Identifying individuals with recent COVID-19 through voice classification using deep learning.

Recently deep learning has attained a breakthrough in model accuracy for the classification of images due mainly to convolutional neural networks. In the present study, we attempted to investigate the presence of subclinical voice feature alteration in COVID-19 patients after the recent resolution of disease using deep learning. The study was a prospective study of 76 post COVID-19 patients and 40 healthy individuals. The diagnoses of post COVID-19 patients were based on more than the eighth week after onset of symptoms. Voice samples of an 'ah' sound, coughing sound and a polysyllabic sentence were collected and preprocessed to log-mel spectrogram. Transfer learning using the VGG19 pre-trained convolutional neural network was performed with all voice samples. The performance of the model using the polysyllabic sentence yielded the highest classification performance of all models. The coughing sound produced the lowest classification performance while the ability of the monosyllabic 'ah' sound to predict the recent COVID-19 fell between the other two vocalizations. The model using the polysyllabic sentence achieved 85% accuracy, 89% sensitivity, and 77% specificity. In conclusion, deep learning is able to detect the subtle change in voice features of COVID-19 patients after recent resolution of the disease.

Comparison between blood hemoglobin concentration determined by point-of-care device and complete blood count in adult patients with dengue.

BACKGROUND: Hematocrit measurement has been an indispensable tool for monitoring plasma leakage and bleeding in dengue patients. However, hematocrit measurement by automated methods is hampered by frequent venipunctures. Utility of point-of-care hemoglobin (POC-Hb) test for monitoring dengue patients has not been established. We evaluated the relationship between hemoglobin measured by POC-Hb testing and hematocrit measured by the automated method in adult dengue patients. METHODOLOGY AND PRINCIPAL FINDINGS: Adult dengue patients were recruited at two university hospitals in Thailand from October 2019 to December 2020. POC-Hb test was performed using capillary blood simultaneously with venipuncture to obtain whole blood for an automated complete blood count (CBC) analysis. The correlation of hemoglobin and hematocrit measurement was evaluated. A total of 44 dengue patients were enrolled. Twenty-nine patients (65.9%) were female, with a median age of 31 years (interquartile range 22-41). Of the enrolled patients, 30 (68.2%), 11 (25.0%), and 3 (6.8%) were classified as dengue without warning signs, with warning signs, and severe dengue, respectively. Seven patients (15.9%) had hemoconcentration, and five patients (11.3%) had bleeding. A total of 216 pairs of POC-Hb and CBC were evaluated. A significant positive correlation was observed between hemoglobin measured by POC-Hb testing and hematocrit measured by an automated CBC (r = 0.869, p <0.001). Bland-Altman analysis between hemoglobin measured by POC-Hb testing and an automated CBC showed a bias of -0.43 (95% limit of agreement of -1.81 and 0.95). Using the cutoff of POC-Hb ≥20% as a criteria for hemoconcentration, the sensitivity and specificity of hemoconcentration detected by POC-Hb device were 71.4% and 100.0%, respectively. CONCLUSIONS: Hemoglobin measurement by POC-Hb testing has a strong correlation with hematocrit in adult patients with dengue fever. However, the sensitivity in detecting hemoconcentration is fair. The adjunct use of capillary POC-Hb testing can decrease the frequency of venipuncture. Further study in children is encouraged.