Abstract：Objective To explore the application value of image-based preliminary artificial intelligence detection in the diagnosis and treatment of neonatal jaundice. Methods The clinical data of 200 neonates with jaundice in Maternity& Child Care Center of Xinyu from October 2018 to December 2019 were retrospectively analyzed. According to the types of jaundice, the neonates were divided into physiological jaundice group (n=102) and pathological jaundice group(n=98). Diag analysis detection system was used to collect relevant data of the two groups of children patients. The neonatal transcutaneous bilirubin (TcB), automated image-based bilirubin (AIB) and new artificial intelligence bilirubin(DiagB) were compared between the two groups. Receiver operating characteristics (ROC) curve was used to describe the diagnostic efficacy of the above quantitative parameters on neonatal pathological jaundice. The accuracy of DiagB in the diagnosis of neonatal jaundice was calculated by using total serum bilirubin (TSB) as the gold standards for the diagnosis of neonatal jaundice. Results The detection values of TcB, AIB and DiagB in pathological jaundice group were higher than those in physiological jaundice group, the differences were statistically significant (P<0.05). The area under ROC curves (AUC) of TcB, AIB and DiagB in the diagnosis of pathological jaundice were 0.774, 0.764 and 0.792, and the cut-off values were 228.59 μmol/L, 213.13 μmol/L and 214.96 μmol/L respectively. The sensitivity,specificity, accuracy rate, positive predictive value, negative predictive value and Kappa value of DiagB in the diagnosis of neonatal pathological jaundice were 74.49% (73/98), 77.45% (79/102), 76.00% (152/200), 76.04% (73/96), 75.96%(79/104) and 0.520. Conclusion Image-based DiagB value detection has high efficacy in diagnosing neonatal pathological jaundice, and it is highly consistent with serum TSB and biochemical test results. Thus it can provide a new safe and non-invasive monitoring method for neonatal pathological jaundice.
胡安辉 ；宗渝敏； 胡小华； 罗勇； 傅思勇. 图片初步人工智能检测在新生儿黄疸诊疗中的应用价值[J]. 中国当代医药, 2021, 28(21): 104-107.
HU An-hui ；ZONG Yu-min ；HU Xiao-hua； LUO Yong； FU Si-yong. Application value of image-based preliminary artificial intelligence detection in the diagnosis and treatment of neonatal jaundice. 中国当代医药, 2021, 28(21): 104-107.
京公网安备 11010502046607号