Congenital anosmia is associated with altered ongoing nasal respiratory airflow
Wed-P2-064
Presented by: Lior Gorodisky
Humans use their nose for two primary purposes: One is to constantly scan the environment for chemical information, and the second is to pass filtered and humidified air to the lungs for ongoing respiration. Humans with isolated congenital anosmia (ICA) use their nose for the second purpose alone. We therefore hypothesized that this difference will manifest in altered patterns of ongoing nasal airflow in ICA.
To address this hypothesis, we used a wearable spirometer to monitor left and right nasal airflow for 24 hours in 21 participants with ICA and 31 normosmic controls. Participants maintained an activity diary throughout the 24 hours. We split the data to wake and sleep, and further parsed the data to 5-minute blocks for which we extracted Respiratory Frequency, Respiratory Magnitude (the height of the respiratory peak), and an estimate of variability in these measures, namely the 5-minute standard deviation.
Using an analysis of variance on the airflow data with conditions of Sense of Smell (ICA/Normosmic) and Level of Arousal (Wake/Sleep) we observed a significant interaction between Sense of Smell and Arousal (F(1,50)=4.72, p=0.035), reflecting significantly less inhalation peaks per minute in ICA during wake (ICA: 19.5±5.8, Normosmics: 23.8±4.5, t(50)=2.8, d’=0.8, p=0.004), but not during sleep (ICA: 13.9±3.5, Normosmics: 15.1±3.5, t(50)=1.22, p=0.23). This decreased inhalation peaks per minute did not influence the number of breaths per minute in ICA compared to Controls (F(1,50)=2.5, p=0.12), but was significant for Arousal (F(1,50)=54.8, p<0.001). These differences, when entered into a SVM classifier, detected ICA at 73% accuracy, with 62% TPR (ICA classified as ICA) and 81% TNR (normosmics classified as normosmics).
In other words, we could identify anosmia by ongoing nasal airflow alone. We predict that using higher-order airflow parameters we will be able to significantly improve the performance of this classifier.
To address this hypothesis, we used a wearable spirometer to monitor left and right nasal airflow for 24 hours in 21 participants with ICA and 31 normosmic controls. Participants maintained an activity diary throughout the 24 hours. We split the data to wake and sleep, and further parsed the data to 5-minute blocks for which we extracted Respiratory Frequency, Respiratory Magnitude (the height of the respiratory peak), and an estimate of variability in these measures, namely the 5-minute standard deviation.
Using an analysis of variance on the airflow data with conditions of Sense of Smell (ICA/Normosmic) and Level of Arousal (Wake/Sleep) we observed a significant interaction between Sense of Smell and Arousal (F(1,50)=4.72, p=0.035), reflecting significantly less inhalation peaks per minute in ICA during wake (ICA: 19.5±5.8, Normosmics: 23.8±4.5, t(50)=2.8, d’=0.8, p=0.004), but not during sleep (ICA: 13.9±3.5, Normosmics: 15.1±3.5, t(50)=1.22, p=0.23). This decreased inhalation peaks per minute did not influence the number of breaths per minute in ICA compared to Controls (F(1,50)=2.5, p=0.12), but was significant for Arousal (F(1,50)=54.8, p<0.001). These differences, when entered into a SVM classifier, detected ICA at 73% accuracy, with 62% TPR (ICA classified as ICA) and 81% TNR (normosmics classified as normosmics).
In other words, we could identify anosmia by ongoing nasal airflow alone. We predict that using higher-order airflow parameters we will be able to significantly improve the performance of this classifier.