• Proceedings of the Korea Information Processing Society Conference
• /
• 2019.10a
• /
• pp.1151-1152
• /
• 2019

발달 장애는 신체 및 정신이 해당하는 나이에 맞게 발달하지 않은 상태로, 다른 아동에 비해 신경정신과적 질환 발생 확률이 높기 때문에 발달장애 아동의 치료는 매우 중요하다. 그러나 주관적 판단에 의해 이루어지는 기존 작업치료의 경우, 정량적 성과 지표를 확인하기 힘들고 대상자 스스로 지속적으로 진행하기에 한계가 있다. 본 연구에서는 치료 모델을 가상 공간상에 구현하여 공간에 구애받지 않고 치료를 진행할 수 있으며, 수행 결과에 대한 자료를 정확하고 지속적으로 기록하며 확인할 수 있도록 하였다. 또한, AI 스피커를 통해 치료에 대한 피드백을 줌으로써, 대상자 스스로 실시하여 치료자의 개입을 줄여 심리적 부담을 덜어 더욱 정확한 수행이 이루어지도록 하였다.

• Journal of The Korean Association of Information Education
• /
• v.18 no.4
• /
• pp.559-566
• /
• 2014

This is to understand the neurological dynamic cognitive processes of math learning based on the abstract mappings( level A2), abstract systems(level A3), and single principles(level A4), which are principles of Fischer's cognitive development theory. Math learning requires flexibility to adapt existing brain function in selecting new neurophysiological activities to learn desired knowledge. This study suggests a general statistical framework for the identification of neurological patterns in different abstract learning change with optimal support. We expected that functional brain networks derived from a simple math learning would change dynamically during the supportive learning associated with different abstract levels. Task based patterns of the brain structure and function on representations of underlying connectivity suggests the possible prediction for the success of the supportive learning.

• Clinical and Experimental Pediatrics
• /
• v.53 no.1
• /
• pp.80-84
• /
• 2010

Purpose : The aims of this study were to investigate the long-term outcomes in children with infantile spasms (IS) and to identify the prognostic factors influencing their neurodevelopment. Methods : We retrospectively evaluated seventy two children over five years old who were treated for IS at Asan Medical Center, Seoul, Korea, between 1994 and 2007. Forty-three children were contacted by telephone or medical follow-up to assess their current neurodevelopmental status. Multiple logistic regression was used to calculate odds ratios (ORs) and 95% confidence interval (95% CIs) of risk factors for unfavorable outcomes.Results : The mean follow-up duration for these 43 children was $7.2{\pm}1.5$ years (range, 4.5 to 13.0 years). Of these, 13 (30.2%) had cryptogenic and 30 (69.8%) had symptomatic IS. Eleven (25.6%) children were initially treated with adrenocorticotrophic hormone (ACTH) therapy, with a mean treatment lag of $1.3{\pm}1.9$ months (range; 0.1 to 7.0 months). Eighteen (41.8%) children clinically responded to initial treatment, as shown by EEG response. Overall, 22 (51.2%) children had at least moderate neurodevelopmental disorders and 2 (4.8%) died. In univariate analysis, etiology (symptomatic) and poor electroclinical response to initial treatment were related to long-term unfavorable outcomes. In multivariate analysis, response to primary treatment was the sole significant independent risk factor with a high OR. Conclusion : Overall prognosis of children with IS was poor. Electroclinical non-responsiveness to initial treatment was related to unfavorable long-term outcomes, indicating that initial control of seizures may be important in reducing the likelihood of poor neurodevelopment.

• The Korean Journal of Zoology
• /
• v.39 no.1
• /
• pp.54-64
• /
• 1996

Normal development of the olfactory placode was studied to describe the sequence of events involved in the development of the olfactory placode. It has been primarily concerned with the morphological differentiation of the sensory neurons, their initial growth, maturation patterns and the contacts of their axons with the primitive prosencephalic vesicle. The olfactory organ first appears at stage 23 as a paired thickening of the two ectodermal layers: the superficial non-nervous layer (NNL) and the inner nervous layer (NL). Receptor cells differentiate from the NL and the supporting cells develop from the NNL. After stage 26 the placodal cells begin to migrate toward the epithelial surface between the NNL cells and their apical processes reach the surface at stage 28. As the apical process reaches the epithelial surface, basal processes (presumptive axons) sprout from the base of the NL cells at stage 29/30. They penetrate the underlying telencephalon by stage 32. Sensory synaptic contacts first appear at stage 37/38. Some placodal cells remain at the olfactory epithelium as basal cells while other placodal cells differentiate into olfactory neurons. The results confirmed that neurons originate exclusively from the nervous layer of the ectoderm while supporting cells originate from the NNL layer. The results also indicate that the development of olfactory neuron is independent of information from the target ftssue.

• Journal of Chest Surgery
• /
• v.38 no.3 s.248
• /
• pp.229-232
• /
• 2005

Lumbar Sympathectomy is a surgery for plantar hyperhidrosis, vascular and other reflex sympathetic diseases and has a various indications and physiologic effects. However it is not performed actively compared to thoracic sympathectomy because of its invasiveness. Therefore, we tried to perform lumbar sympathectomy using mediastinoscopy with small incision and introduce this new surgical technique. Material and Method: From July 2003 to December 2004, 18 patients undewent lumbar sympathectomy with mediastinoscopy at Inje University Sanggye Paik Hospital. There were 12 males and 6 females whose mean age was 24.3$\pm$8.2 years ranging from 18 to 67 30 cases of lumbar sympathectomy was performed with mediastinoscopy of which 24 cases were for plantar hyperhidrosis and 6 cases for other diseases. Result: Mean operation time was 37.2$\pm$12.5 minutes and mean post operation hospital stay was 3.1$\pm$2.2 days. There was one post sympathetic neuralgia and one peritoneal opening. Conclusion: Lumbar sympathectomy using mediastinoscopy is a simple and effective technique and has the advantage of cosmetics, post operative pain and hospital stay. However, further studies with large number of cases should be carried out for better outcome.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.11 no.1
• /
• pp.103-112
• /
• 2007

Recently structures become longer and higher because of the developments of new materials and construction techniques. However, such modern structures are susceptible to excessive structural vibrations, which may induce problems of serviceability and structural damages. In this paper we attempt to control structural vibration using the probabilistic neural network(PNN) and the artificial neural network(ANN) based on the training pattern that consist of only the structural state vector and the control force. The state vectors of the structure and control forces made by linear quadratic regulator(LQR) algorithm are used for training pattern of PNN and ANN. The proposed algorithm is applied for the vibration control of the three story shear building under Northridge earthquake. Control results by the proposed PNN and ANN are compared with each other.

• Neonatal Medicine
• /
• v.18 no.2
• /
• pp.399-403
• /
• 2011

Brain abscesses in neonates are extremely rare and usually occur in patients with certain risk factors. A 1-month-old boy presented at the hospital with fever and irritability. As a result of preterm delivery and low birth weight, he had a history of admission to the neonatal intensive care unit. Neuroimaging revealed a large, space-occupying lesion in both frontal lobes, which was suspected to be an abscess with the midline shifting to the right. With a single aspiration and abscess drainage along with concurrent prolonged parenteral antibiotic therapy, the patient showed an excellent treatment outcome with normal development. The focus will be placed on minimally invasive surgical management as well as positive outcomes.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.22 no.10
• /
• pp.1314-1319
• /
• 2018

The current TW3 - based maximum height prediction technique used in KMAA(Korean Medical Academy of Auxology) is manual and subjective, and it requires a lot of time and effort in the medical treatment, while the interest in the child's growth is very high. In addition, the technique of classifying images using deep learning, especially convolutional neural networks, is used in many fields at a more accurate level than the human eyes, also there is no exception in the medical field. In this paper, we introduce a TW3 algorithm using deep learning, that uses the convolutional neural network to predict the growth level of the left hand bone, to predict the maximum height of child and youth in order to increase the reliability of predictions and improve the convenience of the doctor.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2021.06a
• /
• pp.185-185
• /
• 2021

기후변화로 인한 돌발 강우 등 이상 기후 현상이 증가함에 따라 정확한 강우예측의 중요성은 더 증가하는 추세이다. 전통적인 강우예측의 경우 기상수치모델 또는 외삽법을 이용한 레이더 기반 강우예측 기법을 이용하며, 최근 머신러닝 기술의 발달에 따라 이를 활용한 레이더 자료기반 강우예측기법이 개발되고 있다. 기존 머신러닝을 이용한 강우예측 모델의 경우 주로 시계열 이미지 예측에 적합한 2차원 순환 신경망 기반 기법(Convolutional Long Short-Term Memory, ConvLSTM) 또는 합성곱 신경망 기반 기법(Convolutional Neural Network(CNN) Encoder-Decoder) 등을 이용한다. 본 연구에서는 생성적 적대 신경망 기반 기법(Generative Adversarial Network, GAN)을 이용해 미래 강우예측을 수행하도록 하였다. GAN 방법론은 이미지를 생성하는 생성자와 이를 실제 이미지와 구분하는 구별자가 경쟁하며 학습되어 현재 이미지 생성 분야에서 높은 성능을 보여주고 있다. 본 연구에서 개발한 GAN 기반 모델은 기상청에서 제공된 2016년~2019년까지의 레이더 이미지 자료를 이용하여 초단기, 단기 강우예측을 수행하도록 학습시키고, 2020년 레이더 이미지 자료를 이용해 단기강우예측을 모의하였다. 또한, 기존 머신러닝 기법을 기반으로 한 모델들의 강우예측결과와 GAN 기반 모델의 강우예측결과를 비교분석한 결과, 본 연구를 통해 개발한 강우예측모델이 단기강우예측에 뛰어난 성능을 보이는 것을 확인할 수 있었다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.449-451
• /
• 2022

암호화 프로그램에서 난수생성기는 널리 사용되며 중요한 역할을 하므로 공격의 대상이 되기 쉽고, 따라서 높은 난수성을 확보해야 한다. 최근에는 인공 신경망 기술이 발달함에 따라 난수생성기에 딥러닝 기술을 적용하는 연구들이 다수 진행되었으며, 본 논문에서는 이러한 연구 동향에 대해 알아본다. 크게 난수를 생성하는 연구와 다음에 올 수를 예측하는 예측 공격으로 나뉜다. 공통적으로는 학습해야 할 대상인 난수가 시계열 데이터이므로 대부분의 연구들이 RNN, CNN-1D 신경망을 사용한다. 난수 생성을 위해서는 분류형 신경망이 아닌, 생성형 신경망과 강화학습을 주로 사용하였다. 대부분의 연구들이 NIST SP-800 테스트를 시행하였을 때 높은 난수성을 확보할 수 있었다. 이외에도 최근 양자 컴퓨터가 개발됨에 따라 양자 하드웨어로부터의 양자 난수 생성기에 대한 예측 공격에 관한 연구도 있다. 딥러닝 기반의 난수 생성기에 대해서, 향후에는 기존의 난수생성기보다 빠른 생성 속도를 달성할 수 있는 경량 구현에 대한 연구와 그에 대한 비교 및 평가가 있어야 할 것으로 생각된다.