Browsing by Author "Cubukcu, Hikmet Can"

Now showing 1 - 3 of 3

Detection of COVID-19 by Machine Learning Using Routine Laboratory Tests
(2021) Cubukcu, Hikmet Can; Topcu, Deniz Ilhan; Bayraktar, Nilufer; Gulsen, Murat; Sari, Nuran; Arslan, Ayse Hande; 0000-0002-1219-6368; 0000-0002-7886-3688; 34791032; E-3717-2019; Y-8758-2018
Objectives The present study aimed to develop a clinical decision support tool to assist coronavirus disease 2019 (COVID-19) diagnoses with machine learning (ML) models using routine laboratory test results. Methods We developed ML models using laboratory data (n = 1,391) composed of six clinical chemistry (CC) results, 14 CBC parameter results, and results of a severe acute respiratory syndrome coronavirus 2 real-time reverse transcription-polymerase chain reaction as a gold standard method. Four ML algorithms, including random forest (RF), gradient boosting (XGBoost), support vector machine (SVM), and logistic regression, were used to build eight ML models using CBC and a combination of CC and CBC parameters. Performance evaluation was conducted on the test data set and external validation data set from Brazil. Results The accuracy values of all models ranged from 74% to 91%. The RF model trained from CC and CBC analytes showed the best performance on the present study's data set (accuracy, 85.3%; sensitivity, 79.6%; specificity, 91.2%). The RF model trained from only CBC parameters detected COVID-19 cases with 82.8% accuracy. The best performance on the external validation data set belonged to the SVM model trained from CC and CBC parameters (accuracy, 91.18%; sensitivity, 100%; specificity, 84.21%). Conclusions ML models presented in this study can be used as clinical decision support tools to contribute to physicians' clinical judgment for COVID-19 diagnoses.
Estimation of Low-Density Lipoprotein Cholesterol Concentration Using Machine Learning
(2021) Cubukcu, Hikmet Can; Topcu, Deniz İlhan; 0000-0002-1219-6368; 34635916; E-3717-2019
Objective Low-density lipoprotein cholesterol (LDL-C) can be estimated using the Friedewald and Martin-Hopkins formulas. We developed LDL-C prediction models using multiple machine learning methods and investigated the validity of the new models along with the former formulas. Methods Laboratory data (n = 59,415) on measured LDL-C, high-density lipoprotein cholesterol, triglycerides (TG), and total cholesterol were partitioned into training and test data sets. Linear regression, gradient-boosted trees, and artificial neural network (ANN) models were formed based on the training data. Paired-group comparisons were performed using a t-test and the Wilcoxon signed-rank test. We considered P values .2 to be statistically significant. Results For TG >= 177 mg/dL, the Friedewald formula underestimated and the Martin-Hopkins formula overestimated the LDL-C (P <.001), which was more significant for LDL-C <70 mg/dL. The linear regression, gradient-boosted trees, and ANN models outperformed the aforementioned formulas for TG >= 177 mg/dL and LDL-C <70 mg/dL based on a comparison with a homogeneous assay (P >.001 vs. P <.001) and classification accuracy. Conclusion Linear regression, gradient-boosted trees, and ANN models offer more accurate alternatives to the aforementioned formulas, especially for TG 177 to 399 mg/dL and LDL-C <70 mg/dL.
Optimization Of Patient-Based Real-Time Quality Control Based On The Youden Index
(2022) Topcu, Deniz Ilhan; Cubukcu, Hikmet Can; 35810801
Aim: This study sets out to investigate the utility of exponentially weighted moving average (EWMA) as patient-based real-time quality control (PBRTQC) by conducting a simulation study and subsequent real-patient data implementation to determine optimal EWMA features (weighting factors, control limits, and truncation methods) based on the Youden index. Methods: A simulation experiment was conducted in the first stage to investigate optimal EWMA features for the tests, including aspartate aminotransferase, blood urea nitrogen, and glucose, calcium, creatinine, potassium, sodium, triglycerides, thyroid - stimulating hormone (TSH), and vitamin B12 tests. In the second stage of the study, EWMA was applied to real patient data to elucidate practical utility and achieve final optimal EWMA features. Different degrees of systematic errors (SE) including total allowable error (TEa) as a maximum error level were added to both simulation and patient results, and then the EWMA performance was assessed for different EWMA features. We calculated Youden's index for each combination of EWMA features to find their optimal features to achieve minimum false positive rate (FPR) and maximum error detection rate at the SE level corresponding to TEa. Results: EWMA implementation on real patient data revealed optimal EWMA features for each test. FPR values of creatinine and glucose were 18.48% and 10.17%, respectively, which exceeded the acceptable criteria for FPR (10%). The remaining six analytes showed acceptable FPR. Conclusions: We showed the implementation of EWMA as PBRTQC, and optimization of its features based on the Youden index by conducting extensive performance evaluations and simulations in this study.