RESUMO
Convolutional neural network (CNN) has promoted the development of diagnosis technology of medical images. However, the performance of CNN is limited by insufficient feature information and inaccurate attention weight. Previous works have improved the accuracy and speed of CNN but ignored the uncertainty of the prediction, that is to say, uncertainty of CNN has not received enough attention. Therefore, it is still a great challenge for extracting effective features and uncertainty quantification of medical deep learning models In order to solve the above problems, this paper proposes a novel convolutional neural network model named DM-CNN, which mainly contains the four proposed sub-modules : dynamic multi-scale feature fusion module (DMFF), hierarchical dynamic uncertainty quantifies attention (HDUQ-Attention) and multi-scale fusion pooling method (MF Pooling) and multi-objective loss (MO loss). DMFF select different convolution kernels according to the feature maps at different levels, extract different-scale feature information, and make the feature information of each layer have stronger representation ability for information fusion HDUQ-Attention includes a tuning block that adjust the attention weight according to the different information of each layer, and a Monte-Carlo (MC) dropout structure for quantifying uncertainty MF Pooling is a pooling method designed for multi-scale models, which can speed up the calculation and prevent overfitting while retaining the main important information Because the number of parameters in the backbone part of DM-CNN is different from other modules, MO loss is proposed, which has a fast optimization speed and good classification effect DM-CNN conducts experiments on publicly available datasets in four areas of medicine (Dermatology, Histopathology, Respirology, Ophthalmology), achieving state-of-the-art classification performance on all datasets. DM-CNN can not only maintain excellent performance, but also solve the problem of quantification of uncertainty, which is a very important task for the medical field. The code is available: https://github.com/QIANXIN22/DM-CNN.
Assuntos
Medicina , Redes Neurais de Computação , Incerteza , Algoritmos , Método de Monte CarloRESUMO
Many countries worldwide are aging rapidly, and the complex care needs of older adults generate an unprecedented demand for health services. Common reasons for elderly emergency department (ED) visits frequently involve conditions triggered by preventable infections also known as ambulatory care sensitive conditions (ACSCs). This study aims to describe the trend and the associated disease burden attributable to ACSC-related ED visits made by elderly patients and to characterize their ED use by nursing home residence. We designed a population-based ecological study using administrative data on Taiwan EDs between 2002 and 2013. A total of 563,647 ED visits from individuals aged 65 or over were examined. All elderly ED visits due to ACSCs (tuberculosis, upper respiratory infection, pneumonia, sepsis, cellulitis and urinary tract infection (UTI)) were further identified. Subsequent hospital admissions, related deaths after discharge, total health care costs and disability-adjusted life years (DALYs) were compared among different ACSCs. Prevalence of ACSCs was then assessed between nursing home (NH) residents and non-NH residents. Within the 12-year observation period, we find that there was a steady increase in both the rate of ACSC ED visits and the proportion of elderly with a visit. Overall, pneumonia is the most prevalent among six ACSCs for elderly ED visits (2.10%; 2.06 to 2.14), subsequent hospital admissions (5.77%; 5.59 to 5.94) and associated mortality following admission (17.37%; 16.74 to 18.01). UTI is the second prevalent ACSC consistently across ED visits (2.02%; 1.98 to 2.05), subsequent hospital admissions (2.36%, 2.25 to 2.48) and mortality following admission (10.80%; 10.28 to 11.32). Sepsis ranks third highest in the proportion of hospitalization following ED visit (2.29%; 2.18 to 2.41) and related deaths after hospital discharge (7.39%; 6.95 to 7.83), but it accounts for the highest average total health care expenditure (NT$94,595 ± 120,239; ≈US$3185.02) per case. When examining the likelihood of ACSC-attributable ED use, significantly higher odds were observed in NH residents as compared with non-NH residents for: pneumonia (adjusted odds ratio (aOR): 5.01, 95% confidence interval (CI) 4.50-5.58); UTI (aOR: 4.44, 95% CI 3.97-4.98); sepsis (aOR: 3.54, 95% CI 3.06-4.10); and tuberculosis (aOR: 2.44, 95% CI 1.63-3.65). Here we examined the ACSC-related ED care and found that, among the six ACSCs studied, pneumonia, UTI and sepsis were the leading causes of ED visits, subsequent hospital admissions, related mortality, health care costs and DALYs in Taiwanese NH elderly adults. Our findings suggest that efficient monitoring and reinforcing of quality of care in the residential and community setting might substantially reduce the number of preventable elderly ED visits and alleviate strain on the health care system.
Assuntos
Assistência Ambulatorial , Efeitos Psicossociais da Doença , Serviço Hospitalar de Emergência/economia , Hospitalização/economia , Aceitação pelo Paciente de Cuidados de Saúde , Idoso , Idoso de 80 Anos ou mais , Celulite (Flegmão)/terapia , Feminino , Humanos , Masculino , Pneumonia/terapia , Infecções Respiratórias/terapia , Estudos Retrospectivos , Taiwan , Tuberculose/terapia , Infecções Urinárias/terapiaRESUMO
OBJECTIVES: To evaluate the cost-effectiveness of sorafenib treatment in combination with other therapies versus sorafenib monotherapy among patients with advanced hepatocellular carcinoma (HCC) who are enrolled in Taiwan's National Health Insurance. METHODS: A Markov model was constructed to simulate treatment outcomes and direct medical costs of sorafenib combination therapy and monotherapy from the perspective of the healthcare payer in Taiwan. Both life-years (LYs) and quality-adjusted life-years (QALYs) were used to measure treatment outcomes, and all costs were expressed in 2014 New Taiwan dollars (NT$). Model parameters were acquired primarily using data from population-based administrative databases: the Cancer Registry, National Health Insurance Research Database, and the Death Registry. Willingness-to-pay (WTP) threshold was set at three times the per capita gross domestic product at NT$2,133,930. Deterministic and probabilistic sensitivity analyses were conducted. RESULTS: For advanced HCC patients, sorafenib combined with other treatments might not be a cost-effective option when compared with sorafenib therapy alone. In the base-case analysis, combination treatment with sorafenib was estimated to increase costs by NT$434,788 compared with monotherapy, with a gain of 0.1595 QALYs. The resulting incremental cost-effectiveness ratio (ICER) was NT$2,725,943 per QALY gained. Results were sensitive to health utility values and monthly costs accrued in the progression-free survival state of the combination therapy group. CONCLUSIONS: Our evidence from Taiwan demonstrated that while sorafenib in combination with other therapeutic approaches might improve treatment outcome when compared with sorafenib monotherapy, its ICER exceeded the WTP threshold and was considered not cost-effective.
Assuntos
Antineoplásicos/uso terapêutico , Carcinoma Hepatocelular/tratamento farmacológico , Terapia Combinada , Análise Custo-Benefício , Neoplasias Hepáticas/tratamento farmacológico , Niacinamida/análogos & derivados , Compostos de Fenilureia/uso terapêutico , Intervalo Livre de Doença , Feminino , Humanos , Masculino , Pessoa de Meia-Idade , Modelos Estatísticos , Niacinamida/uso terapêutico , Anos de Vida Ajustados por Qualidade de Vida , Sorafenibe , TaiwanRESUMO
Spatial information plays a critical role in remote sensing and mapping applications such as environment surveying and disaster monitoring. An Unmanned Aerial Vehicle (UAV)-borne mobile mapping system (MMS) can accomplish rapid spatial information acquisition under limited sky conditions with better mobility and flexibility than other means. This study proposes a long endurance Direct Geo-referencing (DG)-based fixed-wing UAV photogrammetric platform and two DG modules that each use different commercial Micro-Electro Mechanical Systems' (MEMS) tactical grade Inertial Measurement Units (IMUs). Furthermore, this study develops a novel kinematic calibration method which includes lever arms, boresight angles and camera shutter delay to improve positioning accuracy. The new calibration method is then compared with the traditional calibration approach. The results show that the accuracy of the DG can be significantly improved by flying at a lower altitude using the new higher specification hardware. The new proposed method improves the accuracy of DG by about 20%. The preliminary results show that two-dimensional (2D) horizontal DG positioning accuracy is around 5.8 m at a flight height of 300 m using the newly designed tactical grade integrated Positioning and Orientation System (POS). The positioning accuracy in three-dimensions (3D) is less than 8 m.