Ahmed, Salahuddin. (2021). Community use of digital auscultation to improve diagnosis of paediatric pneumonia in Sylhet, Bangladesh, [dataset]. University of Edinburgh. Usher Institute. NIHR Global Health Research Unit on Respiratory Health (RESPIRE). https://doi.org/10.7488/ds/3148.
We are collecting data on general physical health, clinical signs, oxygen saturation and lung sound record from children. We are also collecting household socio-economic data of enrolled children. We are collecting those data using tablets. A trained and standardised paediatric listening panel are interpreting lung sounds in five categories – Only wheeze, only crackles, both wheeze and crackles, no wheeze and no crackles, and uninterpretable. A machine learning artificial intelligence technique will classify the lung sound in the same categories. The study will also generate audio recording of four focus group discussions (FGDs) and their transcripts in the native language (Bengali), later transcripts will be translated in English and archive as word document.
Below datafiles will be created from this study:
a) Two monthly children surveillance data file: CHWs are visiting each child in the study area in every two months and collecting child history of respiratory illness symptoms as well as examine the child’s respiratory system and recording signs including respiratory rate
b) Screening data file: Community Health Care Providers (CHCP) are screening all under-5 children who visit community clinic and record the finding which include history of cough or difficult breathing
c) Enrolment data file: CHCP examine all enrolled children and record findings in this data file which include respiratory symptoms and signs, temperature, oxygen saturation, weight, height, mid arm circumference, record lung sounds using the Smartscope
d) Confounding factors data file: CHWs are collecting the socioeconomic status and other confounding data from each enrolled child’s parents or carer
e) Lung sound interpretation by human data file: A trained paediatric listening panel is form. All recorded lung sound files are interpreting by two members and if discordant, then, Dr Eric McCollum interpret the sound file and which is the final interpretation. Those interpretation will create another data file
f) Lung sound interpretation by machine data file: All sound files will be interpreted by a machine learning algorithm and create another data file
g) FGD data files: Bangla and English transcript in word/pdf files
The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336, VAT Registration Number GB 592 9507 00, and is acknowledged by the UK authorities as a “Recognised body” which has been granted degree awarding powers.