RESUMEN
The purpose of this study was to analyze the mediating role of digital health literacy and the moderating effect of parasocial relationships on the relationship between the viewing experience of health exercise-related YouTube content and the intention for health exercise behavior. Based on the health action process approach, this study established a foundational theoretical model to analyze how digital health literacy mediates the impact of media viewing experience on health exercise behavior intention. Additionally, this study examined the moderating effect of parasocial relationships with YouTube creators. For empirical analysis, variables were measured using a self-administration method among 409 randomly sampled consumers of YouTube health exercise content. The collected data were analyzed using a structural equation model incorporating mediation parameters, and a multigroup model analysis was conducted to understand differences based on parasocial relationships. The results revealed that increased YouTube viewing experience enhanced cognitive, skill, and evaluative components of digital health literacy, which were significant factors in increasing health exercise behavior intention. Notably, the mediating effect of cognition played a crucial role, and the strengthening effect of parasocial relationships on this relationship was confirmed. These findings can be utilized as practical foundational data for designing digital health communication strategies, particularly in developing motivational mechanisms that encourage consumers to engage voluntarily and consistently in health behaviors based on online health information.
RESUMEN
Deep learning methods have achieved a lot of success in various applications involving converting wearable sensor data to actionable health insights. A common application areas is activity recognition, where deep-learning methods still suffer from limitations such as sensitivity to signal quality, sensor characteristic variations, and variability between subjects. To mitigate these issues, robust features obtained by topological data analysis (TDA) have been suggested as a potential solution. However, there are two significant obstacles to using topological features in deep learning: (1) large computational load to extract topological features using TDA, and (2) different signal representations obtained from deep learning and TDA which makes fusion difficult. In this paper, to enable integration of the strengths of topological methods in deep-learning for time-series data, we propose to use two teacher networks - one trained on the raw time-series data, and another trained on persistence images generated by TDA methods. These two teachers are jointly used to distill a single student model, which utilizes only the raw time-series data at test-time. This approach addresses both issues. The use of KD with multiple teachers utilizes complementary information, and results in a compact model with strong supervisory features and an integrated richer representation. To assimilate desirable information from different modalities, we design new constraints, including orthogonality imposed on feature correlation maps for improving feature expressiveness and allowing the student to easily learn from the teacher. Also, we apply an annealing strategy in KD for fast saturation and better accommodation from different features, while the knowledge gap between the teachers and student is reduced. Finally, a robust student model is distilled, which can at test-time uses only the time-series data as an input, while implicitly preserving topological features. The experimental results demonstrate the effectiveness of the proposed method on wearable sensor data. The proposed method shows 71.74% in classification accuracy on GENEActiv with WRN16-1 (1D CNNs) student, which outperforms baselines and takes much less processing time (less than 17 sec) than teachers on 6k testing samples.
RESUMEN
We report the use of an advanced magnetic resonance image (MRI) sequence to detect the treatment response after SRS for aggressive vertebral haemangioma (VH). A 63-year-old female patient presented with back pain, bilateral lower extremity weakness (grade IV), and sensory change in the saddle area. MRI revealed a vertebral body mass compressing the spinal cord at T10, which had high T2 and low T1 signal intensity. Three-dimensional volumetric sagittal time-resolved imaging of contrast kinetics (TRICKS) abdominal magnetic resonance angiography (MRA) showed it to be hypervascular. SRS with the Novalis beam shaping system (BrainLAB; Heimstetten®, Germany) was performed on the gross tumor volume of 14.954 mL. 30 Gy was given to the 90% isodose line in 5 fractions. Seven days later, the patient underwent decompressive laminectomy for weakness. Seven months later, the patient's motor weakness was improved to allow for unassisted gait, and back pain and sensory changes resolved. Follow-up MRI revealed no significant change on T1 and T2 signal intensity images. However, TRICKS abdominal MRA demonstrated disapprearance of the hypervascularity. Seven years after SRS, the same signal intensity images showed shrinkage of the mass and resolution of compression of the spinal cord, and the signal intensity of the T1 image was changed to iso- and high signal intensity.
Asunto(s)
Hemangioma , Radiocirugia , Femenino , Humanos , Persona de Mediana Edad , Estudios de Seguimiento , Radiocirugia/métodos , Columna Vertebral , Imagen por Resonancia Magnética/métodos , Hemangioma/diagnóstico por imagen , Hemangioma/radioterapia , Hemangioma/cirugíaRESUMEN
Wearable sensor data analysis with persistence features generated by topological data analysis (TDA) has achieved great successes in various applications, however, it suffers from large computational and time resources for extracting topological features. In this paper, our approach utilizes knowledge distillation (KD) that involves the use of multiple teacher networks trained with the raw time-series and persistence images generated by TDA, respectively. However, direct transfer of knowledge from the teacher models utilizing different characteristics as inputs to the student model results in a knowledge gap and limited performance. To address this problem, we introduce a robust framework that integrates multimodal features from two different teachers and enables a student to learn desirable knowledge effectively. To account for statistical differences in multimodalities, entropy based constrained adaptive weighting mechanism is leveraged to automatically balance the effects of teachers and encourage the student model to adequately adopt the knowledge from two teachers. To assimilate dissimilar structural information generated by different style models for distillation, batch and channel similarities within a mini-batch are used. We demonstrate the effectiveness of the proposed method on wearable sensor data.
RESUMEN
Converting wearable sensor data to actionable health insights has witnessed large interest in recent years. Deep learning methods have been utilized in and have achieved a lot of successes in various applications involving wearables fields. However, wearable sensor data has unique issues related to sensitivity and variability between subjects, and dependency on sampling-rate for analysis. To mitigate these issues, a different type of analysis using topological data analysis has shown promise as well. Topological data analysis (TDA) captures robust features, such as persistence images (PI), in complex data through the persistent homology algorithm, which holds the promise of boosting machine learning performance. However, because of the computational load required by TDA methods for large-scale data, integration and implementation has lagged behind. Further, many applications involving wearables require models to be compact enough to allow deployment on edge-devices. In this context, knowledge distillation (KD) has been widely applied to generate a small model (student model), using a pre-trained high-capacity network (teacher model). In this paper, we propose a new KD strategy using two teacher models - one that uses the raw time-series and another that uses persistence images from the time-series. These two teachers then train a student using KD. In essence, the student learns from heterogeneous teachers providing different knowledge. To consider different properties in features from teachers, we apply an annealing strategy and adaptive temperature in KD. Finally, a robust student model is distilled, which utilizes the time series data only. We find that incorporation of persistence features via second teacher leads to significantly improved performance. This approach provides a unique way of fusing deep-learning with topological features to develop effective models.
RESUMEN
Topological features such as persistence diagrams and their functional approximations like persistence images (PIs) have been showing substantial promise for machine learning and computer vision applications. This is greatly attributed to the robustness topological representations provide against different types of physical nuisance variables seen in real-world data, such as view-point, illumination, and more. However, key bottlenecks to their large scale adoption are computational expenditure and difficulty incorporating them in a differentiable architecture. We take an important step in this paper to mitigate these bottlenecks by proposing a novel one-step approach to generate PIs directly from the input data. We design two separate convolutional neural network architectures, one designed to take in multi-variate time series signals as input and another that accepts multi-channel images as input. We call these networks Signal PI-Net and Image PI-Net respectively. To the best of our knowledge, we are the first to propose the use of deep learning for computing topological features directly from data. We explore the use of the proposed PI-Net architectures on two applications: human activity recognition using tri-axial accelerometer sensor data and image classification. We demonstrate the ease of fusion of PIs in supervised deep learning architectures and speed up of several orders of magnitude for extracting PIs from data. Our code is available at https://github.com/anirudhsom/PI-Net.
RESUMEN
Animal tracking is an important tool for observing behavior, which is useful in various research areas. Animal specimens can be tracked using dynamic models and observation models that require several types of data. Tracking mouse has several barriers due to the physical characteristics of the mouse, their unpredictable movement, and cluttered environments. Therefore, we propose a reliable method that uses a detection stage and a tracking stage to successfully track mouse. The detection stage detects the surface area of the mouse skin, and the tracking stage implements an extended Kalman filter to estimate the state variables of a nonlinear model. The changes in the overall shape of the mouse are tracked using an oval-shaped tracking model to estimate the parameters for the ellipse. An experiment is conducted to demonstrate the performance of the proposed tracking algorithm using six video images showing various types of movement, and the ground truth values for synthetic images are compared to the values generated by the tracking algorithm. A conventional manual tracking method is also applied to compare across eight experimenters. Furthermore, the effectiveness of the proposed tracking method is also demonstrated by applying the tracking algorithm with actual images of mouse. Graphical abstract.
Asunto(s)
Conducta Animal/fisiología , Movimiento/fisiología , Algoritmos , Animales , Procesamiento de Imagen Asistido por Computador/métodos , Ratones , Piel/fisiopatologíaRESUMEN
Spinal subdural abscesses (SSA) are very rare disease. The etiologies of SSA are hematogenous spread, iatrogenic contamination, and local extension. Elevated WBC counts, ESR, and C-reactive protein are usually found in laboratory tests. But they are not sensitive indicators of SSA, especially chronic abscesses patient tend to have a less specific characteristic. We report the case of a healthy man with chronic subdural abscess referred to our hospital as an intradural-extramedullary (IDEM) tumor. The patient presented with voiding difficulty and pain in the back and left leg. In a contrast MRI scan, a rim-enhanced mass-like lesion was seen at the L5/S1 level. But adjacent ill-defined epidural fat enhancement that are unusual imaging manifestation for IDEM tumors was seen. He had no fever and normal WBC, ESR, and CRP. In addition, the patient had no previous infection history or other disease, but he did have an epidural block for back pain at another hospital 2 years previously. So, we repeated the MRI with a high-resolution 3-T scanner. The newly taken MR images in our hospital revealed a clear enlargement of lesion size compared to the previous MRI taken 1 week before in other hospital. We suspected a chronic spinal subdural abscess with recent aggravation and immediately performed surgical evacuation. In the surgical field, tensed dura was observed and pus was identified after opening the abscess capsule. Because chronic spinal subdural abscesses are difficult to diagnose, we could differentiate with IDEM tumor exactly and an exact history taking, contrast MRI are required.
Asunto(s)
Diagnóstico Diferencial , Empiema Subdural/diagnóstico , Empiema Subdural/microbiología , Enfermedades de la Médula Espinal/microbiología , Neoplasias de la Médula Espinal/diagnóstico , Infecciones Estafilocócicas/diagnóstico , Adulto , Enfermedad Crónica , Humanos , Inyecciones Epidurales/efectos adversos , Masculino , Staphylococcus aureus , Espacio Subdural/microbiología , Espacio Subdural/patologíaRESUMEN
OBJECTIVE: To investigate pregnancy outcomes subsequent to ovarian pregnancy treated by surgery. METHODS: A retrospective analysis was conducted of ovarian pregnancies that were treated by surgery at a hospital in Korea between January 1996 and December 2009. RESULTS: Forty-nine women with ovarian pregnancies (1.6% of all ectopic pregnancies) were treated; 28 of these patients who were followed-up for more than a year were included in the study. The most common risk factor for ovarian pregnancy was endometriosis (42.9%). Accurate diagnosis of ovarian pregnancy was made preoperatively in 7 patients (25%). Of the 28 patients, 16 (57.1%) had subsequent pregnancies: 13 (46.4%) were intrauterine pregnancies and 3 (10.7%) were tubal pregnancies. However, no subsequent ovarian pregnancies occurred. In addition, only 1 patient had secondary infertility after surgery for ovarian pregnancy. CONCLUSIONS: After an ovarian pregnancy treated by surgery, the outcome of a subsequent pregnancy is reasonable; there is a high rate of successful subsequent pregnancy and a low rate of subsequent ectopic pregnancy or of infertility.
Asunto(s)
Resultado del Embarazo , Embarazo Ectópico/cirugía , Adulto , Endometriosis/complicaciones , Femenino , Humanos , Infertilidad Femenina/etiología , Complicaciones Posoperatorias/epidemiología , Embarazo , Embarazo Ectópico/diagnóstico , Embarazo Ectópico/etiología , República de Corea/epidemiología , Estudios Retrospectivos , Factores de Riesgo , Resultado del TratamientoRESUMEN
OBJECTIVE: To clinically analyze cases of ectopic ovarian pregnancy and to generate data regarding the evaluation and management of suspected ectopic ovarian pregnancies. STUDY DESIGN: We retrospectively analyzed 49 ovarian pregnancies that were surgically treated at Cheil General Hospital and Women's Healthcare Center between January 1996 and December 2009. We analyzed patient age, parity, symptoms, risk factors, preoperative diagnosis, and ovarian pregnancy type. RESULTS: During the study period, the incidence of ovarian pregnancy was 1.59% of all ectopic pregnancies (49/3081); 45/49 (91.8%) were primary ovarian pregnancies. At the time of diagnosis, mean age was 30.7 years (SD: ± 4.4 years) and mean parity was 0.63 (SD: ± 0.8). The most common presenting symptoms were abdominal pain (42.9%) and vaginal bleeding (28.6%). The most common sonographic findings were fluid surrounding the ovarian pregnancy and ovarian enlargement. In regard to surgical treatment, ovarian wedge resection was most often performed (85.7% of cases), followed by oophorectomy (8.2% of cases). The most common risk factors were endometriosis (16 patients) and a history of abdominal surgery (19 patients). CONCLUSIONS: Ovarian pregnancies are extremely rare and difficult to diagnose both pre- and intra-operatively. Our data may assist surgeons in understanding the clinical presentation of ovarian pregnancy and in counseling patients. Larger studies are warranted to gather more data on this rare form of ectopic pregnancy.
Asunto(s)
Embarazo Ectópico/epidemiología , Adulto , Femenino , Humanos , Ovario/diagnóstico por imagen , Embarazo , Embarazo Ectópico/diagnóstico por imagen , Prevalencia , República de Corea/epidemiología , Estudios Retrospectivos , Factores de Riesgo , UltrasonografíaRESUMEN
INTRODUCTION: The purpose of this study was to compare the efficacy of a laparoscopy-assisted surgical staging with a traditional laparotomy staging for the treatment of endometrial cancer. METHODS: We retrospectively analyzed the medical records of 465 patients with endometrial adenocarcinoma treated by surgery between January 1996 and December 2007. RESULTS: There were no significant differences between the laparoscopy and the laparotomy groups in age, body mass index, and histologic type. However, in the laparotomy group, grade and surgical stage were higher, the diseases were more chronic, and more postoperative adjuvant treatments were necessary. One hundred seven (76.4%) of 140 patients in the laparoscopy group and 260 (80.0%) of the 325 patients in the laparotomy group had lymphadenectomy, and the median numbers of pelvic and paraaortic lymph nodes obtained were not statistically different. The laparoscopy group showed shorter postoperative hospital stay and lower blood loss, and the operating time was also shorter than that in the laparotomy group. There was no significant difference in intraoperative or postoperative complications, and the operative technique did not influence survival rates after adjusting several confounding factors. CONCLUSIONS: Our data of 12 years with a large number of patients show no differences in complications and impacts on survival between laparoscopy and laparotomy. Laparoscopy has advantages of shorter operating time and other advantages over laparotomy previously reported. Therefore, laparoscopy can be considered a good therapeutic option for endometrial cancer.