“Timbre” Pilot Study Conducted Using Training & Validation Data Provisioned by UCSF R2D2 for Screening of Pulmonary Tuberculosis Using Cough (Acoustic Sounds), Clinical & Demographic Inputs

Abstract

TimBre from Docturnal offers multidirectional screening of Lung Ailments – Pulmonary Tuberculosis, Pneumonia, Covid19 & COPD. Detailed studies of TimBre in the past used third party Microphone Array that focused on a XY arrangement that provided high fidelity cough sounds with an average length of >5 seconds and real-time demographic data such as Height, Weight, BMI [1]. In the current study, cough sounds were harvested from 7 different countries (India, Vietnam, Philippines, Uganda, Tanzania, Madagascar, SA) using Mobile Phones from different manufacturers & recorded solicited coughs in a clinic for a duration of 0.5 seconds. A plethora of demographic and clinical variables were provided of which a subset was used by TimBre algorithm. Most importantly, the .WAV files were recorded in a single channel at a sampling rate of 44.1kHz & 16 bits. The study details two approaches wherein the first method was to concatenate all the 0.5 second WAV files based on a timestamp provided for each StudyID in the training & scoring set while the second method involved using the 0.5 second snippets as-is in both training and validation sets without any concatenation. The first approach on the TEST set yielded a sensitivity and specificity (table-1) of 68.6% and 71.7% respectively with an AUC of 0.75 while the second approach yielded a sensitivity & specificity (table-2) of 75.41% and 68.30% respectively with an AUC of 0.78 as reported by UCSF R2D2 team. Both the approaches used a combination of Clinical, Demographic and Spectral Variables. Some additional variables included were derived (BMI) & excluded (Spectral) based on the feature importance scores. The ML model performed better in the second approach and we anticipate it to improve further once an additional 714,922 .WAV files harvested as Longitudinal coughs shall be appended to the training set as a part of a subsequent pilot study