China
Africa
Explore chapters and articles related to this topic
Classification of Breast Thermograms using a Multi-layer Perceptron with Back Propagation Learning
Published in K. Gayathri Devi, Kishore Balasubramanian, Le Anh Ngoc, Machine Learning and Deep Learning Techniques for Medical Science, 2022
Comparison of the performance of a classifier with other classifiers to categorize thermograms is done using various evaluation metrics. Accuracy, sensitivity, specificity, PPV and NPV are calculated using the data in the confusion matrix, which uses the values of - True-Positive (TP) True-Negative (TN), False Positive (FP) and False Negative (FN). True-Positive: malignancy is correctly classified; True-Negative: benign cases correctly identified as benign; False-Positive: benign cases incorrectly classified as malignant; False-Negative: malignancy incorrectly identified as benign. The confusion matrix is a 2 × 2 matrix that helps to evaluate the performance of supervised classification algorithms. The diagonal grid values in the matrix show the number of cases that are correctly classified and the off-diagonal values show the falsely classified cases. Accuracy gives the percentage of correct classification. However, it is not enough alone to reveal how well the model predicted ‘benign’ and ‘malignant’ cases independently. Sensitivity is the ability of a classifier to detect malignancy while specificity is the ability to detect benign cases. PPV reflects the malignant possibility of positive result while NPV reflects the benign possibility of a negative result as shown in Figure 3.8.
Critical appraisal of studies on diagnostic tests
Published in O. Ajetunmobi, Making Sense of Critical Appraisal, 2021
The negative predictive value (NPV) of a test is defined as the: Proportion of negative test results for which disease is absent, orProbability that the patient is a true negative (without disease), having tested negative, orThe chance of no disease given a negative result.
Critical Appraisal Skills
Published in John C Watkinson, Raymond W Clarke, Louise Jayne Clark, Adam J Donne, R James A England, Hisham M Mehanna, Gerald William McGarry, Sean Carrie, Basic Sciences Endocrine Surgery Rhinology, 2018
Paul Nankivell, Christopher Coulson
In an ideal world, a positive test would mean that someone has disease, and a negative test would mean they do not have disease. Unfortunately, this is rarely the case. When a test is carried out, there are four possible outcomes: The test can correctly detect disease that is present (a true positive result).The test can detect disease when it is really absent (a false-positive result).The test can correctly identify that a disease is not present (a true negative result).The test identifies someone as free of a disease when it is really present (a false-negative result). These possible outcomes are illustrated in Table 44.1.
Evaluation of health economic impact of initial diagnostic modality selection for colorectal cancer liver metastases in suspected patients in China, Japan and the USA
Published in Journal of Medical Economics, 2023
Michael Blankenburg, Mostafa Elhamamy, Diana Zhang, Naoto Fujikawa, Alice Corbin, Guanyi Jin, James Harris, Gesine Knobloch
After initial imaging, patients received either a positive (including true positive and false positive), negative (including true negative and false negative) or indeterminate diagnosis. Patients with a positive CRCLM diagnosis underwent treatment and these patients were designated as having resectable or unresectable disease with patients designated as resectable undergoing surgical resection (including all false-positive patients) and patients designated as unresectable undergoing one out of a range of possible treatment options. The distribution of patients to surgery and non-surgical treatment options were derived from expert interviews in each of the three countries. Patients classified as indeterminate proceeded for further imaging or biopsy and patients with a negative CRCLM diagnosis did not receive treatment and were not associated with any further steps.
Evaluation of the health economic impact of initial diagnostic modality selection in patients suspected of having HCC in China and the USA
Published in Journal of Medical Economics, 2022
Michael Blankenburg, Mostafa Elhamamy, Diana Zhang, Alice Corbin, Guanyi Jin, James Harris, Gesine Knobloch
The decision tree structures comprised 10 steps for both the US (Figure 1) and China (Figure 2). Patients were initially imaged with MDCT, ECCM-MRI, EOB-MRI, or CEUS, and biopsy was considered from the second round of diagnosis. After initial imaging, patients were either given a positive (including true positive and false positive), negative (including true negative and false negative), or indeterminate diagnosis. Patients with a positive HCC diagnosis underwent treatment according to Barcelona Clinic Liver Cancer (BCLC) stage in the US, and according to tumor number and size in China. For the US model inputs, the distribution of patients between stages and the distribution of treatment options for a specific disease stage was derived from expert interviews. For China model inputs, these distributions were derived from a previous study10. Patients classified as indeterminate continued for further imaging or biopsy, and patients with a negative HCC diagnosis did not receive treatment and were not associated with any further steps.
Commentary: statistical analysis of 2 × 2 tables in biomarker studies 2) study design and statistical tests
Published in Biomarkers, 2022
Screening and diagnostic testing are similar procedures, but their difference depends on the context. A screening test is public health/population centred whereas a diagnostic test is medically/individual centred. A highly sensitive test should be used as a screening test to detect potential positives (and to avoid missing a true positive at the expense, potentially, of false positives) (SnOUT). Screening tests almost always require a second, diagnostic test to confirm the condition. Usually, only positives from a screening test will be tested in this, so a diagnostic test should ideally have high specificity with a low false positive rate so that a negative result means a true negative (SpIN). It should also have a high sensitivity to minimize false negatives if negatives from the first test are tested. When specificity (TN/(FP + TN)) tends to 1, FP tends to 0; similarly, as sensitivity (TP/(TP + FN)) tends to 1, FN tends to 0. A negative result on a highly sensitive test is thus almost certainly a true negative (SnOUT). A positive result on a highly specific test is almost certainly a true positive (SpIN). Note that in a sequential approach, the prevalence of positive cases increases from the screening to the diagnostic tests: the prior probability of being positive increasing in the diagnostic test. It is important, therefore, that the sensitivity is very high in the screening test and the specificity is very high in the diagnostic test and that the respective specificities and sensitivities in these tests are also relatively high to ensure correct results.