Development and Technical Validation of a Smartphone-Based Cry Detection Algorithm.

Ahnjili Zhuparris,Max Van Gent,Eva Dessing,Vasileios Exadaktylos,Gertjan A Driessen,Adam F Cohen,Matthijs D Kruizinga,Robert Jan Doll,Frederik E Stuurman

doi:10.3389/fped.2021.651356

Abstract

Introduction: The duration and frequency of crying of an infant can be indicative of its health. Manual tracking and labeling of crying is laborious, subjective, and sometimes inaccurate. The aim of this study was to develop and technically validate a smartphone-based algorithm able to automatically detect crying.Methods: For the development of the algorithm a training dataset containing 897 5-s clips of crying infants and 1,263 clips of non-crying infants and common domestic sounds was assembled from various online sources. OpenSMILE software was used to extract 1,591 audio features per audio clip. A random forest classifying algorithm was fitted to identify crying from non-crying in each audio clip. For the validation of the algorithm, an independent dataset consisting of real-life recordings of 15 infants was used. A 29-min audio clip was analyzed repeatedly and under differing circumstances to determine the intra- and inter- device repeatability and robustness of the algorithm.Results: The algorithm obtained an accuracy of 94% in the training dataset and 99% in the validation dataset. The sensitivity in the validation dataset was 83%, with a specificity of 99% and a positive- and negative predictive value of 75 and 100%, respectively. Reliability of the algorithm appeared to be robust within- and across devices, and the performance was robust to distance from the sound source and barriers between the sound source and the microphone.Conclusion: The algorithm was accurate in detecting cry duration and was robust to various changes in ambient settings.

Highlights

The duration and frequency of crying of an infant can be indicative of its health
The training set consisted of 897 5-s audio clips, as well as 1,263 non-crying 5-s clips
The median cry duration of the infants recorded at home was shorter (1.4 min, IQR 0.58–2.6) compared to children recorded during their admission to the hospital (5.8 min, IQR 2.2–16.7)

Summary

Introduction

The duration and frequency of crying of an infant can be indicative of its health. Manual tracking and labeling of crying is laborious, subjective, and sometimes inaccurate. Smartphone-Based Cry Detection to record cry behavior, such as parent- or nurse- reported cry duration, are subjective and vulnerable to observer bias [5]. More objective manual annotating of audio recordings is labor intensive and may be subject to privacy-concerns by parents. An objective, automated and unobtrusive method to quantify crying behavior in an at-home and clinical setting may improve the diagnostic process in excessively crying infants, allow for objective determination of treatment effects by physicians, and enable researchers to include objectively determined cry duration as digital biomarker in clinical trials. Given the importance for researchers to study the relationship between an infant’s crying patterns and their health, automatic detection and quantification of infant cries from an audio signal is an essential step in remote baby monitoring applications [6]

Objectives

Methods

Results

Conclusion