Pure tone audiogram classification using deep learning techniques

OBJECTIVE

Pure tone audiometry has played a critical role in audiology as the initial diagnostic tool, offering vital insights for subsequent analyses. This study aims to develop a robust deep learning framework capable of accurately classifying audiograms across various commonly encountered tasks.

DESIGN, SETTING, AND PARTICIPANTS

This single-centre retrospective study was conducted in accordance with the STROBE guidelines. A total of 12 518 audiograms were collected from 6259 patients aged between 4 and 96 years, who underwent pure tone audiometry testing between February 2018 and April 2022 at Tongji Hospital, Tongji Medical College, Wuhan, China. Three experienced audiologists independently annotated the audiograms, labelling the hearing loss in degrees, types and configurations of each audiogram.

MAIN OUTCOME MEASURES

A deep learning framework was developed and utilised to classify audiograms across three tasks: determining the degrees of hearing loss, identifying the types of hearing loss, and categorising the configurations of audiograms. The classification performance was evaluated using four commonly used metrics: accuracy, precision, recall and F1-score.

RESULTS

The deep learning method consistently outperformed alternative methods, including K-Nearest Neighbors, ExtraTrees, Random Forest, XGBoost, LightGBM, CatBoost and FastAI Net, across all three tasks. It achieved the highest accuracy rates, ranging from 96.75% to 99.85%. Precision values fell within the range of 88.93% to 98.41%, while recall values spanned from 89.25% to 98.38%. The F1-score also exhibited strong performance, ranging from 88.99% to 98.39%.

CONCLUSIONS

This study demonstrated that a deep learning approach could accurately classify audiograms into their respective categories and could contribute to assisting doctors, particularly those lacking audiology expertise or experience, in better interpreting pure tone audiograms, enhancing diagnostic accuracy in primary care settings, and reducing the misdiagnosis rate of hearing conditions. In scenarios involving large-scale audiological data, the automated classification system could be used as a research tool to efficiently provide a comprehensive overview and statistical analysis. In the era of mobile audiometry, our deep learning framework can also help patients quickly and reliably understand their self-tested audiograms, potentially encouraging timely consultations with audiologists for further evaluation and intervention.

mGluR1/IP3/ERK signaling pathway regulates vestibular compensation in ON UBCs of the cerebellar flocculus

AIMS

To investigate the role of mGluR1α in cerebellar unipolar brush cells (UBC) in mediating vestibular compensation (VC), using mGluR1α agonist and antagonist to modulate ON UBC neurons, and explore the mGluR1/IP3/extracellular signal-regulated kinase (ERK) signaling pathway.

METHODS

First, AAV virus that knockdown ON UBC (mGluR1α) were injected into cerebellar UBC by stereotactic, and verified by immunofluorescence and western blot. The effect on VC was evaluated after unilateral labyrinthectomy (UL). Second, saline, (RS)-3,5-dihydroxyphenylglycine (DHPG), and LY367385 were injected into tubes implanted in rats at different time points after UL separately. The effect on ON UBC neuron activity was evaluated by immunofluorescence. Then, Phosphoinositide (PI) and p-ERK1/2 levels of mGluR1α were analyzed by ELISA after UL. The protein levels of p-ERK and total ERK were verified by western blot. In addition, the effect of mGluR1α activation or inhibition on VC-related behavior was observed.

RESULTS

mGluR1α knockdown induced VC phenotypes. DHPG increased ON UBC activity, while LY367385 reduced ON UBC activity. DHPG group showed an increase in PI and p-ERK1/2 levels, while LY367385 group showed a decrease in PI and p-ERK1/2 levels in cerebellar UBC of rats. The western blot results of p-ERK and total ERK confirm and support the observations. DHPG alleviated VC-related behavior phenotypes, while LY367385 exacerbated vestibular decompensation-like behavior induced by UL.

CONCLUSION

mGluR1α activity in cerebellar ON UBC is crucial for mediating VC through the mGluR1/IP3/ERK signaling pathway, which affects ON UBC neuron activity and contributes to the pathogenesis of VC.

The Effects of Unilateral Labyrinthectomy on Monoamine Neurotransmitters in the Medial Vestibular Nucleus of Rats

BACKGROUND

This study aimed to investigate the effects of unilateral labyrinthectomy (UL) on monoamine neurotransmitters in the medial vestibular nucleus (MVN) of rats.

METHODS

Adult Sprague-Dawley rats were utilized for the vestibular impaired animal model through UL. The success of the model establishment and the recovery process were evaluated using vestibular behavioral tests, including spontaneous nystagmus, postural asymmetry, and balance beam test. Additionally, the expression levels of c-Fos protein in the MVN were assessed by immunofluorescence. Furthermore, changes in the expression levels of monoamine neurotransmitters, including 5-hydroxytryptamine (5-HT), norepinephrine (NE), dopamine (DA), and histamine in the MVN, were analyzed using high-performance liquid chromatography (HPLC) at different time points after UL (4 h, 8 h, 1 day, 2 days, 4 days, and 7 days).

RESULTS

Compared to the sham control group, the UL group exhibited the most pronounced vestibular impairment symptoms at 4 h post-UL, which significantly decreased at 4 days and almost fully recovered by 7 days. Immunofluorescence results showed a notable upregulation of c-Fos expression in the MVN subsequent to the UL-4 h, serving as a reliable indicator of heightened neuronal activity. In comparison with the sham group, HPLC analysis showed that the levels of 5-HT and NE in the ipsilesional MVN of the UL group were significantly elevated within 4 days after UL, and peaked on 1 day and 2 days, respectively. DA showed an increasing trend at different time points up to 7 days post-UL, while histamine levels significantly increased only at 1 day post-UL.

CONCLUSIONS

UL-induced dynamic changes in monoamine neurotransmitters during the early compensation period in the rat MVN may be associated with the regulation of the central vestibular compensation mechanism by the MVN.

Dispelling Mist That Obscures Positional Vertigo in Vestibular Migraine

(1) Background: Patients with vestibular migraine (VM) often present with positional vertigo. A portion of these patients have features of benign paroxysmal positional vertigo (BPPV). It is a challenge to rapidly identify the BPPV component of VM associated with positional vertigo. (2) Methods: Retrospective data collected from 60 VM and 47 VM + BPPV patients were used to build a diagnostic model, and then prospective data from 47 patients were used for the external validation. All patients had VM manifesting as positional vertigo, with or without accompanying BPPV. The clinical manifestations and the results of vestibular function tests were comprehensively analyzed using logistic regression. (3) Results: The univariate and multivariate analyses showed that the age, symptom duration, tinnitus, ear fullness, nausea, head shaking nystagmus, the direction of the Dix-Hallpike and roll tests, and horizontal gain could help differentiate between the two groups. A nomogram and an online calculator were generated. The C-index was 0.870. The diagnostic model showed good discriminative power and calibration performance during internal and external validation. (4) Conclusions: This study provided a new perspective for diagnosing VM with positional vertigo by identifying the BPPV component and, for the first time, offers a prediction model integrating multiple predictors.

Time-feature attention-based convolutional auto-encoder for flight feature extraction

Quick Access Recorders (QARs) provide an important data source for Flight Operation Quality Assurance (FOQA) and flight safety. It is generally characterized by large volume, high-dimensionality and high frequency, and these features result in extreme complexities and uncertainties in its usage and comprehension. In this study, we proposed a Time-Feature Attention (TFA)-based Convolutional Auto-Encoder (TFA-CAE) network model to extract essential flight features from QAR data. As a case study, we used the QAR data landing at the Kunming Changshui International Airport and Lhasa Gonggar International Airport as the experimental data. The results show that (1) the TFA-CAE model performs the best in extracting representative flight features in comparison to some traditional or similar approaches, such as Principal Component Analysis (PCA), Convolutional Auto-Encoder (CAE), Self-Attention-based CAE (SA-CAE), Gate Recurrent Unit based Auto-Encoder (GRU-AE) and TFA-GRU-AE models; (2) flight patterns corresponding to different runways can be recognized; and (3) anomalous flights can effectively deviate from many observations. Overall, the TFA-CAE model provides a well-established technique for further usage of QAR data, such as flight risk detection or FOQA.

Identification of potential candidate genes and regulatory pathways related to reproductive capacity in hypothalamus and pituitarium of male ducks (Anas platyrhynchos) by differential transcriptome analysis

The improvement of reproductive capacity of poultry is important for the poultry industry. The existing studies on reproductive capacity mainly focus on the testis tissue, but few reports on regulationary effect of brain neuroendocrime on reproductive capacity have been available. The hypothalamus-pituitarium-gonad (HPG) axis is an important pathway regulating spermatogenesis and sexual behavior. This study analyzed the gene expression in the hypothalamus and pituitary tissues of male ducks in high-semen-quality group (DH), low-semen-quality group (DL), and non-response group (DN) by RNA-sequencing. A total of 1980 differentially expressed genes (DEGs) were identified, and significantly less DEGs were found in pituitary gland than in hypothalamus. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses showed that these DEGs were mainly enriched in nerve-related and synapse-related biological processes, mitochondrial inner membrane formation pathway, and ribosome structure pathway. Notably, the neuroactive ligand-receptor interaction pathway significantly enriched in all three comparisons (DH vs. DL, DH vs. DN, and DL vs. DN) was related to different reproductive performance such as semen quality and sexual response. Furthermore, six genes, including POMC, CPLX2, HAPLN2, EGR4, TOX3, and MSH4, were identified as candidate genes regulating reproductive capacity. Our findings provide new insights into the regulation mechanisms underlying the reproductive performance of male poultry, and offer a valuable reference for duck breeding programs aimed at promoting reproductive capacity.

Effects of ultra-long fermentation time on the microbial community and flavor components of light-flavor Xiaoqu Baijiu based on fermentation tanks

Microbial structure and succession of fermented grains play a significant role in Baijiu's flavor and quality. In this study, high-throughput sequencing (HTS) coupled with headspace solid-phase microextraction-gas chromatography-mass spectrometry (HS-SPME-GC-MS) were used to analyze the microbial community structures and flavor components in the fermented grains at the end of fermentation from different fermentation time of light-flavor Xiaoqu Baijiu. HTS results showed that Lactobacillus acetotolerans, Lactobacillus helveticus, Lactobacillus buchneri, Wickerhamomyces, Saccharomyces, and Condenascus were identified as the dominant microbes, but Lactobacillus (96.28%) exhibited obvious advantages at the end of ultra-long fermentation time (day 98). HS-SPME-GC-MS analysis revealed that esters and alcohols had the most abundance in fermented grains of day 98, containing high concentrations of ethyl acetate, diethyl succinate, phenylethyl alcohol, isoamyl alcohol, and n-propanol, which were related to the succession of Lactobacillus and yeast communities. Interestingly, the content of n-propanol in the ultra-long fermentation time samples (day 98) was 6 times of that in normal fermented grains (day 14), which may be caused by higher abundance of Lactobacillus in day 98 samples. Monte Carlo permutation test showed residual starch, acidity, and amino nitrogen (p < 0.05) were important factors affecting the microbial community. Together, these results shed light on the physicochemical changes, microbial dynamics, and key flavor components of fermented grains at the end of fermentation from different fermentation time and provide a strategy for further improvement of Baijiu quality.