Online Continual Learning in Acoustic Scene Classification: An Empirical Study.

Numerous deep learning methods for acoustic scene classification (ASC) have been proposed to improve the classification accuracy of sound events. However, only a few studies have focused on continual learning (CL) wherein a model continually learns to solve issues with task changes. Therefore, in this study, we systematically analyzed the performance of ten recent CL methods to provide guidelines regarding their performances. The CL methods included two regularization-based methods and eight replay-based methods. First, we defined realistic and difficult scenarios such as online class-incremental (OCI) and online domain-incremental (ODI) cases for three public sound datasets. Then, we systematically analyzed the performance of each CL method in terms of average accuracy, average forgetting, and training time. In OCI scenarios, iCaRL and SCR showed the best performance for small buffer sizes, and GDumb showed the best performance for large buffer sizes. In ODI scenarios, SCR adopting supervised contrastive learning consistently outperformed the other methods, regardless of the memory buffer size. Most replay-based methods have an almost constant training time, regardless of the memory buffer size, and their performance increases with an increase in the memory buffer size. Based on these results, we must first consider GDumb/SCR for the continual learning methods for ASC.

The Influence of Dental Hygienists’ Self-Leadership on Organizational Commitment and Quality of Medical Services / 치위생과학회지

Background@#Self-leadership, an action strategy that can maximize individual capabilities, can affect the organizational commitment of dental hygienists and ultimately improve the quality of medical services. This study aims to demonstrate the need for self-leadership and organizational commitment for dental hygienists and develop measures to improve the quality of medical services. @*Methods@#An online survey of dental hygienists working at dental hospitals and clinics in Seoul and Gyeonggi province, Republic of Korea was conducted from March 28 to May 1, 2022. A total of 341 questionnaires were returned and analyzed. The measurement tools were modified and supplemented based on the theories and models developed by Manz for self-leadership, Mowday for organizational commitment, and Cronin and Taylor for medical services. Descriptive statistics, independent t-tests, ANOVA, simple regression, and multiple regression analyses were performed using SPSS 25.0. @*Results@#In leadership education, self-leadership is based on participation experience, the number of participants, and when and where it is received. Organizational commitment comes from participation experience, and the quality of medical services has been found to affect participation experience and location. Self-leadership had an effect on the quality of medical services (β=0.497, t=10.551, p＜0.001; β =0.599, t=13.783, p＜0.001; β=0.353, t=7.601, p＜0.001) and organizational commitment was found to have a mediating effect. @*Conclusion@#Dental hygienists’ self-leadership has a positive effect on the quality of medical services through the formation of appropriate interrelationships within the organization. Therefore, self-leadership programs should be developed, participated in, and promoted to improve the self-leadership of dental hygienists. Moreover, hospitals should improve their environment to provide and improve self-leadership education.

Analysis and Validation of Cross-Modal Generative Adversarial Network for Sensory Substitution.

Visual-auditory sensory substitution has demonstrated great potential to help visually impaired and blind groups to recognize objects and to perform basic navigational tasks. However, the high latency between visual information acquisition and auditory transduction may contribute to the lack of the successful adoption of such aid technologies in the blind community; thus far, substitution methods have remained only laboratory-scale research or pilot demonstrations. This high latency for data conversion leads to challenges in perceiving fast-moving objects or rapid environmental changes. To reduce this latency, prior analysis of auditory sensitivity is necessary. However, existing auditory sensitivity analyses are subjective because they were conducted using human behavioral analysis. Therefore, in this study, we propose a cross-modal generative adversarial network-based evaluation method to find an optimal auditory sensitivity to reduce transmission latency in visual-auditory sensory substitution, which is related to the perception of visual information. We further conducted a human-based assessment to evaluate the effectiveness of the proposed model-based analysis in human behavioral experiments. We conducted experiments with three participant groups, including sighted users (SU), congenitally blind (CB) and late-blind (LB) individuals. Experimental results from the proposed model showed that the temporal length of the auditory signal for sensory substitution could be reduced by 50%. This result indicates the possibility of improving the performance of the conventional vOICe method by up to two times. We confirmed that our experimental results are consistent with human assessment through behavioral experiments. Analyzing auditory sensitivity with deep learning models has the potential to improve the efficiency of sensory substitution.

Accelerating On-Device Learning with Layer-Wise Processor Selection Method on Unified Memory.

Recent studies have applied the superior performance of deep learning to mobile devices, and these studies have enabled the running of the deep learning model on a mobile device with limited computing power. However, there is performance degradation of the deep learning model when it is deployed in mobile devices, due to the different sensors of each device. To solve this issue, it is necessary to train a network model specific to each mobile device. Therefore, herein, we propose an acceleration method for on-device learning to mitigate the device heterogeneity. The proposed method efficiently utilizes unified memory for reducing the latency of data transfer during network model training. In addition, we propose the layer-wise processor selection method to consider the latency generated by the difference in the processor performing the forward propagation step and the backpropagation step in the same layer. The experiments were performed on an ODROID-XU4 with the ResNet-18 model, and the experimental results indicate that the proposed method reduces the latency by at most 28.4% compared to the central processing unit (CPU) and at most 21.8% compared to the graphics processing unit (GPU). Through experiments using various batch sizes to measure the average power consumption, we confirmed that device heterogeneity is alleviated by performing on-device learning using the proposed method.

Multi-Path and Group-Loss-Based Network for Speech Emotion Recognition in Multi-Domain Datasets.

Speech emotion recognition (SER) is a natural method of recognizing individual emotions in everyday life. To distribute SER models to real-world applications, some key challenges must be overcome, such as the lack of datasets tagged with emotion labels and the weak generalization of the SER model for an unseen target domain. This study proposes a multi-path and group-loss-based network (MPGLN) for SER to support multi-domain adaptation. The proposed model includes a bidirectional long short-term memory-based temporal feature generator and a transferred feature extractor from the pre-trained VGG-like audio classification model (VGGish), and it learns simultaneously based on multiple losses according to the association of emotion labels in the discrete and dimensional models. For the evaluation of the MPGLN SER as applied to multi-cultural domain datasets, the Korean Emotional Speech Database (KESD), including KESDy18 and KESDy19, is constructed, and the English-speaking Interactive Emotional Dyadic Motion Capture database (IEMOCAP) is used. The evaluation of multi-domain adaptation and domain generalization showed 3.7% and 3.5% improvements, respectively, of the F1 score when comparing the performance of MPGLN SER with a baseline SER model that uses a temporal feature generator. We show that the MPGLN SER efficiently supports multi-domain adaptation and reinforces model generalization.

An Energy-Efficient Method for Human Activity Recognition with Segment-Level Change Detection and Deep Learning.

Human activity recognition (HAR), which is important in context awareness services, needs to occur continuously in daily life, owing to which an energy-efficient method is needed. However, because human activities have a longer cycle than HAR methods, which have analysis cycles of a few seconds, continuous classification of human activities using these methods is computationally and energy inefficient. Therefore, we propose segment-level change detection to identify activity change with very low computational complexity. Additionally, a fully convolutional network (FCN) with a high recognition rate is used to classify the activity only when activity change occurs. We compared the accuracy and energy consumption of the proposed method with that of a method based on a convolutional neural network (CNN) by using a public dataset on different embedded platforms. The experimental results showed that, although the recognition rate of the proposed FCN model is similar to that of the CNN model, the former requires only 10% of the network parameters of the CNN model. In addition, our experiments to measure the energy consumption on the embedded platforms showed that the proposed method uses as much as 6.5 times less energy than the CNN-based method when only HAR energy consumption is compared.

Endoglin haploinsufficiency is associated with differential regulation of extracellular matrix production during skin fibrosis and cartilage repair in mice.

Transforming growth factor (TGF)-ß is a multifunctional growth factor with potent pro-fibrotic effects. Endoglin is a TGF-ß co-receptor that strongly regulates TGF-ß signaling in a variety of cell types. Although aberrant regulation of TGF-ß signaling is known to play a key role in fibrotic diseases such as scleroderma and impaired cartilage repair, the significance of endoglin function in regulating these processes is poorly understood. Here we examined whether endoglin haploinsufficiency regulates extracellular (ECM) protein expression and fibrotic responses during bleomycin induced skin fibrosis and surgically induced osteoarthritis, using endoglin-heterozygous (Eng+/-) mice and wild-type (Eng+/+) littermates. Skin fibrosis was induced by injecting mice intradermally with bleomycin or vehicle. Osteoarthritis was induced surgically by destabilization of medial meniscus. Dermal thickness, cartilage integrity and ECM protein expression were then determined. Eng+/- mice subjected to bleomycin challenge show a marked decrease in dermal thickness (P < 0.005) and reduced collagen content and decreased collagen I, fibronectin, alpha-smooth muscle actin levels as compared to Eng+/+ mice, both under basal and bleomycin treated conditions. Eng+/- mice undergoing surgically induced osteoarthritis show no differences in the degree of cartilage degradation, as compared to Eng+/+ mice, although chondrocytes isolated from Eng+/- display markedly enhanced collagen II levels. Our findings suggest that endoglin haploinsufficiency in mice ameliorates bleomycin-induced skin fibrosis suggesting that endoglin represents a pro-fibrotic factor in the mouse skin. However, endoglin haploinsufficiency does not protect these mice from surgically indiced cartilage degradation, demonstrating differential regulation of endoglin action during skin and cartilage repair.

Variants of Lipolysis-Related Genes in Korean Patients with Very High Triglycerides.

We investigated the prevalence and characteristics of variants of five lipolysis-related genes in Korean patients with very high triglycerides (TGs). Twenty-six patients with TG levels >885 mg/dL were selected from 13545 Korean subjects. Five candidate genes, LPL, APOC2, GPIHBP1, APOA5, and LMF1, were sequenced by targeted next-generation sequencing. Predictions of functional effects were performed and matched against public databases of variants. Ten rare variants of three genes were found in nine (34.6%) patients (three in LPL, four in APOA5, and three in LMF1). Five were novel and all variants were suspected of being disease-causing. Nine were heterozygous, and one (3.8%) had a homozygous rare variant of LPL. Six common variants of four genes were observed in 25 (96.2%) patients (one in LPL, one in GPIHBP1, two in APOA5, and two in LMF1). The c.G41T variant of GPIHBP1 and c.G533T variant of APOA5 were most frequent and found in 15 (57.7%) and 14 (53.8%) patients, respectively. Rare homozygous variants of the genes were very uncommon, while diverse rare heterozygous variants were commonly identified. Taken together, most study subjects may be manifesting the combined effects of rare heterozygous variants and common variants.

Variants of Lipolysis-Related Genes in Korean Patients with Very High Triglycerides

We investigated the prevalence and characteristics of variants of five lipolysis-related genes in Korean patients with very high triglycerides (TGs). Twenty-six patients with TG levels >885 mg/dL were selected from 13545 Korean subjects. Five candidate genes, LPL, APOC2, GPIHBP1, APOA5, and LMF1, were sequenced by targeted next-generation sequencing. Predictions of functional effects were performed and matched against public databases of variants. Ten rare variants of three genes were found in nine (34.6%) patients (three in LPL, four in APOA5, and three in LMF1). Five were novel and all variants were suspected of being disease-causing. Nine were heterozygous, and one (3.8%) had a homozygous rare variant of LPL. Six common variants of four genes were observed in 25 (96.2%) patients (one in LPL, one in GPIHBP1, two in APOA5, and two in LMF1). The c.G41T variant of GPIHBP1 and c.G533T variant of APOA5 were most frequent and found in 15 (57.7%) and 14 (53.8%) patients, respectively. Rare homozygous variants of the genes were very uncommon, while diverse rare heterozygous variants were commonly identified. Taken together, most study subjects may be manifesting the combined effects of rare heterozygous variants and common variants.

In-hospital contact investigation among health care workers after exposure to pulmonary tuberculosis in an intermediate tuberculosis prevalence area: A prospective study.

We carried out in-hospital contact investigations of patients with pulmonary tuberculosis and analyzed the prevalence of latent tuberculosis infection (LTBI) among health care workers (HCWs) after TB exposure. A prospective study was conducted of 872 HCWs who were exposed to 55 index cases diagnosed with active pulmonary TB. HCWs after TB exposure were evaluated both TST and chest X-ray at the time of enrollment and 12 weeks after exposure; 625 HCWs (71.6%) underwent both initial assessments; 41 HCWs (6.6%) had a positive TST result. After 12 weeks, 71.1% of HCWs with initial negative TST (n = 415) underwent a second assessment. Ten HCWs had TST conversion. One HCW (0.2%) developed active pulmonary TB. In multivariable analysis, age over 30 years was associated with TST conversion (p = .02). Point prevalence of latent TB was 6.6%, and incidence of LTBI was estimated as 2.4 per 100 HCWs. Strict infection control measures should be emphasized in intermediate TB-burden, BCG-vaccinated countries, especially in HCWs with high risk for TB exposure.

Statin Intensity and Clinical Outcome in Patients with Stable Coronary Artery Disease and Very Low LDL-Cholesterol.

BACKGROUND: Although intensive statin therapy is recommended for high risk patients, evidence of its benefit in patients with stable coronary artery disease (CAD) and very low low-density lipoprotein-cholesterol (LDL-C) has been very rare. In this study, we investigated whether higher statin intensity reduces cardiovascular risks in this population. METHODS: In this retrospective study, a total of 5234 patients with stable CAD were screened at three tertiary hospitals in Korea; 449 patients (mean age: 65 years, male: 69%) with LDL-C <80 mg/dL were finally analyzed. The statin intensities were classified according to the 2013 American College of Cardiology/American Heart Association guidelines. Patients who received statins equivalent to or weaker than atorvastatin 10 mg (group 1) were compared with those who took statins equivalent to or stronger than atorvastatin 20 mg (group 2). The impact of statin intensity on major adverse cardiac events (MACE) was evaluated during follow-up. RESULTS: Group 1 and group 2 consisted of 181 patients (40.3%) and 268 patients (59.7%), respectively. The mean LDL-C level decreased to 52 and 57 mg/dL in group 1 and group 2, respectively, during follow-up. In a median follow-up of 4.5 years, patients of group 2 had a lower incidence of MACE (30 [16.6%] vs. 12 [4.5%], p <0.001), which were mostly related to a lower incidence of coronary revascularization. Cox proportional hazard analyses identified the statin intensity of group 2 (adjusted hazard ratio: 0.25, confidence interval: 0.11-0.55, p <0.001) and the baseline high-density lipoprotein-cholesterol level as independent determinants of MACE. CONCLUSION: This study provides evidence that higher intensity statins are beneficial for cardiovascular outcomes in patients with stable CAD and very low LDL-C. Statins equivalent to or stronger than atorvastatin 20 mg are more effective than lower intensity statins.

The Effect of FLT1 Variant on Long-Term Cardiovascular Outcomes: Validation of a Locus Identified in a Previous Genome-Wide Association Study.

BACKGROUND: Data on genetic variants that can predict follow-up cardiovascular events are highly limited, particularly for Asians. The aim of this study was to validate the effects of two variants in FLT1 and 9p21 on long-term cardiovascular outcomes in high-risk Korean patients. METHODS: We examined the prognostic values of the rs9508025 and rs1333049 variants that were found to be associated with coronary artery disease (CAD) risk in a previous Korean genome-wide association study. A total of 2693 patients (mean age: 55.2 years; male: 55.2%) with CAD or its risk factors at baseline were enrolled and followed for major adverse cardiac events (MACE). RESULTS: During the mean follow-up of 8.8 years, 15.4% of the patients experienced MACE. Kaplan-Meier curves showed that MACE-free survival was different according to the genotype of rs9508025 (log rank p = 0.02), whereas rs1333049 genotype did not correlate with the prognosis. Multivariate Cox proportional hazard analysis showed that C-allele of rs9508025 was significantly associated with a high rate of MACE, while rs1333049 was not. Further analyses demonstrated that the association of the rs9508025 variant with MACE was mainly due to its relation to coronary revascularization, which was also associated with the rs1333049 variant. In an additional analysis, rs9508025 was found to be an independent determinant of the outcome only in the subgroup with history of CAD. CONCLUSIONS: rs9508025 in FLT1 was significantly associated with long-term cardiovascular events, particularly in patients with prior CAD. The association of rs1333049 in 9p21 was not significant.

Mouth rehabilitation of a patient with severely worn dentition with vertical dimension increase / 대한치과보철학회지

Treating patients with severely worn dentition often requires comprehensive prosthetic treatments. Oral rehabilitation and long-term stability could be achieved by analyzing the cause of the wear and changing vertical dimension. The temporary restoration procedure is executed and patient's adaptation is evaluated by long-term observation when treatment is accompanied with increasing vertical dimension. In this case, 54 year-old male presented with worn dentition and esthetic dissatisfaction. In oral manifestations, severely worn dentition with attrition and erosion is diagnosed. To evaluate change of vertical dimension, interocclusal rest space, facial appearance, pronunciation, clinical crown length, distance from incisor edge to labial vestibule are evaluated. Consequently, 2.5 mm increase of vertical dimension that based on articulator incisal pin is determined. After 8 weeks stabilization period with temporary prostheses, definitive prostheses are fabricated. After 6 months follow up period, satisfactory outcomes were achieved both in functional and esthetic aspects through this procedure.

TGF-b signaling in cartilage homeostasis and osteoarthritis.

Healthy cartilage is maintained by a delicate balance between the anabolic and catabolic activities of articular chondrocytes. This involves actions of numerous cytokines and growth factors that regulate the synthesis and degradation of extracellular matrix components which maintain the functional integrity of the joint. An imbalance between the activities of these anabolic and catabolic factors leads to cartilage degradation resulting in osteoarthritis (OA), a chronic degenerative joint disorder characterized by destruction of articular cartilage, alterations of subchondral bone and synovial fibrosis. Among the cytokines and growth factors that have been studied in the context of cartilage homeostasis and OA, transforming growth factor-beta TGF-beta has emerged as an important molecule that plays a critical role in the development, growth, maintenance and repair of articular cartilage. Deregulation of its signaling and responses has been shown to be involved in OA. Several components of the TGF-beta pathway, including extracellular, cell surface and intracellular molecules, display altered expression or action in OA. In this review, we discuss the regulatory mechanisms of TGF-beta signaling and link these mechanisms to cartilage function, highlighting the important role of TGF-beta in maintaining cartilage function and integrity. We also summarize the alterations in the molecular events of TGF-beta signaling and responses that may contribute to OA progression and discuss the potential of targeting the TGF-beta signaling pathway for the development of novel therapies for OA.

Delirium tremens associated with fracture of mandible: Report of A Case

Delirium tremens is a severe state of alcohol-related withdrawal syndrome, which is precipitated in the chronic alcoholic who are suddenly deprived of alcohol because of traumatic injury or other diseases. About 5 % of alcoholics show evidence of severe withdrawal symptoms. Symptoms begin within 1 to 3 days after stopping ethanol intake. These include a state of confusion sometimes accompanied by visual, tactile, or auditory hallucinations. The diagnosis is made when the course progresses beyond the usual symptoms of withdrawal to include confusion, severe agitation, and generalized seizures. The likelihood of developing severe withdrawal symptoms increases with concomitant infections or medical problems, a prior history of withdrawal seizures of DTs, and higher quantity and frequency of drinking. Most periods of severe withdrawal begin and end abruptly, rarely lasting longer than 3 to 5 days. The mortality risk for DTs is quite low but increases with preexisting medical illnesses or organ system failure. We experienced a case of DTs associated with fracture of mandible. The patient was a 36-years-old male who was admitted to our hospital via emergency room after suffering from a traffic accident. He developed DTs 3 days after admission and eventually expired. The report of a case and review of literatures are as follows.

Conservative care of pathologic fracture by osteoradionecrosis of mandible: report of two cases

The main pathogenesis of osteoradionecrosis is the initial radiation cellular kill, plus the progressive tissue hypoxia owing to endarteritis, plus late expressions of lethal mutations in stromal cells. Its treatment comprised (1) avoidance of local physical, chemical & biological irritants, (2) wound irrigation, (3) antibiotics & analgesics, (4) superficial sequestrectomy, (5) hyperbaric oxygen therapy, (6) resection and reconstructive surgery, and (7) supportive therapy for systemic condition. Because of the cumulative effect of radiation therapy, it is important to manage the osteoradionecrosis conservatively & continuously, with emotional & physical support. The authors think that conservative care is more essential to the patient of poor systemic condition, than the aggressive resection & reconstructive surgery. We report two cases of pathologic fracture by osteoradionecrosis of mandible that were cared conservatively in the department of dentistry, Wonju Christian Hospital. The contents of treatment are continuous wound dressing, medications (antibiotics and analgesics) in acute infection, incision and drainage on new abscess, emotional support (stress management), proper nutritional and physical supports (exercise and physical therapy). The prognosis of continuous conservative care is relatively favorable.