Results
Baseline characters of participants
The baseline characters of participants are summarized in Tab 1. Thoroughtout the study(Fig 1.B), 322 infants developed symptoms or signs in the monitoring room. All of them were performed LUS, and clinic data were collected. Twelve of them were excluded according to criteria(4 with complex malformation or congenital lung diseases, 3 were transferred to another hospital; 5 without qualified LUS images). 310 infants enrolled in total, and their LUS images were obtained at one of following timing—-(67, 85, 92, 37, 29 infants; respectively at 0.5h, 1h, 2h, 4h, 6h after birth). After three-day follow, 236 infants were confirmed to be healthy, 74 were admitted to NICU due to pulmonary issues, including RDS(28/74), TTN(31/74), pneumonia(6/74) and MAS(6/74). The GA range was 34w+0d to 41w+5d, and the birth weight range was 1990g to 4520g. Most patients and healthy infants were scanned at the timing of 2 hours after birth. Patients were more likely to have a higher respiratory rate(65(87.8%) vs 57(24.2%), p<0.01) and higher incidence of lower TcSO2(bellow 95% for more than 3 mins, 23(31.1%) vs 18(7.6%), p<0.01).
High-risk patterns and Low-risk patterns
There are six LUS image patterns can be found in these late preterm and term infants right after their birth(Fig 2). These patterns can be categorized as ”High-risk” and ”Low-risk” ones. The low-risk patterns, which means less likelihood of lung issues, include ”pure A-line” pattern, ”Scarce discrete B-line” pattern, ”Moderate discrete B-line” pattern, and ”Abundant discrete B-line”. In fact, the three kinds of ”discrete B-line” can be easily discerned when performing LUS and there is no need to identify each of them, as they are significantly different from ”pure A-line” and high-risk patterns, and they are less likely to be a sign of lung issues. High-risk patterns inlude ”Coalesced B-line” and ”Consolidation”.
The reliability of LUS to predict respiratory support in late preterm and term infants
The ROC analysis for any respiratory support needs using High-risk patterns yielded an AUC of 0.95(95%CI, 0.92-0.98, p<0.001). Correspondingly, the ROC analysis for any respiratory support needs using Low-risk patterns yielded an AUC of 0.89(95%CI, 0.85-0.93, p<0.001). By contrast, as for RR(respiratory rate) which we used conventionally to predict respiratory support need, its ROC only yield an AUC of 0.70(95%CI, 0.64-0.76, p<0.001). The ROC of LUS(high-patterns) and RR have a significant difference(p<0.01)(Fig 3). Table 2 shows reliability data for LUS and RR to predict any need for respiratory support. The ROC analysis for the subgroups also was conducted, and the result of hood oxygen support was shown(Tab 2), but due to an insufficient number of patients who only obtained CPAP, MV, and PS treatment(23/74, 18/74, 29/74, individually), the ROC for they may not reliable and results were not shown.