Creating a lung ultrasound information score to measure the information content of pediatric point-of-care lung ultrasound reports.

Designing point-of-care lung ultrasound reporting interfaces requires methods for comparison of the information content of the reports generated. Objective We developed an information score that quantifies the detail in the sonographer's report. This is distinct from ultrasound severity scores, which reflect disease severity. Methods Our subject matter experts created a criteria-based, regex-assignable ultrasound report information score. This score was designed to measure the quality of the ultrasound report, with weighting for important clinical findings while retaining its distinctness from ultrasound severity scores. We then tested this score using a publicly available dataset of point-of-care lung ultrasounds performed in a single pediatric emergency department between 30/November/2019 and 14/August/2021. We iteratively refined the initial code we had written with Claude to create a Stata command packaging a set of regex commands that assigned the information score as intended. We compared its results with ultrasound severity and measured its relation with Guiraud's R, a measure of lexical diversity. Results We were able to score 415 of the 427 ultrasounds by code; of these, 220/415 (53%) were scored as zero (normal or almost normal). Patients ranged in age from 14 days to 21 years. The score behaved as intended when tested using the dataset. It t retained its distinctiveness from the ultrasound severity score while not being wholly divorced from it. The lung ultrasound information score also correlated modestly with lexical diversity while not being unduly affected by report length. Conclusion Our information score is an objective, regex-assignable method for comparing information content in point-of-care lung ultrasound reports.