A5011699970- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- COVID-19 diagnosis using AI
- Artificial Intelligence in Healthcare and Education
- Oral microbiology and periodontitis research
- Pressure Ulcer Prevention and Management
- Colorectal Cancer Screening and Detection
- Cardiac electrophysiology and arrhythmias
- Machine Learning in Healthcare
- Sepsis Diagnosis and Treatment
- ECG Monitoring and Analysis
- Global Cancer Incidence and Screening
- Pectus Deformity Diagnosis and Treatment
- Generative Adversarial Networks and Image Synthesis
- Natural Language Processing Techniques
- linguistics and terminology studies
- Advanced Steganography and Watermarking Techniques
- Vascular Procedures and Complications
- Oral Health Pathology and Treatment
- Digital Media Forensic Detection
- Speech and dialogue systems
- Antimicrobial agents and applications
- Complementary and Alternative Medicine Studies
- Venous Thromboembolism Diagnosis and Management
- Bacterial biofilms and quorum sensing
Hekinan Technical High School
2022-2024
New York University
2020-2022
City Hospital
2022
NYU Langone Health
2020
Tsurumi University
2011-2014
Fukuoka University
2011-2013
Fukuoka University Hospital
2011
Abstract During the coronavirus disease 2019 (COVID-19) pandemic, rapid and accurate triage of patients at emergency department is critical to inform decision-making. We propose a data-driven approach for automatic prediction deterioration risk using deep neural network that learns from chest X-ray images gradient boosting model routine clinical variables. Our AI prognosis system, trained data 3661 patients, achieves an area under receiver operating characteristic curve (AUC) 0.786 (95% CI:...
Deep neural networks (DNNs) show promise in image-based medical diagnosis, but cannot be fully trusted since they can fail for reasons unrelated to underlying pathology. Humans are less likely make such superficial mistakes, use features that grounded on science. It is therefore important know whether DNNs different than humans. Towards this end, we propose a framework comparing human and machine perception diagnosis. We frame the comparison terms of perturbation robustness, mitigate...
Real-world datasets often combine data collected under different experimental conditions. This yields larger datasets, but also introduces spurious correlations that make it difficult to model the phenomena of interest. We address this by learning two embeddings independently represent interest and correlations. The embedding representing is correlated with target variable $y$, invariant environment $e$. In contrast, invariance $e$ achieve on real-world datasets. Our primary contribution an...
Abstract Background: Cumulative interceptive supportive therapy (CIST) is currently used as a guideline for treating peri‐implant diseases. The objectives of this study were to determine the detection rate and measure number periodontopathic bacteria in lesions different CIST levels thereby characterize disease from bacteriological viewpoint. Methods: This included 105 patients who had both residual natural teeth implants with disease. A total divided into A, B, C D according classification....
Indwelling foreign-body infections are a critical medical problem, especially in immunocompromised patients. To examine the pathogenicity of biofilm-forming bacteria settling on foreign materials, mice implanted with plastic discs were infected Staphylococcus aureus. After opening wide subcutaneous pocket dorsal side or without temporal leukocytopenia, sheet was placed left space; subsequently, planktonic state dispersed over space. Bacterial numbers examined 7 days after inoculation. In...
When using a free flap to reconstruct facial deformity caused by Romberg's disease, it is important prevent the from sagging after operation. We report new method of reconstructive surgery subscapular adipofascial this problem.Three female patients (ages 27, 28, and 34 years) with Parry-Romberg syndrome underwent microsurgical scapular transfer for buccal defects. This operation requires making gingivobuccal sulcus incision forming pocket fat reconstruction dissecting over periosteum...
Abstract Background We aimed to investigate the association between ventricular repolarization instability and sustained tachycardia fibrillation (VT/VF) occurring within 48 h (acute‐phase VT/VF) after onset of acute coronary syndrome (ACS) prognostic role heart rate variability (HRV) discharge from hospital. Methods studied 572 ACS patients with a left ejection fraction >35%. The was assessed by beat‐to‐beat T‐wave amplitude (TAV) using high‐resolution 24‐h Holter ECGs recorded at median...
We propose generative multitask learning (GMTL), a simple and scalable approach to causal representation for learning. Our makes minor change the conventional inference objective, improves robustness target shift. Since GMTL only modifies it can be used with existing methods without requiring additional training. The improvement in comes from mitigating unobserved confounders that cause targets, but not input. refer them as \emph{target-causing confounders}. These induce spurious...
Supervised multi-modal learning involves mapping multiple modalities to a target label. Previous studies in this field have concentrated on capturing isolation either the inter-modality dependencies (the relationships between different and label) or intra-modality within single modality label). We argue that these conventional approaches rely solely inter- may not be optimal general. view problem from lens of generative models where we consider as source interaction them. Towards end,...
Acute heart failure is an important cause of unplanned hospitalizations and poses a significant burden through increased mortality frequent hospitalizations. Heart with preserved ejection fraction (HFpEF) presents as diverse condition characterized by complex cardiovascular non-cardiovascular pathology. This study aimed to identify distinct clinical phenotypes in acute decompensated HFpEF (ADHF) using cluster analysis assess their prognostic significance. We applied latent class 1,281 ADHF...
褥瘡は全ての診療科に関わる重要な疾病である。その評価方法であるDESIGNシステムの当院の看護師への浸透度をアンケート形式で調査した。有効回答数584(総数620，回収率94％)の中でDESIGNシステムについて認知しているものが79％，実際に採点したことがあるものは全体の35％であった。勤務年数との関係では経験年数の増加に応じて認知，実施経験が有意に上昇していた。経験した診療科による浸透度の差は，神経疾患(脳神経外科・神経内科)でやや高い傾向がみられたが，平均値と有意な差はなかった。判定項目別の難易度を調査したところ，滲出液(E)，大きさ(S)は55％，49％がそれぞれ迷わず判定できると答えたのに対し，深さ(D)，炎症/感染(I)，肉芽組織(G)，ポケット(P)，壊死組織(N)ではそれぞれ88％，88％，84％，78％，75％が判定にときどき迷うまたはいつも迷うことが解った。これらの情報は今後のDESIGNの浸透を目指す教育やDESIGN自身の再評価に向けての基礎データになると考えられた。
Breast cancer is the most common in women, and hundreds of thousands unnecessary biopsies are done around world at a tremendous cost. It crucial to reduce rate that turn out be benign tissue. In this study, we build deep neural networks (DNNs) classify biopsied lesions as being either malignant or benign, with goal using these second readers serving radiologists further number false positive findings. We enhance performance DNNs trained learn from small image patches by integrating global...
Deep neural networks (DNNs) show promise in image-based medical diagnosis, but cannot be fully trusted since their performance can severely degraded by dataset shifts to which human perception remains invariant. If we better understand the differences between and machine perception, potentially characterize mitigate this effect. We therefore propose a framework for comparing diagnosis. The two are compared with respect sensitivity removal of clinically meaningful information, regions an...
褥瘡診療において全身状態の悪化とともに褥瘡が発生・悪化するのはしばしば経験されることである。逆に創は改善しているのに全身状態が悪化し死に至る例にも遭遇する。褥瘡の状態と患者の転帰に因果関係があるかについて福岡大学病院で2008年3月から2009年8月までに診療した入院中の褥瘡患者92例について検討した。褥瘡の重症度評価はDESIGN-Rを用いた。背景疾患では循環器疾患が24％，悪性新生物が23％，神経系疾患が21％と多く，この三者で68％を占めた。転帰では49％が治癒，34％が治癒しないまま退院，17％が治癒しないまま死亡していた。褥瘡の重症度と原疾患，年齢，転帰を比較したところ，いずれの原疾患，年齢，転帰においても初診時のDESIGN-R値に有意差はみられなかった。非治癒死亡例において初診日と最終診察日のDESIGN-R総和に有意差は認められなかった。以上の観察から当院の疾患群においては初診時の褥瘡の重症度は患者の転帰を予測する要因と成らず，またその推移は患者の生命予後と関係しない可能性が考えられた。
During the coronavirus disease 2019 (COVID-19) pandemic, rapid and accurate triage of patients at emergency department is critical to inform decision-making. We propose a data-driven approach for automatic prediction deterioration risk using deep neural network that learns from chest X-ray images gradient boosting model routine clinical variables. Our AI prognosis system, trained data 3,661 patients, achieves an area under receiver operating characteristic curve (AUC) 0.786 (95% CI:...
Deep neural networks (DNNs) show promise in breast cancer screening, but their robustness to input perturbations must be better understood before they can clinically implemented. There exists extensive literature on this subject the context of natural images that potentially built upon. However, it cannot assumed conclusions about will transfer from mammogram images, due significant differences between two image modalities. In order determine whether transfer, we measure sensitivity a...
Background: Experimental studies have shown that enhanced regional heterogeneity of repolarization and autonomic imbalances due to acute myocardial ischemia are associated with the development sustained ventricular tachycardia fibrillation (VT/VF). We examined prognostic role instability heart rate variability in coronary syndrome (ACS) patients a left ejection fraction >35%.Methods: prospectively enrolled 572 ACS admitted two hospitals between April 2012 March 2013. The was assessed by...