The artificial intelligence revolution has been spurred forward by the availability of large-scale datasets. In contrast, the paucity of large-scale medical datasets hinders the application of machine learning in healthcare. The lack of publicly available multi-centric and diverse datasets mainly stems from confidentiality and privacy concerns around sharing medical data. To demonstrate a feasible path forward in medical image imaging, we conduct a case study of applying a differentially private federated learning framework for analysis of histopathology images, the largest and perhaps most complex medical images. We study the effects of IID and non-IID distributions along with the number of healthcare providers, i.e., hospitals and clinics, and the individual dataset sizes, using The Cancer Genome Atlas (TCGA) dataset, a public repository, to simulate a distributed environment. We empirically compare the performance of private, distributed training to conventional training and demonstrate that distributed training can achieve similar performance with strong privacy guarantees. We also study the effect of different source domains for histopathology images by evaluating the performance using external validation. Our work indicates that differentially private federated learning is a viable and reliable framework for the collaborative development of machine learning models in medical image analysis.
The artificial intelligence revolution has been spurred forward by the availability of large-scale datasets. In contrast, the paucity of large-scale medical datasets hinders the application of machine learning in healthcare. The lack of publicly available multi-centric and diverse datasets mainly stems from confidentiality and privacy concerns around sharing medical data. To demonstrate a feasible path forward in medical image imaging, we conduct a case study of applying a differentially private federated learning framework for analysis of histopathology images, the largest and perhaps most complex medical images. We study the effects of IID and non-IID distributions along with the number of healthcare providers, i.e., hospitals and clinics, and the individual dataset sizes, using The Cancer Genome Atlas (TCGA) dataset, a public repository, to simulate a distributed environment. We empirically compare the performance of private, distributed training to conventional training and demonstrate that distributed training can achieve similar performance with strong privacy guarantees. We also study the effect of different source domains for histopathology images by evaluating the performance using external validation. Our work indicates that differentially private federated learning is a viable and reliable framework for the collaborative development of machine learning models in medical image analysis.
The research aims to study the impact of the change in the market value resulting from the increase in the capital of a sample of banks listed in the Iraq Stock Exchange. Which imposed a capital increase by the Central Bank of Iraq to increase the capital efficiency of those banks. The research attempts to shed light on the concepts associated with the research variables. Theoretically, by reviewing the concepts and topics related to capital increase and the market value of stocks. Applied by measuring the correlation and regression with statistical significance between the two research variables, which contributes to verify the research hypotheses. Data were collected from the Iraq Stock Exchange and the Iraq Securities Commission. The research community is represented by the listed bank sector, whose stocks were traded on the Iraq Stock Exchange, and they amounted to (41) banks. The research sample identified (7) listed banks that did not record any missing data during the research period, with a number of (48) observations from the beginning of 2007 to the end of 2018 quarterly, with four observations per year. SPSS program was used in statistical analysis and Microsoft Excel program to collect data, format it and represent it in graphical forms. The most important results of the research concluded on the effect of the increase in capital on the market value of the stocks of the research sample in a positive direction. However, this does not agree with the research hypothesis that the increase in capital leads to a decrease in the market value. The most important recommendations concluded that it is necessary to use more effective measures and models for financial research because of the inaccuracy of the traditional results in measuring the large fluctuations that occur in the financial data.
Introduction Wide experimental evidence has proved Alpha 1 antitrypsin, which is released from the liver, to protect lung tissue from damage during inflammation. There is experimental evidence also suggests such protection is extended to myocardial tissues during myocardial ischemia/infarction. Objective The aim of the study was to determine the levels and the time course of A1AT in the plasma of myocardial infarction patients over 96 hours and compare it with healthy subject as control. Methods Blood samples were collected from patients (n=40) arriving at emergency department at 1, 4, 24, 48 and 96 hrs after diagnosis and admission to King Salman Cardiac Center, in King Fahad Medical City in Riyadh, Saudi Arabia. Blood samples were collected in purple top tubes containing EDTA to prevent blood coagulation. Blood plasma was separated by centrifugation with in 3 hrs of collection. Plasma was decanted and stored until analyzed. Alpha 1 antitrypsin was determined in all patients’ blood samples using human Eliza kits( abcam USA). In addition control blood samples were collected from 20 healthy individuals. Results A1AT levels were significantly less in MI patients (1025±77 ug/ml at 96 hrs, P<001) compared to (1655±128 ug/ml) (mean ± SEM) in controls. It was very interesting to find those patients win whom the decrease in A1AT was >50 % of control values, they did not survive and passed away (deceased) (673+71 ug/ml, n=3). Also the levels of A1AT after an MI attack was comparable and not significant in hypertensive compared to non‐hypertensive or in STEMI and Non‐STEMI patients but was significantly (P<0.03) higher in non‐smokers (1457+163, n=27) compared to smokers (1153±155, n=10). Conclusion The plasma level of A1AT is suggested to be important in providing some protection to myocardial cells during an MI attack. A1AT levels seem to drop shortly following an MI attack. The drop in the level is larger in smokers. In those patients in whom the level of A1AT drops to more about 50 % of normal values they are less likely to survive the MI attack. More studies are needed to confirm the above findings. Support or Funding Information We acknowledge the financial support from the research center at King Fahad Medical City, Riyadh, Saudi Arabia
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.