1.Computerized geometric features of carpal bone for bone age estimation.
Chi-Wen HSIEH ; Tai-Lang JONG ; Yi-Hong CHOU ; Chui-Mei TIU
Chinese Medical Journal 2007;120(9):767-770
BACKGROUNDBone age development is one of the significant indicators depicting the growth status of children. However, bone age assessment is an heuristic and tedious work for pediatricians. We developed a computerized bone age estimation system based on the analysis of geometric features of carpal bones.
METHODSThe geometric features of carpals were extracted and analyzed to judge the bone age of children by computerized shape and area description. Four classifiers, linear, nearest neighbor, back-propagation neural network, and radial basis function neural network, were adopted to categorize bone age. Principal component and discriminate analyses were employed to improve assorting accuracy.
RESULTSThe hand X-ray films of 465 boys and 444 girls served as our database. The features were extracted from carpal bone images, including shape, area, and sequence. The proposed normalization area ratio method was effective in bone age classification by simulation. Besides, features statistics showed similar results between the standard of the Greulich and Pyle atlas and our database.
CONCLUSIONSThe bone area has a higher discriminating power to judge bone age. The ossification sequence of trapezium and trapezoid bones between Taiwanese and the atlas of the GP method is quite different. These results also indicate that carpal bone assessment with classification of neural networks can be correct and practical.
Age Determination by Skeleton ; Carpal Bones ; anatomy & histology ; Child ; Child, Preschool ; Female ; Humans ; Infant ; Male ; Neural Networks (Computer)
2.Artificial intelligence predicts direct-acting antivirals failure among hepatitis C virus patients: A nationwide hepatitis C virus registry program
Ming-Ying LU ; Chung-Feng HUANG ; Chao-Hung HUNG ; Chi‐Ming TAI ; Lein-Ray MO ; Hsing-Tao KUO ; Kuo-Chih TSENG ; Ching-Chu LO ; Ming-Jong BAIR ; Szu-Jen WANG ; Jee-Fu HUANG ; Ming-Lun YEH ; Chun-Ting CHEN ; Ming-Chang TSAI ; Chien-Wei HUANG ; Pei-Lun LEE ; Tzeng-Hue YANG ; Yi-Hsiang HUANG ; Lee-Won CHONG ; Chien-Lin CHEN ; Chi-Chieh YANG ; Sheng‐Shun YANG ; Pin-Nan CHENG ; Tsai-Yuan HSIEH ; Jui-Ting HU ; Wen-Chih WU ; Chien-Yu CHENG ; Guei-Ying CHEN ; Guo-Xiong ZHOU ; Wei-Lun TSAI ; Chien-Neng KAO ; Chih-Lang LIN ; Chia-Chi WANG ; Ta-Ya LIN ; Chih‐Lin LIN ; Wei-Wen SU ; Tzong-Hsi LEE ; Te-Sheng CHANG ; Chun-Jen LIU ; Chia-Yen DAI ; Jia-Horng KAO ; Han-Chieh LIN ; Wan-Long CHUANG ; Cheng-Yuan PENG ; Chun-Wei- TSAI ; Chi-Yi CHEN ; Ming-Lung YU ;
Clinical and Molecular Hepatology 2024;30(1):64-79
Background/Aims:
Despite the high efficacy of direct-acting antivirals (DAAs), approximately 1–3% of hepatitis C virus (HCV) patients fail to achieve a sustained virological response. We conducted a nationwide study to investigate risk factors associated with DAA treatment failure. Machine-learning algorithms have been applied to discriminate subjects who may fail to respond to DAA therapy.
Methods:
We analyzed the Taiwan HCV Registry Program database to explore predictors of DAA failure in HCV patients. Fifty-five host and virological features were assessed using multivariate logistic regression, decision tree, random forest, eXtreme Gradient Boosting (XGBoost), and artificial neural network. The primary outcome was undetectable HCV RNA at 12 weeks after the end of treatment.
Results:
The training (n=23,955) and validation (n=10,346) datasets had similar baseline demographics, with an overall DAA failure rate of 1.6% (n=538). Multivariate logistic regression analysis revealed that liver cirrhosis, hepatocellular carcinoma, poor DAA adherence, and higher hemoglobin A1c were significantly associated with virological failure. XGBoost outperformed the other algorithms and logistic regression models, with an area under the receiver operating characteristic curve of 1.000 in the training dataset and 0.803 in the validation dataset. The top five predictors of treatment failure were HCV RNA, body mass index, α-fetoprotein, platelets, and FIB-4 index. The accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of the XGBoost model (cutoff value=0.5) were 99.5%, 69.7%, 99.9%, 97.4%, and 99.5%, respectively, for the entire dataset.
Conclusions
Machine learning algorithms effectively provide risk stratification for DAA failure and additional information on the factors associated with DAA failure.