quantify the value of preference-agnostic parameters. We apply this framework to maintenance of azithromycin therapy for chronic obstructive pulmonary disease using a discrete choice experiment to estimate preference weights for attribute level associated with treatment. We identify the positivity threshold on 12-month moderate or severe exacerbation risk that would maximize the NB of treatment in terms of severe exacerbations avoided. RESULTS: In the case study, the prevention of moderate and severe exacerbations (benefits) and the risk of hearing loss and gastrointestinal symptoms (harms) emerged as important attributes. Four hundred seventy seven respondents completed the discrete choice experiment survey. Relative to each percent risk of severe exacerbation, preference weights for each percent risk of moderate exacerbation, hearing loss, and gastrointestinal symptoms were 0.395 (95% confidence interval [CI] 0.338-0.456), 1.180 (95% CI 1.071-1.201), and 0.253 (95% CI 0.207-0.299), respectively. The optimal threshold that maximized NB was to treat patients with a 12-month risk of moderate or severe exacerbations ≥12%. CONCLUSION: The proposed methodology can be applied to many contexts where the objective is to devise positivity thresholds that need to incorporate stakeholder preferences. Applying this framework to chronic obstructive pulmonary disease pharmacotherapy resulted in a stakeholder-informed treatment threshold that was substantially lower than the implicit thresholds in contemporary guidelines. PLAIN LANGUAGE SUMMARY: Doctors often compare disease markers (such as laboratory results) or risk scores for a patient with cut-off values from guidelines to decide which patients need to be treated. For example, guidelines recommend that patients whose 10-year risk of heart attack is more than 10% be given statin pills. However, guidelines that recommend such treatment rules might not consider what matters most to patients (like how much they do not like side effects of the drugs). In this study, we propose a mathematical method where preferences of individuals on the trade-off between treatment benefits and harms can be used to determine the best treatment rule. We apply this method to the choice of antibiotic therapy for patients with lung airway diseases. We find that, given patient and public preferences on treatment benefit and risks, those with a 12% or more risk of experiencing a lung attack should receive antibiotic therapy. This patient-oriented cut-off is significantly lower than the cut-off values currently used by guidelines, which are in the 60%-70% range. We recommend applying this method whenever scientists must make recommendations on treatment rules where patient or public preferences might influence those rules.

Correlation between coronary calcification and cardiac structure in non-dialysis patients with chronic kidney disease.

AIMS: We aim to explore the correlation between coronary artery calcification (CAC) score (CACS) and cardiac structure and function in chronic kidney disease (CKD) patients, create a clinical prediction model for severe CAC associated with cardiac ultrasound indexes. METHODS AND RESULTS: The study included 178 non-dialysis CKD patients who underwent CACS testing and collected general information, serological indices, cardiac ultrasound findings and follow-up on renal function, heart failure (HF) manifestations and re-hospitalization. The mean age of participants in the study cohort was 67.4 years; 59% were male, and 66.9% of patients had varying degrees of comorbid CAC. CKD patients with CACS > 100 were older, predominantly male and had a higher proportion of smoking, diabetes and hypertension (P < 0.05) compared with those with CACS = 0 and 0 < CACS ≤ 100, and had higher brain natriuretic peptide, serum magnesium and fibrinogen levels were also higher (P < 0.05). CACS was positively correlated with left atrial inner diameter (LAD), left ventricular end-diastolic inner diameter (LVDd), left ventricular volume at diastole (LVVd), output per beat (SV) and mitral orifice early diastolic blood flow velocity/early mitral annular diastolic myocardial motion velocity (E/e) (P < 0.05). We tested the associations between varying degrees of CAC and HF and heart valve calcification using multivariable-adjusted regression models. The risk of HF in patients with severe CAC was about 1.95 times higher than that in patients without coronary calcification, and the risk of heart valve calcification was 2.46 times higher than that in patients without coronary calcification. Heart valve calcification and HF diagnosis, LAD and LVDd are essential in predicting severe CAC. During a mean follow-up time of 18.26 ± 10.17 months, 65 (36.52%) patients had a composite renal endpoint event, of which 36 (20.22%) were admitted to renal replacement therapy. Patients with severe CAC had a higher risk of progression of renal function, re-admission due to cardiovascular and renal events and more pronounced symptoms of HF (P < 0.05). CONCLUSIONS: There is a correlation between CACS and cardiac structure and function in non-dialysis CKD patients, which may mainly involve abnormalities in left ventricular structure and cardiac diastolic function. CAC may affect renal prognosis and quality of survival in CKD patients. Based on clinical information, HF, valvular calcification status and indicators related to left ventricular hypertrophy can identify people at risk for severe CAC.

Development of a diagnostic model for detecting mild cognitive impairment in young and middle-aged patients with obstructive sleep apnea: a prospective observational study.

Objectives: Obstructive sleep apnea (OSA) is a common sleep-disordered breathing condition linked to the accelerated onset of mild cognitive impairment (MCI). However, the prevalence of undiagnosed MCI among OSA patients is high and attributable to the complexity and specialized nature of MCI diagnosis. Timely identification and intervention for MCI can potentially prevent or delay the onset of dementia. This study aimed to develop screening models for MCI in OSA patients that will be suitable for healthcare professionals in diverse settings and can be effectively utilized without specialized neurological training. Methods: A prospective observational study was conducted at a specialized sleep medicine center from April 2021 to September 2022. Three hundred and fifty consecutive patients (age: 18-60 years) suspected OSA, underwent the Montreal Cognitive Assessment (MoCA) and polysomnography overnight. Demographic and clinical data, including polysomnographic sleep parameters and additional cognitive function assessments were collected from OSA patients. The data were divided into training (70%) and validation (30%) sets, and predictors of MCI were identified using univariate and multivariate logistic regression analyses. Models were evaluated for predictive accuracy and calibration, with nomograms for application. Results: Two hundred and thirty-three patients with newly diagnosed OSA were enrolled. The proportion of patients with MCI was 38.2%. Three diagnostic models, each with an accompanying nomogram, were developed. Model 1 utilized body mass index (BMI) and years of education as predictors. Model 2 incorporated N1 and the score of backward task of the digital span test (DST_B) into the base of Model 1. Model 3 expanded upon Model 1 by including the total score of digital span test (DST). Each of these models exhibited robust discriminatory power and calibration. The C-statistics for Model 1, 2, and 3 were 0.803 [95% confidence interval (CI): 0.735-0.872], 0.849 (95% CI: 0.788-0.910), and 0.83 (95% CI: 0.763-0.896), respectively. Conclusion: Three straightforward diagnostic models, each requiring only two to four easily accessible parameters, were developed that demonstrated high efficacy. These models offer a convenient diagnostic tool for healthcare professionals in diverse healthcare settings, facilitating timely and necessary further evaluation and intervention for OSA patients at an increased risk of MCI.

Clinician perspectives and recommendations regarding design of clinical prediction models for deteriorating patients in acute care.

BACKGROUND: Successful deployment of clinical prediction models for clinical deterioration relates not only to predictive performance but to integration into the decision making process. Models may demonstrate good discrimination and calibration, but fail to match the needs of practising acute care clinicians who receive, interpret, and act upon model outputs or alerts. We sought to understand how prediction models for clinical deterioration, also known as early warning scores (EWS), influence the decision-making of clinicians who regularly use them and elicit their perspectives on model design to guide future deterioration model development and implementation. METHODS: Nurses and doctors who regularly receive or respond to EWS alerts in two digital metropolitan hospitals were interviewed for up to one hour between February 2022 and March 2023 using semi-structured formats. We grouped interview data into sub-themes and then into general themes using reflexive thematic analysis. Themes were then mapped to a model of clinical decision making using deductive framework mapping to develop a set of practical recommendations for future deterioration model development and deployment. RESULTS: Fifteen nurses (n = 8) and doctors (n = 7) were interviewed for a mean duration of 42 min. Participants emphasised the importance of using predictive tools for supporting rather than supplanting critical thinking, avoiding over-protocolising care, incorporating important contextual information and focusing on how clinicians generate, test, and select diagnostic hypotheses when managing deteriorating patients. These themes were incorporated into a conceptual model which informed recommendations that clinical deterioration prediction models demonstrate transparency and interactivity, generate outputs tailored to the tasks and responsibilities of end-users, avoid priming clinicians with potential diagnoses before patients were physically assessed, and support the process of deciding upon subsequent management. CONCLUSIONS: Prediction models for deteriorating inpatients may be more impactful if they are designed in accordance with the decision-making processes of acute care clinicians. Models should produce actionable outputs that assist with, rather than supplant, critical thinking.

Conditional Similarity Triplets Enable Covariate-Informed Representations of Single-Cell Data.

Single-cell technologies enable comprehensive profiling of diverse immune cell-types through the measurement of multiple genes or proteins per cell. In order to translate data from immune profiling assays into powerful diagnostics, machine learning approaches are used to compute per-sample immunological summaries, or featurizations that can be used as inputs to models for outcomes of interest. Current supervised learning approaches for computing per-sample representations are optimized based only on the outcome variable to be predicted and do not take into account clinically-relevant covariates that are likely to also be measured. Here we expand the optimization problem to also take into account such additional patient covariates to directly inform the learned per-sample representations. To do this, we introduce CytoCoSet, a set-based encoding method, which formulates a loss function with an additional triplet term penalizing samples with similar covariates from having disparate embedding results in per-sample representations. Overall, incorporating clinical covariates leads to improved prediction of clinical phenotypes.

Implementation of a dynamic model updating pipeline provides a systematic process for maintaining performance of prediction models.

OBJECTIVES: We describe the steps for implementing a dynamic updating pipeline for clinical prediction models and illustrate the proposed methods in an application of 5-year survival prediction in cystic fibrosis. STUDY DESIGN AND SETTING: Dynamic model updating refers to the process of repeated updating of a clinical prediction model with new information to counter performance degradation. We describe 2 types of updating pipeline: "proactive updating" where candidate model updates are tested any time new data are available, and "reactive updating" where updates are only made when performance of the current model declines or the model structure changes. Methods for selecting the best candidate updating model are based on measures of predictive performance under the 2 pipelines. The methods are illustrated in our motivating example of a 5-year survival prediction model in cystic fibrosis. Over a dynamic updating period of 10 years, we report the updating decisions made and the performance of the prediction models selected under each pipeline. RESULTS: Both the proactive and reactive updating pipelines produced survival prediction models that overall had better performance in terms of calibration and discrimination than a model that was not updated. Further, use of the dynamic updating pipelines ensured that the prediction model's performance was consistently and frequently reviewed in new data. CONCLUSION: Implementing a dynamic updating pipeline will help guard against model performance degradation while ensuring that the updating process is principled and data-driven.

Artificial Intelligence-Augmented Clinical Decision Support Systems for Pregnancy Care: Systematic Review.

BACKGROUND: Despite the emerging application of clinical decision support systems (CDSS) in pregnancy care and the proliferation of artificial intelligence (AI) over the last decade, it remains understudied regarding the role of AI in CDSS specialized for pregnancy care. OBJECTIVE: To identify and synthesize AI-augmented CDSS in pregnancy care, CDSS functionality, AI methodologies, and clinical implementation, we reported a systematic review based on empirical studies that examined AI-augmented CDSS in pregnancy care. METHODS: We retrieved studies that examined AI-augmented CDSS in pregnancy care using database queries involved with titles, abstracts, keywords, and MeSH (Medical Subject Headings) terms. Bibliographic records from their inception to 2022 were retrieved from PubMed/MEDLINE (n=206), Embase (n=101), and ACM Digital Library (n=377), followed by eligibility screening and literature review. The eligibility criteria include empirical studies that (1) developed or tested AI methods, (2) developed or tested CDSS or CDSS components, and (3) focused on pregnancy care. Data of studies used for review and appraisal include title, abstract, keywords, MeSH terms, full text, and supplements. Publications with ancillary information or overlapping outcomes were synthesized as one single study. Reviewers independently reviewed and assessed the quality of selected studies. RESULTS: We identified 30 distinct studies of 684 studies from their inception to 2022. Topics of clinical applications covered AI-augmented CDSS from prenatal, early pregnancy, obstetric care, and postpartum care. Topics of CDSS functions include diagnostic support, clinical prediction, therapeutics recommendation, and knowledge base. CONCLUSIONS: Our review acknowledged recent advances in CDSS studies including early diagnosis of prenatal abnormalities, cost-effective surveillance, prenatal ultrasound support, and ontology development. To recommend future directions, we also noted key gaps from existing studies, including (1) decision support in current childbirth deliveries without using observational data from consequential fetal or maternal outcomes in future pregnancies; (2) scarcity of studies in identifying several high-profile biases from CDSS, including social determinants of health highlighted by the American College of Obstetricians and Gynecologists; and (3) chasm between internally validated CDSS models, external validity, and clinical implementation.

Predictive etiological classification of acute ischemic stroke through interpretable machine learning algorithms: a multicenter, prospective cohort study.

BACKGROUND: The prognosis, recurrence rates, and secondary prevention strategies varied significantly among different subtypes of acute ischemic stroke (AIS). Machine learning (ML) techniques can uncover intricate, non-linear relationships within medical data, enabling the identification of factors associated with etiological classification. However, there is currently a lack of research utilizing ML algorithms for predicting AIS etiology. OBJECTIVE: We aimed to use interpretable ML algorithms to develop AIS etiology prediction models, identify critical factors in etiology classification, and enhance existing clinical categorization. METHODS: This study involved patients with the Third China National Stroke Registry (CNSR-III). Nine models, which included Natural Gradient Boosting (NGBoost), Categorical Boosting (CatBoost), Extreme Gradient Boosting (XGBoost), Random Forest (RF), Light Gradient Boosting Machine (LGBM), Gradient Boosting Decision Tree (GBDT), Adaptive Boosting (AdaBoost), Support Vector Machine (SVM), and logistic regression (LR), were employed to predict large artery atherosclerosis (LAA), small vessel occlusion (SVO), and cardioembolism (CE) using an 80:20 randomly split training and test set. We designed an SFS-XGB with 10-fold cross-validation for feature selection. The primary evaluation metrics for the models included the area under the receiver operating characteristic curve (AUC) for discrimination and the Brier score (or calibration plots) for calibration. RESULTS: A total of 5,213 patients were included, comprising 2,471 (47.4%) with LAA, 2,153 (41.3%) with SVO, and 589 (11.3%) with CE. In both LAA and SVO models, the AUC values of the ML models were significantly higher than that of the LR model (P < 0.001). The optimal model for predicting SVO (AUC [RF model] = 0.932) outperformed the optimal LAA model (AUC [NGB model] = 0.917) and the optimal CE model (AUC [LGBM model] = 0.846). Each model displayed relatively satisfactory calibration. Further analysis showed that the optimal CE model could identify potential CE patients in the undetermined etiology (SUE) group, accounting for 1,900 out of 4,156 (45.7%). CONCLUSIONS: The ML algorithm effectively classified patients with LAA, SVO, and CE, demonstrating superior classification performance compared to the LR model. The optimal ML model can identify potential CE patients among SUE patients. These newly identified predictive factors may complement the existing etiological classification system, enabling clinicians to promptly categorize stroke patients' etiology and initiate optimal strategies for secondary prevention.

Construction of a clinical prediction model for the diagnosis of immune thrombocytopenia based on clinical laboratory parameters.

PURPOSE: Primary immune thrombocytopenia (ITP) is an autoimmune bleeding disorder characterized by isolated thrombocytopenia that is often misdiagnosed due to the lack of a gold standard for diagnosis and currently relies on exclusionary approaches. This project combines several laboratory parameters to construct a clinical prediction model for adult ITP patients. METHODS: A total of 428 patients with thrombocytopenia who visited the West China Hospital of Sichuan University between January 2021 and March 2023 were enrolled. Based on the diagnostic criteria, we divided those patients into an ITP group and a non-ITP group. A total of 34 laboratory parameters were analyzed via univariate analysis and correlation analysis, and the least absolute shrinkage and selection operator regression analysis was used to establish the model. The training and validation sets were divided at a ratio of 7:3, and we used a fivefold cross-validation method to construct the model. RESULTS: The model included the following variables: red blood cell, mean corpuscular hemoglobin concentration, red blood cell distribution width-standard deviation, platelet variability index score, immature platelet fraction, lymphocyte absolute value. The prediction model exhibited good performance, with a sensitivity of 0.89 and a specificity of 0.83 in the training set and a sensitivity of 0.90 and a specificity of 0.87 in the validation set. CONCLUSION: The clinical prediction model can assess the probability of ITP in thrombocytopenic patients and has good predictive accuracy for the diagnosis of ITP.

Application of a combined clinical prediction model based on enhanced T1-weighted image(T1WI) full volume histogram in peripheral nerve invasion (PNI) and lymphatic vessel invasion (LVI) in rectal cancer.

PURPOSE: This study aims to use a combined clinical prediction model based on enhanced T1-weighted image(T1WI) full volume histogram to predict preoperative peripheral nerve invasion (PNI) and lymphatic vessel invasion (LVI) in rectal cancer. METHODS: We included a total of 68 PNI patients and 80 LVI patients who underwent surgical resection and pathological confirmation of rectal cancer. According to the PNI/LVI status, patients were divided into PNI positive group (n = 39), the PNI negative group (n = 29), LVI positive group (n = 48), and the LVI negative group (n = 32). External validation included a total of 42 patients with nerve and vascular invasion in patients with surgically resected and pathologically confirmed rectal cancer at another healthcare facility, with a PNI positive group (n = 32) and a PNI-negative group (n = 10) as well as an LVI positive group (n = 35) and LVI-negative group (n = 7). All patients underwent 3.0T magnetic resonance T1WI enhanced scanning. We use Firevoxel software to delineate the region of interest (ROI), extract histogram parameters, and perform univariate analysis, LASSO regression, and multivariate logistic regression analysis in sequence to screen for the best predictive factors. Then, we constructed a clinical prediction model and plotted it into a column chart for personalized prediction. Finally, we evaluate the performance and clinical practicality of the model based on the area under curve (AUC), calibration curve, and decision curve. RESULTS: Multivariate logistic regression analysis found that variance and the 75th percentile were independent risk factors for PNI, while maximum and variance were independent risk factors for LVI. The clinical prediction model constructed based on the above factors has an AUC of 0.734 (95% CI: 0.591-0.878) for PNI in the training set and 0.731 (95% CI: 0.509-0.952) in the validation set; The training set AUC of LVI is 0.701 (95% CI: 0.561-0.841), and the validation set AUC is 0.685 (95% CI: 0.439-0.932). External validation showed an AUC of 0.722 (95% CI: 0.565-0.878) for PNI; and an AUC of 0.706 (95% CI: 0.481-0.931) for LVI. CONCLUSIONS: This study indicates that the combination of enhanced T1WI full volume histogram and clinical prediction model can be used to predict the perineural and lymphovascular invasion status of rectal cancer before surgery, providing valuable reference information for clinical diagnosis.

Addressing the implementation challenge of risk prediction model due to missing risk factors: The submodel approximation approach.

Clinical prediction models have been widely acknowledged as informative tools providing evidence-based support for clinical decision making. However, prediction models are often underused in clinical practice due to many reasons including missing information upon real-time risk calculation in electronic health records (EHR) system. Existing literature to address this challenge focuses on statistical comparison of various approaches while overlooking the feasibility of their implementation in EHR. In this article, we propose a novel and feasible submodel approach to address this challenge for prediction models developed using the model approximation (also termed "preconditioning") method. The proposed submodel coefficients are equivalent to the corresponding original prediction model coefficients plus a correction factor. Comprehensive simulations were conducted to assess the performance of the proposed method and compared with the existing "one-step-sweep" approach as well as the imputation approach. In general, the simulation results show the preconditioning-based submodel approach is robust to various heterogeneity scenarios and is comparable to the imputation-based approach, while the "one-step-sweep" approach is less robust under certain heterogeneity scenarios. The proposed method was applied to facilitate real-time implementation of a prediction model to identify emergency department patients with acute heart failure who can be safely discharged home.

Acute pulmonary embolism pretest probability estimation by d-dimer test, our modified, new ECG score and clinical prediction rules.

Objectives: We investigated whether a sufficiently sensitive D-dimer test could exclude acute pulmonary embolism (acPE) as a stand-alone diagnostic test and compared our previously published, modified ECG score with the Wells and Geneva scores in the estimation of acPE pretest probability. Methods: We retrospectively evaluated 345 patients who underwent chest CT angiography (CTA) for the suspicion of acPE. The pretest probability of acPE was assessed in 120 D-dimer negative [DD (-)] and 225 D-dimer positive [DD (+)] patients. Results: Chest CTA verified acPE in 57/345 (16.5 %) patients and in 1/120 (0.8 %) DD (-) patient. In DD (-) patients the test accuracy (TA) and specificity (SP) of the ECG score (98 %, 99 %) were better than those of the Wells score (92.5 %, 92.4 %) (p = 0.063 and p < 0.05 respectively) and the Geneva score (76.7 %, 76.5 %) (p < 0.001 for both), the Wells score TA and SP were greater than those of the Geneva score (p < 0.001 for both). In DD (+) patients the SPs, TAs and positive predictive values (PPV) of the ECG score (94 %, 78.6 %, 69 %) and the Wells score (91.8 %, 75.1 %, 48 %) were greater than those of the Geneva score (71.3 %, 64.9 %, 38.2 %) (p < 0.001 for both SP and TA respectively, and p < 0.001 for PPV of the ECG score vs. the Geneva score and p < 0.05 for PPV of the Wells score vs. Geneva score), their sensitivities (SE) (36.4 %, 23.6 %) were less than that of the Geneva score (47.5 %) (p < 0.05 and p < 0.001 respectively). The ECG score's TA in a trend, its SE and PPV were significantly (p < 0.01 and p < 0.001) better than those of the Wells score. Conclusion: In contrast to the current guidelines, a stand-alone high sensitivity DD (-) test, without prediction rules, could reliably exclude acPE. Our ECG score slightly outperformed the Wells score, the ECG score and Wells score far outperformed the Geneva score in the estimation of acPE pretest probability. An acPE diagnosis with the ECG score, in addition to the supportive diagnosis with the clinical prediction rules, may further increase the chance of true DD positivity.