‹ Back to the team's academic timeline
PAPER BRIEF · Research digest
Segmenting infant cries — inspiration and expiration across 226 cry segments
Construction and validation of an automated time label segmentation method for infant cries. · BMC pediatrics
An infant's cry is an important window into physiological and psychological states, and distinguishing typical from atypical cries has clinical value early in development. Yet most cry studies do not separate "inspiration" from "expiration" when analysing — precisely because it is hard.
This study developed an audio data-analysis tool that automatically segments inspiration and expiration in infant cries and extracts duration, frequency and intensity per segment. In total 226 cry segments were collected and analysed: sound parameters differed significantly between preterm and full-term infants. The pipeline achieves automatic recognition and digital capture of infant cries, offering an objective way to understand what lies behind them.
Three boundaries must be stated. First, 226 counts cry audio segments, not infants. Second, this is a methods construction-and-validation study; parameter differences are not diagnostic conclusions, and nothing involving infants is phrased as screening or diagnosis. Third, the acknowledgements thank the Chengdu Medical College medical sound database and Chengdu Zhiju Data Technology Co., Ltd. for professional discussion — acknowledgement means the company contributed discussion, not that it developed the method or owns the data. The CV entry marks # after Xu Fan's name; this site keeps the CV record, with symbol meaning and corresponding authorship per the published footnotes.
Fact-checking boundaries
- 226 segments, inspiration/expiration split, three parameter families and the preterm–full-term differences all come from the abstract (full text XML verified locally).
- Acknowledgement sentences are quoted verbatim from the papers' Acknowledgements sections in the evidence ledger.
- The company's stated role is limited to "acknowledged for professional discussion"; the database belongs to Chengdu Medical College.
- The # after author names is a CV annotation; not sufficient to establish corresponding authorship.
AcknowledgementsAcknowledgments: We thank the undergraduate candidates SQC, WZH, LYF, who performed clinical examinations in community hospitals and provided valuable suggestions for this study. Dr Huang Kun provided valuable software segmentation experience for this study. The valuable support from the medical sound database from Chengdu Medical College (http://ama.cmc.edu.cn) and professional discussion from Chengdu Zhiju Data Science Technology Co., Ltd. (www.zjdatast.cn).
Paper abstract (original)
Infant cries serve as critical indicators of an infant's physiological and psychological states, holding essential clinical value in distinguishing between normal and pathological conditions during early development. However, most studies on infant cries failed to segment between inhalation and exhalation during the crying analysis because segmentation remains great challenge. Therefore, we developed an audio data analysis tool capable of accurately segmenting inhalation and exhalation phases in infant cries, followed by extraction of the duration, frequency, and intensity parameters of each segment. Here, 226 sound clips of infants' cries were collected and analyzed. The results revealed the significant differences in sound parameters between premature and term infants. The whole process allows automatic identification and digital collection of infant cries, and it provides an objective method to evaluate the significance behind the cries.