Diagnostic Prediction Model Research Articles

Prognostic nomogram for proliferative verrucous leukoplakia

BackgroundProliferative verrucous leukoplakia (PVL) is a special type of leukoplakia characterized by high rate of malignant transformation into oral squamous cell carcinoma (OSCC). This study aimed to analyze the canceration risk and prognostic factors of PVL and establish effective diagnostic and prognostic predictive models. Materials and methodsA total of 467 patients were enrolled, including 170 cases of PVL. The independent risk and prognostic factors of PVL were analyzed by univariable and multivariable logistic regression. Nomogram models were constructed to predict the canceration risk and prognosis of PVL. The predictive power was evaluated by Hosmer–Lemeshow test, receiver operating characteristic (ROC) curve, calibration curve and decision curve analysis. ResultsMultivariable logistic regression analyses identified that canceration risk factors of PVL included sex, lesion sites, clinical presentation, non-smoker and oral epithelial dysplasia (OED). The independent prognostic factors of PVL were sex, clinical presentation, local irritants and OED. Diagnosis and prognostic nomogram models were constructed. The areas under the ROC curve were 0.945 and 0.893, respectively. The calibration plots showed strong agreement between the prediction and observation. Decision curve analysis indicated that the models provided significant clinical benefits for patients. ConclusionOur study established and validated the diagnosis and prognostic predictive nomogram models, which were accurate to predict the canceration risk and prognostic factors of PVL, providing individualized clinical decisions for clinical work.

Intelligent diagnosis of Kawasaki disease from real-world data using interpretable machine learning models

ObjectiveThis study aimed to leverage real-world electronic medical record data to develop interpretable machine learning models for diagnosis of Kawasaki disease while also exploring and prioritizing the significant risk factors. MethodsA comprehensive study was conducted on 4087 pediatric patients at the Children’s Hospital of Chongqing, China. The study collected demographic data, physical examination results, and laboratory findings. Statistical analyses were performed using IBM SPSS Statistics, Version 26.0. The optimal feature subset was used to develop intelligent diagnostic prediction models based on the Light Gradient Boosting Machine, Explainable Boosting Machine (EBM), Gradient Boosting Classifier (GBC), Fast Interpretable Greedy-Tree Sums, Decision Tree, AdaBoost Classifier, and Logistic Regression. Model performance was evaluated in three dimensions: discriminative ability via receiver operating characteristic curves, calibration accuracy using calibration curves, and interpretability through SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-Agnostic Explanations). ResultsIn this study, Kawasaki disease was diagnosed in 2971 participants. Analysis was conducted on 31 indicators, including red blood cell distribution width and erythrocyte sedimentation rate. The EBM model demonstrated superior performance relative to other models, with an area under the curve of 0.97, second only to the GBC model. Furthermore, the EBM model exhibited the highest calibration accuracy and maintained its interpretability without relying on external analytical tools such as SHAP and LIME, thus reducing interpretation biases. Platelet distribution width, total protein, and erythrocyte sedimentation rate were identified by the model as significant predictors for the diagnosis of Kawasaki disease. ConclusionThis study used diverse machine learning models for early diagnosis of Kawasaki disease. The findings demonstrated that interpretable models such as EBM outperformed traditional machine learning models in terms of both interpretability and performance. Ensuring consistency between predictive models and clinical evidence is crucial for the successful integration of artificial intelligence into real-world clinical practice.

Identification of Blood Biomarkers Related to Energy Metabolism and Construction of Diagnostic Prediction Model Based on Three Independent Alzheimer's Disease Cohorts.

Blood biomarkers are crucial for the diagnosis and therapy of Alzheimer's disease (AD). Energy metabolism disturbances are closely related to AD. However, research on blood biomarkers related to energy metabolism is still insufficient. This study aims to explore the diagnostic and therapeutic significance of energy metabolism-related genes in AD. AD cohorts were obtained from GEO database and single center. Machine learning algorithms were used to identify key genes. GSEA was used for functional analysis. Six algorithms were utilized to establish and evaluate diagnostic models. Key gene-related drugs were screened through network pharmacology. We identified 4 energy metabolism genes, NDUFA1, MECOM, RPL26, and RPS27. These genes have been confirmed to be closely related to multiple energy metabolic pathways and different types of T cell immune infiltration. Additionally, the transcription factors INSM2 and 4 lncRNAs were involved in regulating 4 genes. Further analysis showed that all biomarkers were downregulated in the AD cohorts and not affected by aging and gender. More importantly, we constructed a diagnostic prediction model of 4 biomarkers, which has been validated by various algorithms for its diagnostic performance. Furthermore, we found that valproic acid mainly interacted with these biomarkers through hydrogen bonding, salt bonding, and hydrophobic interaction. We constructed a predictive model based on 4 energy metabolism genes, which may be helpful for the diagnosis of AD. The 4 validated genes could serve as promising blood biomarkers for AD. Their interaction with valproic acid may play a crucial role in the therapy of AD.

Development and validation of a diagnostic prediction model for children with pertussis

To develop and validate a diagnostic prediction model based on blood parameters for predicting the pertussis in children. A retrospective study of 477 children with suspected pertussis at Zigong First People’s Hospital was performed between January 2020 and December 2021. The patients were randomly divided into training cohort and validation cohort. Stepwise regression and R software was performed to develop and validate the model. Stepwise regression analysis showed that white blood cell (WBC), hematocrit (HCT), lymphocyte (LYMPH), C-reactive protein (CRP) and platelet distribution width to mean platelet volume ratio (PDW-MPV-R) were found to be independent factors associated with pertussis. The model containing WBC, CRP and PDW-MPV-R had the best performance. The area under curve (ROC, 0.77 for the training cohort and 0.80 for the validation cohort) of the model indicated satisfactory discriminative ability. The sensitivity and specificity of the model were 72.1% and 72.6% in training cohort and 74% and 72.1%, respectively, in validation cohort. Based on the ROC analysis, calibration plots, and decision curve analysis, we concluded that the model exhibited excellent performance. A model based on blood parameters is sufficiently accurate to predict the probability of pertussis in children, and may provide some reference for clinical decisions.

Predicting biomarkers related to idiopathic pulmonary fibrosis: Robust ranking aggregation analysis and animal experiment verification

Idiopathic pulmonary fibrosis (IPF) is a progressive and incurable lung disease characterized by unknown etiology. This study employs robust ranking aggregation to identify consistent differential genes across multiple datasets, aiming to enhance prognostic evaluation and facilitate the development of more effective immunotherapy strategies for IPF. Using the GSE10667, GSE110147, and GSE24206 datasets, the analysis identifies 92 robust differentially expressed genes (DEGs), including SPP1, IGF1, ASPN, and KLHL13, highlighted as potential biomarkers through machine learning and experimental validation. Additionally, significant differences in immune cell types between IPF samples and controls, such as Plasma cells, Macrophages M0, Mast cells resting, T cells CD8, and NK cells resting, inform the construction of diagnostic and survival prediction models, demonstrating good applicability. These findings provide insights into IPF pathophysiology and suggest potential therapeutic targets.

Diagnostic models to predict structural spinal osteoarthritis on lumbar radiographs in older adults with back pain: Development and internal validation

ObjectiveIt is difficult for health care providers to diagnose structural spinal osteoarthritis (OA), because current guidelines recommend against imaging in patients with back pain. Therefore, the aim of this study was to develop and internally validate multivariable diagnostic prediction models based on a set of clinical and demographic features to be used for the diagnosis of structural spinal OA on lumbar radiographs in older patients with back pain. DesignThree diagnostic prediction models, for structural spinal OA on lumbar radiographs (i.e. multilevel osteophytes, multilevel disc space narrowing (DSN), and both combined), were developed and internally validated in the ‘Back Complaints in Older Adults’ (BACE) cohort (N ​= ​669). Model performance (i.e. overall performance, discrimination and calibration) and clinical utility (i.e. decision curve analysis) were assessed. Internal validation was performed by bootstrapping. ResultsMean age of the cohort was 66.9 years (±7.6 years) and 59% were female. All three models included age, gender, back pain duration and duration of spinal morning stiffness as predictors. The combined model additionally included restricted lateral flexion and spinal morning stiffness severity, and exhibited the best model performance (optimism adjusted c-statistic 0.661; good calibration with intercept −0.030 and slope of 0.886) and acceptable clinical utility. The other models showed suboptimal discrimination, good calibration and acceptable decision curves. ConclusionAll three models for structural spinal OA displayed lesuboptimal discrimination and need improvement. However, these internally validated models have potential to inform primary care clinicians about a patient with risk of having structural spinal OA on lumbar radiographs. External validation before implementation in clinical care is recommended.

Endoplasmic reticulum stress promotes hepatocellular carcinoma by modulating immunity: a study based on artificial neural networks and single-cell sequencing

IntroductionHepatocellular carcinoma (HCC) is characterized by the complex pathogenesis, limited therapeutic methods, and poor prognosis. Endoplasmic reticulum stress (ERS) plays an important role in the development of HCC, therefore, we still need further study of molecular mechanism of HCC and ERS for early diagnosis and promising treatment targets.MethodThe GEO datasets (GSE25097, GSE62232, and GSE65372) were integrated to identify differentially expressed genes related to HCC (ERSRGs). Random Forest (RF) and Support Vector Machine (SVM) machine learning techniques were applied to screen ERSRGs associated with endoplasmic reticulum stress, and an artificial neural network (ANN) diagnostic prediction model was constructed. The ESTIMATE algorithm was utilized to analyze the correlation between ERSRGs and the immune microenvironment. The potential therapeutic agents for ERSRGs were explored using the Drug Signature Database (DSigDB). The immunological landscape of the ERSRGs central gene PPP1R16A was assessed through single-cell sequencing and cell communication, and its biological function was validated using cytological experiments.ResultsAn ANN related to the ERS model was constructed based on SRPX, THBS4, CTH, PPP1R16A, CLGN, and THBS1. The area under the curve (AUC) of the model in the training set was 0.979, and the AUC values in three validation sets were 0.958, 0.936, and 0.970, respectively, indicating high reliability and effectiveness. Spearman correlation analysis suggests that the expression levels of ERSRGs are significantly correlated with immune cell infiltration and immune-related pathways, indicating their potential as important targets for immunotherapy. Mometasone was predicted to be the most promising treatment drug based on its highest binding score. Among the six ERSRGs, PPP1R16A had the highest mutation rate, predominantly copy number mutations, which may be the core gene of the ERSRGs model. Single-cell analysis and cell communication indicated that PPP1R16A is predominantly distributed in liver malignant parenchymal cells and may reshape the tumor microenvironment by enhancing macrophage migration inhibitory factor (MIF)/CD74 + CXCR4 signaling pathways. Functional experiments revealed that after siRNA knockdown, the expression of PPP1R16A was downregulated, which inhibited the proliferation, migration, and invasion capabilities of HCCLM3 and Hep3B cells in vitro.ConclusionThe consensus of various machine learning algorithms and artificial intelligence neural networks has established a novel predictive model for the diagnosis of liver cancer associated with ERS. This study offers a new direction for the diagnosis and treatment of HCC.

Development and validation of a diagnostic nomogram for frailty in cancer patients

BackgroundThe presence of frailty decreases the overall survival of cancer patients. An accurate and operational diagnostic method is needed to help clinicians choose the most appropriate treatment to improve patient outcomes. MethodsData were collected from 10 649 cancer patients who were prospectively enrolled in the Investigation on Nutritional Status and its Clinical Outcomes of Common Cancers (INSCOC) project in China from July 2013 to August 2022. The training cohort and validation cohort were randomly divided at a ratio of 7:3. The multivariable logistic regression analysis, multivariate Cox regression analyses, and the least absolute shrinkage and selection operator (LASSO) method were used to develop the nomogram. The concordance index and calibration curve were used to assess the diagnostic utility of the nomogram model. ResultsThe 10 risk factors associated with frailty in cancer patients were age, AJCC stage, liver cancer, hemoglobin, radiotherapy, surgery, hand grip strength (HGS), calf circumference (CC), PG-SGA score and QOL from the QLQ-C30. The diagnostic nomogram model achieved a good C index of 0.847 (95% CI, 0.832–0.862, P < 0.001) in the training cohort and 0.853 (95% CI, 0.83–0.876, P < 0.001) in the validation cohort. The prediction nomogram showed 1-, 3-, and 5-year mortality C indices in the training cohort of 0.708 (95% CI, 0.686–0.731), 0.655 (95% CI, 0.627–0.683), and 0.623 (95% CI, 0.568–0.678). The 1-, 3-, and 5-year C indices in the validation cohort were similarly 0.743 (95% CI, 0.711–0.777), 0.680 (95% CI, 0.639–0.722), and 0.629 (95% CI, 0.558–0.700). In addition, the calibration curves and decision curve analysis (DCA) were well-fitted for both the diagnostic model and prediction model. ConclusionsThe nomogram model provides an accurate method to diagnose frailty in cancer patients. Using this model could lead to the selection of more appropriate therapy and a better prognosis for cancer patients.

A predictive model for ischemic colitis: Integrating clinical and laboratory parameters

AbstractObjectiveWe aimed to develop a predictive model for the clinical diagnosis of ischemic colitis (IC).MethodsClinical data were collected from patients with acute IC lesions who were diagnosed and admitted to Beijing Tsinghua Changgung Hospital from January 2016 to December 2022. These patients were included in the IC case group in this retrospective observational study. The control group comprised patients aged ≥40 years who were diagnosed with abdominal pain during the same period, excluding those with IC. All patients were divided into a training and test sets based on the time window. Least absolute shrinkage and selection operator regression was used to screen risk factors for the occurrence of IC. Logistic stepwise regression (maximum likelihood ratio method) was performed in multifactorial analysis, and a diagnostic prediction model for IC was established using R language. The area under the receiver operating characteristic (ROC) curve (AUC) was examined to assess differentiation using working ROC curves. We used bootstrap resampling (1000 times) for internal validation. Model calibration curves and decision curve analysis (DCA) were also applied.ResultsOur study indicates that constipation, hematochezia, neutrophil counts, and specific abdominal computed tomography (CT) (plain scan) findings, including intestinal wall edema and thickening, intestinal lumen stenosis, and dilation, are independent predictors of IC. The predictive model exhibited high discriminative ability with an AUC of 0.9788 in the training set, and the calibration and DCA curves demonstrated excellent model performance. After validation, the AUC remained robust at 0.9868, underscoring the model's reliability in predicting IC.ConclusionAccording to our model, constipation accompanied by hematochezia necessitates careful consideration of IC. Abdominal CT (plain scan) is an effective diagnostic tool for IC, and it is common for patients to exhibit elevated neutrophil counts. The predictive model, demonstrating high discriminative ability and accuracy, shows promise for practical application in clinical settings, aiding in the early diagnosis and management of IC.

Identifying Diagnostic Markers and Constructing Predictive Models for Oxidative Stress in Multiple Sclerosis.

Multiple sclerosis (MS) is a chronic disease characterized by inflammation and neurodegeneration of the central nervous system. Despite the significant role of oxidative stress in the pathogenesis of MS, its precise molecular mechanisms remain unclear. This study utilized microarray datasets from the GEO database to analyze differentially expressed oxidative-stress-related genes (DE-OSRGs), identifying 101 DE-OSRGs. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses indicate that these genes are primarily involved in oxidative stress and immune responses. Through protein-protein interaction (PPI) network, LASSO regression, and logistic regression analyses, four genes (MMP9, NFKBIA, NFKB1, and SRC) were identified as being closely related to MS. A diagnostic prediction model based on logistic regression demonstrated good predictive power, as shown by the nomogram curve index and DAC results. An immune-cell infiltration analysis using CIBERSORT revealed significant correlations between these genes and immune cell subpopulations. Abnormal oxidative stress and upregulated expression of key genes were observed in the blood and brain tissues of EAE mice. A molecular docking analysis suggested strong binding potentials between the proteins of these genes and several drug molecules, including isoquercitrin, decitabine, benztropine, and curcumin. In conclusion, this study identifies and validates potential diagnostic biomarkers for MS, establishes an effective prediction model, and provides new insights for the early diagnosis and personalized treatment of MS.

Predictive modeling of lower extreme deep vein thrombosis following radical gastrectomy for gastric cancer: based on multiple machine learning methods

Postoperative venous thromboembolic events (VTEs), such as lower extremity deep vein thrombosis (DVT), are major risk factors for gastric cancer (GC) patients following radical gastrectomy. Accurately predicting and managing these risks is crucial for optimal patient care. This retrospective case‒control study involved 693 GC patients from our hospital who underwent radical gastrectomy. We collected plentiful and comprehensive clinical indicators including a total of 49 baseline, preoperative, surgical and pathological clinical data. Using univariate logistic regression, we identified potential risk factors, followed by feature selection through the Boruta algorithm. We then constructed the final predictive model using multivariate logistic regression and evaluated it using receiver operating characteristic (ROC) curve analysis, calibration plots, decision curve analysis, and other methods. Additionally, we applied various machine learning techniques, including decision trees and random forests, to assess our model’s predictive strength. This retrospective case‒control study involved 693 GC patients from our hospital who underwent radical gastrectomy. We collected plentiful and comprehensive clinical indicators including a total of 49 baseline, preoperative, surgical and pathological clinical data. Using univariate logistic regression, we identified potential risk factors, followed by feature selection through the Boruta algorithm. We then constructed the final predictive model using multivariate logistic regression and evaluated it using receiver operating characteristic (ROC) curve analysis, calibration plots, decision curve analysis, and other methods. Additionally, we applied various machine learning techniques, including decision trees and random forests, to assess our model’s predictive strength. Univariate logistic analysis revealed 14 risk factors associated with postoperative lower limb DVT. Based on the Boruta algorithm, six significant clinical factors were selected, namely, age, D-dimer (D-D) level, low-density lipoprotein, CA125, and calcium and chloride ion levels. A nomogram was developed using the outcomes from the multivariate logistic regression analysis. The predictive model showed high accuracy, with an area under the curve of 0.936 in the training set and 0.875 in the validation set. Various machine learning algorithms confirmed its strong predictive capacity. MR analysis revealed meaningful causal relationships between key clinical factors and DVT risk. Based on various machine learning methods, we developed an effective predictive diagnostic model for postoperative lower extremity DVT in GC patients. This model demonstrated excellent predictive value in both the training and validation sets. This novel model is a valuable tool for clinicians to use in identifying and managing thrombotic risks in this patient population.

ESCCPred: a machine learning model for diagnostic prediction of early esophageal squamous cell carcinoma using autoantibody profiles.

Esophageal squamous cell carcinoma (ESCC) is a deadly cancer with no clinically ideal biomarkers for early diagnosis. The objective of this study was to develop and validate a user-friendly diagnostic tool for early ESCC detection. The study encompassed three phases: discovery, verification, and validation, comprising a total of 1309 individuals. Serum autoantibodies were profiled using the HuProtTM human proteome microarray, and autoantibody levels were measured using the enzyme-linked immunosorbent assay (ELISA). Twelve machine learning algorithms were employed to construct diagnostic models, and evaluated using the area under the receiver operating characteristic curve (AUC). The model application was facilitated through R Shiny, providing a graphical interface. Thirteen autoantibodies targeting TAAs (CAST, FAM131A, GABPA, HDAC1, HDGFL1, HSF1, ISM2, PTMS, RNF219, SMARCE1, SNAP25, SRPK2, and ZPR1) were identified in the discovery phase. Subsequent verification and validation phases identified five TAAbs (anti-CAST, anti-HDAC1, anti-HSF1, anti-PTMS, and anti-ZPR1) that exhibited significant differences between ESCC and control subjects (P < 0.05). The support vector machine (SVM) model demonstrated robust performance, with AUCs of 0.86 (95% CI: 0.82-0.89) in the training set and 0.83 (95% CI: 0.78-0.88) in the test set. For early-stage ESCC, the SVM model achieved AUCs of 0.83 (95% CI: 0.79-0.88) in the training set and 0.83 (95% CI: 0.77-0.90) in the test set. Notably, promising results were observed for high-grade intraepithelial neoplasia, with an AUC of 0.87 (95% CI: 0.77-0.98). The web-based implementation of the early ESCC diagnostic tool is publicly accessible at https://litdong.shinyapps.io/ESCCPred/ . This study provides a promising and easy-to-use diagnostic prediction model for early ESCC detection. It holds promise for improving early detection strategies and has potential implications for public health.

Refining a non-invasive prediction model for neurosyphilis diagnosis by using immunoassay to detect serum anti-TP0435 (TP17) and TP0574 (TP47) IgG antibodies: two-centre cross-sectional retrospective study in China

ObjectivesInvasive lumbar puncture is the conventional method for diagnosing neurosyphilis (NS). We investigated a non-invasive alternative method to detect serum Treponema pallidum-specific antibodies against highly immunogenic antigens TP0171 (TP15), TP0435 (TP17), and TP0574 (TP47) by using luciferase immunosorbent assay. MethodsA total of 816 HIV-negative patients suspected of NS from the Beijing and Guangzhou cohorts were retrospectively selected and tested for serum anti-TP15, TP17, and TP47 IgG antibodies. Two diagnostic prediction models were developed using stepwise logistic regression in the Beijing cohort, and evaluated in the Guangzhou cohort for external validation. ResultsSerum antibodies against TP15, TP17, and TP47 showed moderate capability for NS diagnosis in the Beijing cohort and the corresponding area under the receiver operating characteristic curves (AUCs) were 0.722 [95% confidence interval (CI): 0.680–0.762)], 0.780 (95% CI: 0.741–0.817), and 0.774 (95% CI: 0.734–0.811), respectively. An expanded NS prediction model integrated with anti-TP17 and anti-TP47 antibodies showed better performance than the base NS diagnostic model without anti-TP17 and anti-TP47 antibodies with the AUC of 0.874 (95% CI: 0.841–0.906) vs. 0.845 (95% CI: 0.809–0.881) (p = 0.007) in the development cohort, and 0.934 (95% CI: 0.909–0.960) vs. 0.877 (95% CI: 0.840–0.914) (p < 0.001) in validation cohort, respectively. Decision curve analysis revealed that the net benefit of the expanded model exceeded that of the base model when the threshold probability was between 0.10 and 0.95 in both the development and external validation cohorts. DiscussionSerum antibodies against TP17 and TP47 exhibited promising diagnostic capability for NS and significantly enhanced the predictive accuracy of model for NS diagnosis. Our study highlights the potential of serum treponemal antibody detection as a non-invasive method for NS diagnosis to substitute invasive lumbar puncture in NS diagnosis.

Development of tuberculosis treatment decision algorithms in children below 5 years hospitalised with severe acute malnutrition in Zambia and Uganda: a prospective diagnostic cohort study

In children with severe acute malnutrition (SAM) tuberculosis is common, challenging to diagnose, and often fatal. We developed tuberculosis treatment decision algorithms (TDAs) for children under the age of 5 years with SAM. In this prospective diagnostic study, we enrolled and followed up children aged <60 months hospitalised with SAM at three tertiary hospitals in Zambia and Uganda from 4 November 2019 to 20 June 2022. We included children aged 2-59 months with SAM as defined by WHO and hospitalised following the WHO clinical criteria. We excluded children with current or history of antituberculosis treatment within the preceding 3 months. They underwent tuberculosis symptom screening, clinical assessment, chest X-ray, abdominal ultrasound, Xpert MTB/RIF Ultra (Ultra) and culture on respiratory and stool samples with 6 months follow-up. Tuberculosis was retrospectively defined using the 2015 standard case definition for childhood tuberculosis. We used logistic regression to develop diagnostic prediction models for a one-step diagnosis and a two-step screening and diagnostic approaches. We derived scores from models using WHO-recommended thresholds for sensitivity and proposed TDAs. This study is registered with ClinicalTrials.gov, NCT04240990. Of 1906 children hospitalised with SAM during the study period, 1230 were screened, 1152 were eligible and 603 were enrolled. Of the 603 children enrolled-median age 15 (inter-quartile range (IQR): 11-20) months and 65 (11.0%) living with HIV-114 (18.9%) were diagnosed with tuberculosis, including 51 (8.5%) with microbiological confirmation and 104 (17.2%) initiated treatment at a median of 6(IQR: 2-10) days after inclusion. 108 children were retrospectively classified as having tuberculosis resulting in a prevalence of 17.9% (95% confidence intervals (CI): 15.1; 21.2). 75 (69.4%) children with tuberculosis reported cough of any duration, 32 (29.6%) cough ≥2 weeks and 11 (10.2%) tuberculosis contact history. 535 children had complete data and were included in the diagnostic prediction model. The one-step diagnostic model had 15 predictors, including Ultra, clinical, radiographic, and abdominal features, an area under the receiving operating curve (AUROC) of 0.910, and derived TDA sensitivity of 86.14% (95% CI: 78.07-91.56) and specificity of 80.88% (95% CI: 76.91-84.30). The two-step model had AUROCs of 0.750 and 0.912 for screening and diagnosis, respectively, and derived combined TDA sensitivity of 79.21% (95% CI: 70.30-85.98) and a specificity of 83.64% (95% CI: 79.87-86.82). Tuberculosis prevalence was high among hospitalised children with SAM, with atypical clinical features. TDAs achieved satisfactory diagnostic accuracy and could be used to improve diagnosis in this vulnerable group. Unitaid.

Machine-Learning for Phenotyping and Prognostication of Myocardial Infarction and Injury in Suspected Acute Coronary Syndrome

BackgroundClinical work-up for suspected acute coronary syndrome (ACS) is resource intensive. ObjectivesThis study aimed to develop a machine learning model for digitally phenotyping myocardial injury and infarction and predict 30-day events in suspected ACS patients. MethodsTraining and testing data sets, predominantly derived from electronic health records, included suspected ACS patients presenting to 6 and 26 South Australian hospitals, respectively. All index presentations and 30-day death and myocardial infarction (MI) were adjudicated using the Fourth Universal Definition of MI. We developed 2 diagnostic prediction models which phenotype myocardial injury and infarction according to the Fourth UDMI (chronic myocardial injury vs acute myocardial injury patterns, the latter further differentiated into acute non-ischaemic myocardial injury, Types 1 and 2 MI) using eXtreme Gradient Boosting (XGB) and deep-learning (DL). We also developed an event prediction model for risk prediction of 30-day death or MI using XGB. Analyses were performed in Python 3.6. ResultsThe training and testing data sets had 6,722 and 8,869 participants, respectively. The diagnostic prediction XGB and deep learning models achieved an area under the curve of 99.2% ± 0.1% and 98.8% ± 0.2%, respectively, for differentiating an acute myocardial injury pattern from no injury or chronic myocardial injury pattern and achieved 95.5% ± 0.2% and 94.6% ± 0.9%, respectively, for differentiating type 1 MI from type 2 MI or acute nonischemic myocardial injury. The 30-day death/MI event prediction model achieved an area under the curve of 88.5% ± 0.5%. ConclusionsMachine learning models can digitally phenotype suspected ACS patients at index presentation and predict subsequent events within 30 days. These models require external validation in a randomized clinical trial to evaluate their impact in clinical practice.

Machine learning constructs a diagnostic prediction model for calculous pyonephrosis

In order to provide decision-making support for the auxiliary diagnosis and individualized treatment of calculous pyonephrosis, the study aims to analyze the clinical features of the condition, investigate its risk factors, and develop a prediction model of the condition using machine learning techniques. A retrospective analysis was conducted on the clinical data of 268 patients with calculous renal pelvic effusion who underwent ultrasonography-guided percutaneous renal puncture and drainage in our hospital during January 2018 to December 2022. The patients were included into two groups, one for pyonephrosis and the other for hydronephrosis. At a random ratio of 7:3, the research cohort was split into training and testing data sets. Single factor analysis was utilized to examine the 43 characteristics of the hydronephrosis group and the pyonephrosis group using the T test, Spearman rank correlation test and chi-square test. Disparities in the characteristic distributions between the two groups in the training and test sets were noted. The features were filtered using the minimal absolute value shrinkage and selection operator on the training set of data. Auxiliary diagnostic prediction models were established using the following five machine learning (ML) algorithms: random forest (RF), xtreme gradient boosting (XGBoost), support vector machines (SVM), gradient boosting decision trees (GBDT) and logistic regression (LR). The area under the curve (AUC) was used to compare the performance, and the best model was chosen. The decision curve was used to evaluate the clinical practicability of the models. The models with the greatest AUC in the training dataset were RF (1.000), followed by XGBoost (0.999), GBDT (0.977), and SVM (0.971). The lowest AUC was obtained by LR (0.938). With the greatest AUC in the test dataset going to GBDT (0.967), followed by LR (0.957), XGBoost (0.950), SVM (0.939) and RF (0.924). LR, GBDT and RF models had the highest accuracy were 0.873, followed by SVM, and the lowest was XGBoost. Out of the five models, the LR model had the best sensitivity and specificity is 0.923 and 0.887. The GBDT model had the highest AUC among the five models of calculous pyonephrosis developed using the ML, followed by the LR model. The LR model was considered be the best prediction model when combined with clinical operability. As it comes to diagnosing pyonephrosis, the LR model was more credible and had better prediction accuracy than common analysis approaches. Its nomogram can be used as an additional non-invasive diagnostic technique.

Differentiation between viral and autoimmune limbic encephalitis: a prospective cohort study with development and validation of a diagnostic model.

Distinguishing between viral encephalitis (VE) and autoimmune limbic encephalitis (ALE) presents a clinical challenge due to the overlap in symptoms. We aimed to develop and validate a diagnostic prediction model to differentiate VE and ALE. A prospective observational multicentre cohort study, which continuously enrolled patients diagnosed with either ALE or VE from October 2011 to April 2023. The demographic data, clinical features, and laboratory test results were collected and subjected to logistic regression analyses. The model was displayed as a web-based nomogram and then modified into a scored prediction tool. Model performance was assessed in both derivation and external validation cohorts. A total of 2423 individuals were recruited, and 1001 (496 VE, 505 ALE) patients were included. Based on the derivation cohort (389 VE, 388 ALE), the model was developed with eight variables including age at onset, acuity, fever, headache, nausea/vomiting, psychiatric or memory complaints, status epilepticus, and CSF white blood cell count. The model showed good discrimination and calibration in both derivation (AUC 0.890; 0.868-0.913) and external validation (107 VE, 117 ALE, AUC 0.872; 0.827-0.917) cohorts. The scored prediction tool had a total point that ranged from - 4 to 10 also showing good discrimination and calibration in both derivation (AUC 0.885, 0.863-0.908) and external validation (AUC 0.868, 0.823-0.913) cohorts. The prediction model provides a reliable and user-friendly tool for differentiating between the VE and ALE, which would benefit early diagnosis and appropriate treatment and alleviate economic burdens on both patients and society.

Regression–Based Diagnostic Models for Early Lung Cancer Integrating Conventional Indicators with Tumor Markers

The aim of this research was to develop a lung cancer diagnostic and predictive model that integrates traditional laboratory indicators with tumor markers. This model is intended to facilitate early screening and assist in the process of identifying or detecting lung cancer through a cost-effective, rapid, and convenient approach, ultimately enhancing the early detection rate of lung cancer. A retrospective study was conducted on 66 patients diagnosed with lung cancer and 159 patients with benign pulmonary conditions. Data including general clinical information, conventional laboratory test results, and tumor marker levels were collected. Data analysis was conducted using SPSS 26.0 (Statistical Product and Service Solutions 26.0). The lung cancer diagnosis and prediction model is created using a composite index established through binary logistic regression. The combined diagnostic prediction models, incorporating both traditional indicators and tumor markers, demonstrated a greater area under the curve (AUC) when compared to the diagnostic prediction model based solely on tumor markers and their combination testing. The values of cut-off point, AUC, accuracy, sensitivity, specificity, positive and negative detection rate and accuracy rate are 0.1805, 0.959, 86.67%, 0.955, 0.830, 95.45%, 83.02% and 89.33 respectively and it is shown that the combined diagnostic model display notable efficacy and clinical relevance in aiding the early diagnosis of lung cancer.

Diagnostics Based Patient Classification for Clinical Decision Support Systems

The widespread adoption of Electronic Healthcare Records has resulted in an abundance of healthcare data. This data holds significant potential for improving healthcare services by providing valuable clinical insights and enhancing clinical decision-making. This paper presents a patient classification methodology that utilizes a multiclass and multilabel diagnostic approach to predict the patient's clinical class. The proposed model effectively handles comorbidities while maintaining a high level of accuracy. The implementation leverages the MIMIC III database as a data source to create a phenotyping dataset and train the models. Various machine learning models are employed in this study. Notably, the natural language processing-based One-Vs-Rest classifier achieves the best classification results, maintaining accuracy and F1 scores even with a large number of classes. The patient diagnostic class prediction model, based on the International Classification of Diseases 9, showcased in this paper, has broad applications in diagnostic support, treatment prediction, clinical assistance, recommender systems, clinical decision support systems, and clinical knowledge discovery engines.

Non-invasive diagnosis of nature of pulmonary nodules using the whole genome methylation sequencing of circulating cell-free DNA.

e20038 Background: The frequent false-positive results of lung cancer screening of low-dose computed tomography (LDCT) remains a challenge. In this study, we aimed to establish a novel panel of cancer-specific methylated regions detected from circulating cell-free DNA (cfDNA) for differentiating malignant pulmonary nodules (PNs) from benign lesions in high-risk patients. Methods: We performed DNA methylation profiling by high throughput DNA bisulfite sequencing in tissue samples (nodule size &lt; 3 cm in diameter) to learn methylation patterns that differentiate cancerous tumors from benign lesions. Then we filtered out differential methylation patterns in circulating cfDNA that are consistent with tissue expression patterns and built an assay for plasma sample classification. Area under the curve was computed for each receiving operation curve, and 95% confidence intervals were also estimated by bootstrapping with 1,000 iterations. Results: We first performed the whole genome methylation sequencing of 100 tissue samples from PNs to learn cancer-specific methylation patterns, and obtained 2,406 differential hypermethylated regions and 16,028 differential hypomethylated regions. These tissue-derived DNA methylation markers were further filtered using a training set of 42 plasma samples and 11 markers were ultimately identified to build a diagnostic prediction model. This methylation panel provided promising results in an independent validation set of additional 17 plasma samples, with a sensitivity of 66.67% (95% CI: 0.25–1.00) and a specificity of 79.17% for differentiating patients with malignant tumor (N = 12) from patients with benign lesions (n = 5). A further validation study of this methylation panel in the cancer genome atlas data set demonstrated significant discriminatory performance in distinguishing patients with lung cancer from subjects without malignancy (area under the curve = 0.88). Conclusions: Our model based on the measurement of circulating cfDNA methylation provide a minimally invasive procedure for differentiation of early-stage lung cancer and nonmalignant pulmonary nodules, and may guide management of positive LDCT screening results. Additional clinical studies with larger sample size are needed to establish the robustness of this model.