From: Clinlabomics: leveraging clinical laboratory data by data mining strategies
Application fields | Year | Sample size | Best models of analysis | Objective and achievement |
---|---|---|---|---|
Clinical prediction | 2019 [64] | 149,000 physical samples | Deep Neural Network (DNN) | Biological aging prediction |
2016 [65] | 62,419 physical samples | Deep Neural Network (DNN) | Biological aging and Smoking status prediction | |
2021 [71] | 285,965 diabetes patients and 1,221,598 healthy human samples | Extreme Gradient Boosting (XGBoost) | Risk prediction for diabetes | |
2017 [72] | 79 paraquat poisoning patients (41 living and 38 deceased) | Support Vector Machine (SVM) | Predicting the prognosis of paraquat poisoning patients | |
2020 [73] | 235 patients (89 benign ovarian tumors and 146 ovarian cancer) | Decision Tree Model | Predicting ovarian cancer | |
2021 [80] | 1823 COVID-19 patients | Extreme Gradient Boosting (XGBoost) | Predicting the mortality of patients with COVID-19 | |
Clinical diagnosis | 2012 [75] | 203 iron deficiency anemia patients | Artificial Neural Network (ANN) | Iron deficiencyanemia diagnosis and iron serum level prediction |
2020 [76] | 355 asthma patients and 1,480 Healthy individuals | Mahalanobis–Taguchi System (MTS) | Asthma diagnosis | |
2019 [77] | 551 chronic kidney disease patients | Logistic Regression Model (LR) | CKD severity diagnosis and surveillance | |
2020 [79] | 177 positive subjects and 102 negative subjects | Random Forest (RF) | COVID-19 infection diagnosis | |
2019 [81] | 15,176 Neurological patients | The Smart Blood Analytics (SBA) Machine Learning (ML) Algorithm | Brain tumors diagnosis | |
2019 [82] | 183 lung cancer patients and 94 patients without lung cancer | Random Forest (RF) | Lung cancer diagnosis | |
2021 [83] | 1168 colorectal cancer patients and 1269 healthy subjects | Logistic Regression Model (LR) | Colorectal cancer diagnosis | |
Clinical labortory management | 2018 [85] | 10,799 training samples and 9839 testing samples | Support Vector Machine (SVM) | Identifying wrong blood in tube errors prior to test reporting |
2021 [86] | 141,396 samples | Artificial Neural Network (ANN) | Identifying mislabeled samples | |
2021 [90] | 192 clotted samples and 2889 normal blood samples | Back Propagation Neural Network (BPNN) | Identifying clotted specimens in coagulation testing | |
2018 [91] | 4619 samples of urine steroid profiles | Tree-based Model | Aiding the Interpretation of urine steroid profiles | |
2022 [92] | 202 consecutive chronic lymphocytic leukemia patients | Deep Neural Network (DNN) | Improving flow cytometry workflow efficiency for detecting of minimal residual disease of chronic lymphocytic leukemia | |
2022 [95] | 254 healthy samples, 8800 physical examination population and 7700 outpatient samples | Normally distributed data: Transformed Hoffmann, Transformed Bhattacahrya, Kosmic and RefineR Algorithms Data with obvious skewness: Expectation Maximization (EM) Algorithm combined with Box-Cox Transformation | Establishing reference intervals for thyroid-related hormones in older adults | |
2019 [99] | 212,554 urine samples | Extreme Gradient Boosting (XGBoost) | Screening urine microbiological inoculation samples |