Avci, Veli; Huyut, Mehmet Tahir; Velichko, Andrei; Belyaev, Maksim

Introduction: Acute appendicitis is the commonest cause of surgical abdominal pain, yet diagnosis in children remains challenging; delays increase the risk of perforation, peritonitis and sepsis. We sought to develop a rapid, inexpensive and interpretable clinical-decision support system (CDSS) that leverages routine blood tests (RBT) to assist early paediatric triage. Materials and Methods: In this retrospective single-centre study (January 2020–December 2024) we analysed 275 emergency-department encounters for abdominal pain (75 histology-confirmed appendicitis, 200 controls). The six-stage pipeline comprised (1) cohort selection; (2) exploratory logistic-regression screening of RBT variables; (3) training of Random Forest, Gradient Boosting and LightGBM ensembles (with/without SMOTE) under 10 × 10 stratified cross-validation; (4) SHAP-based feature interpretation; (5) exhaustive generation of every two- and three-parameter arithmetic biomarker from seven RBT features; and (6) derivation of probability-threshold curves and a three-zone rule tree for the top biomarker. Performance was reported with accuracy (ACC), Matthews correlation coefficient (MCC), AUC-ROC, sensitivity, specificity, F1-Score PPV and NPV. Results: Logistic regression and SHAP confirmed CRP, WBC and neutrophil count as strong positive predictors, whereas MPV and PDW were protective; PLT remained non-informative. All three ensemble classifiers surpassed 97% accuracy, 98% AUC-ROC and 0.93 MCC, with no gain from SMOTE. An extensive formula search, the best two-parameter marker was Neutrophil ÷ PDW (MCC = 0.73, specificity 95%). Its ensemble curve crosses P = 0.5 five times; practical cut-offs of < 0.633 (strongly indicate healthy) and > 0.794 (strongly indicate appendicitis) retain high NPV (∼91%) and PPV (∼86%). Among triple formulas that do not rely on PLT, the leading biomarker was CRP+WBC+Neutrophil (MCC = 0.85, PPV 92%, NPV 95%). The ensemble curve intersects at P = 0.5 at three points; values >27 strongly predict appendicitis, <23 indicates a healthy state, and values 23–27 leave a small uncertain band. A rule-based CDSS built on these two biomarkers correctly classified all controls (specificity 100%), sensitivity 95%, achieved 91% overall accuracy, and offers interpretable, electronic health records (EHRs)-ready cut-offs for paediatric appendicitis triage. Conclusion: Routine haematology-biochemistry data, interpreted through ensemble learning and engineered biomarkers, can deliver fast, transparent and highly accurate support for paediatric appendicitis triage. Given its zero false-positive rate, the proposed CDSS is best suited to in-hospital monitoring, where minimising false negatives is critical. Prospective multi-centre validation is warranted.

Huyut, Mehmet Tahir

The author regrets that the following ethical approval number is missing from the article: Ref. No. 03-15/2021.

Huyut, M.T.

Objectives: When the prognosis of COVID-19 disease can be detected early, the intense-pressure and loss of workforce in health-services can be partially reduced. The primary-purpose of this article is to determine the feature-dataset consisting of the routine-blood-values (RBV) and demographic-data that affect the prognosis of COVID-19. Second, by applying the feature-dataset to the supervised machine-learning (ML) models, it is to identify severely and mildly infected COVID-19 patients at the time of admission. Material and methods: The sample of this study consists of severely (n = 192) and mildly (n = 4010) infected-patients hospitalized with the diagnosis of COVID-19 between March-September, 2021. The RBV-data measured at the time of admission and age-gender characteristics of these patients were analyzed retrospectively. For the selection of the features, the minimum-redundancy-maximum-relevance (MRMR) method, principal-components-analysis and forward-multiple-logistics-regression analyzes were used. The features set were statistically compared between mild and severe infected-patients. Then, the performances of various supervised-ML-models were compared in identifying severely and mildly infected-patients using the feature set. Results: In this study, 28 RBV-parameters and age-variable were found as the feature-dataset. The effect of features on the prognosis of the disease has been clinically proven. The ML-models with the highest overall-accuracy in identifying patient-groups were found respectively, as follows: local-weighted-learning (LWL)-97.86%, K-star (K*)-96.31%, Naive-Bayes (NB)-95.36% and k-nearest-neighbor (KNN)-94.05%. Also, the most successful models with the highest area-under-the-receiver-operating-characteristic-curve (AUC) values in identifying patient groups were found respectively, as follows: LWL-0.95%, K*-0.91%, NB-0.85% and KNN-0.75%. Conclusion: The findings in this article have significant a motivation for the healthcare professionals to detect at admission severely and mildly infected COVID-19 patients.