38 research outputs found
Predicting far-infrared maps of galaxies via machine learning techniques
No full textContext. The ultraviolet (UV) to sub-millimetre spectral energy distribution of galaxies can be roughly divided into two sections: the stellar emission (attenuated by dust) at UV to near-infrared wavelengths and dust emission at longer wavelengths. In Dobbels et al. (2020, A&A, 634, A57), we show that these two sections are strongly related, and we can predict the global dust properties from the integrated UV to mid-infrared emission with the help of machine learning techniques.
Aims. We investigate if these machine learning techniques can also be extended to resolved scales. Our aim is to predict resolved maps of the specific dust luminosity, specific dust mass, and dust temperature starting from a set of surface brightness images from UV to mid-infrared wavelengths.
Methods. We used a selection of nearby galaxies retrieved from the DustPedia sample, in addition to M31 and M33. These were convolved and resampled to a range of pixel sizes, ranging from 150 pc to 3 kpc. We trained a random forest model which considers each pixel individually.
Results. We find that the predictions work well on resolved scales, with the dust mass and temperature having a similar root mean square error as on global scales (0.32 dex and 3.15 K on 18″ scales respectively), and the dust luminosity being noticeably better (0.11 dex). We find no significant dependence on the pixel scale. Predictions on individual galaxies can be biased, and we find that about two-thirds of the scatter can be attributed to scatter between galaxies (rather than within galaxies).
Conclusions. A machine learning approach can be used to create dust maps, with its resolution being only limited to the input bands, thus achieving a higher resolution than Herschel. These dust maps can be used to improve global estimates of dust properties, they can lead to a better estimate of dust attenuation, and they can be used as a constraint on cosmological simulations that trace dust
An applied deep learning approach for estimating soybean relative maturity from UAV imagery to aid plant breeding decisions
No full textFor a global breeding organization, identifying the next generation of superior crops is vital for its success. Recognizing new genetic varieties requires years of in-field testing to gather data about the crop’s yield, pest resistance, heat resistance, etc. At the conclusion of the growing season, organizations need to determine which varieties will be advanced to the next growing season (or sold to farmers) and which ones will be discarded from the candidate pool. Specifically for soybeans, identifying their relative maturity is a vital piece of information used for advancement decisions. However, this trait needs to be physically observed, and there are resource limitations (time, money, etc.) that bottleneck the data collection process. To combat this, breeding organizations are moving towards advanced image capturing devices. In this paper, we develop a robust and automatic approach for estimating the relative maturity of soybeans using a time series of UAV images. An end-to-end hybrid model combining Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) is proposed to extract features and capture the sequential behavior of time series data. The proposed deep learning model was tested on six different environments across the United States Results suggest the effectiveness of our proposed CNN-LSTM model compared to the local regression method. Furthermore, we demonstrate how this newfound information can be used to aid in plant breeding advancement decisions.This article is published as Moeinizade, Saba, Hieu Pham, Ye Han, Austin Dobbels, and Guiping Hu. "An applied deep learning approach for estimating soybean relative maturity from UAV imagery to aid plant breeding decisions." Machine Learning with Applications 7 (2022): 100233.
DOI: 10.1016/j.mlwa.2021.100233
Copyright 2021 The Author(s).
Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).
Posted with permission
Dust and stellar property estimates via machine learning techniques
No full textLarge surveys have been performed from the ultraviolet (UV) to the far-infrared (FIR). Some galaxies are observed over this whole wavelength range, and through SED fitting we get accurate estimates of their stellar and dust properties. Unfortunately, most galaxies are only detected at a limited part of this spectrum. With machine learning techniques, we can use the UV-FIR galaxies as a blueprint: we learn the mapping from their fluxes to their properties. For example, a mapping from UV-NIR to dust mass can be established, and then applied to galaxies that lack FIR data. We present this approach using DustPedia and H-ATLAS data, and show the superiority over energy balance SED fitting. Besides what can be directly estimated from the SEDs alone, this technique implicitly uses relations that follow from galaxy evolution. To avoid a black box, we take special care to estimate uncertainties on our predictions and to interpret the model.</p
« Littéradanse », Quand la danse contemporaine s’empare du texte littéraire
No full textL’ouvrage de Mélanie Mesager, publié en 2018 et intitulé Littéradanse, Quand la chorégraphie s’empare du texte littéraire. Fanny de Chaillé, Daniel Dobbels, Antoine Dufeu et Jonah Bokaer est issu d’un travail mené en 2015 dans le cadre d’un Master 2 dirigé par Julie Perrin à l’Université Paris 8. Il s’agit d’une étude menée à partir de quatre spectacles de danse contemporaine à la croisée du mouvement dansé et du texte littéraire. L’auteur entend ici donner corps au concept de « littéradanse » en s’attachant à des œuvres chorégraphiques qui ne se réduisent ni au texte énoncé ni à la partition dansée, mais qui naissent de leur mise en présence. Elle propose plusieurs façons d’appréhender les inflexions réciproques de la littérature et de la danse, tout en cherchant à décrypter comment un sens mouvant se construit dans le temps du spectacle.Mélanie Mesager’s book, published in 2018 and entitled Littéradanse, Quand la chorégraphie s’empare du texte littéraire. Fanny de Chaillé, Daniel Dobbels, Antoine Dufeu et Jonah Bokaer is the result of a work carried out in 2015 as part of a Master 2 program led by Julie Perrin at the University of Paris 8. This is a study based on four contemporary dance performances at the crossroads of dance and literary text. The author intends here to give substance to the concept of “littéradanse” by focusing on choreographic works that are not to be reduced to the text, nor to the dance performed. She proposes several ways of understanding the reciprocal inflections of literature and dance, while seeking to figure out how a moving meaning is constructed during the performance
Advances in ITP - therapy and quality of life - a patient survey.
No full textCurrent guidelines recommend glucocorticoids and splenectomy as standard 1(st) and 2(nd) line treatments for chronic immune thrombocytopenia (ITP). We sought to find out how German ITP-patients are treated with respect to these guidelines. Members of a patient support association ≥18 years with a self-reported history of chronic ITP>12 months were surveyed with a web-based questionnaire. 122 questionnaires were evaluated. 70% of patients had chronic ITP for more than 5 years and 20% an average platelet count of ≤30·10(9)/L. 41% of the patients reported haematomas or petechiae more than once or twice and up to 12 times or more per year and 17% oropharyngeal and nasal bleeds. 11% had been admitted to hospital during the last 12 months. 88% had received or currently receive glucocorticoids, 27% were splenectomised. IVIG had been given to 55%, rituximab to 22%, anti-D to 12%, ciclosporin to 7%, while complementary and alternative medical treatments had been used by 36%. 50 women responded to questions concerning pregnancy. 14 (28%) had been advised not to become pregnant. 23 reported pregnancies and 10 (44%) required treatment for their ITP during pregnancy. Glucocorticoids are the most common therapy for chronic ITP but complementary and alternative treatments already come second and less than ⅓ of patients are splenectomised. This and the frequent use of complementary medicines suggests patients' dissatisfaction with conventional approaches. Many patients receive off-label therapies. There is a major need for adequate counselling and care for pregnant ITP-patients
Patient self-management in kidney transplantation. definition, measurement, and intervention
No full textWhile one-year graft survival rates for deceased donor transplants have soared from about 40% in 1975 to more than 90% in 2005 [1], the long-term perspective has changed very little. From 1996-2005, 10-year deceased donor graft survival has remained at about 40%, only slightly above that of the 1987-1995 period [1]. Furthermore, the gain in graft survival between 1988 and 1995, based on calculated real half-lives, has been reported as 4.7 or 8.4 months, for first or further deceased donor transplants, respectively. These numbers reveal that estimates of doubled half-lives rom 1988 to 1995, which were based on projected half-lives, were far from accurate [2]. The remarkable short-term improvements have thus not translated into long-term advantages [1, 2]. Improving long-term post-transplantation outcomes should therefore be a priority of transplant recipient management. Investing in chronic illness management, which focuses on improving patient self-management and medication adherence, is a promising pathway in that direction. Chronic illness management has lately emerged as a response to the reported dramatic global increase in chronic conditions [3]. A chronic condition is defined as one that is never completely cured [4] and that requires ongoing long-term management of the illness, coexisting morbidities, treatments, or measures to prevent further disability [3]. Such management imposes a heavy burden on current health care systems. The gravity of the situation is increased by the application of acute care models (i.e., prioritizing the treatment and cure of peoples’ acute and urgent symptoms), which have limited effects on chronic conditions [3]. Effective chronic care models, i.e., care that improves chronically ill patient populations’ conditions, are characterized by continuity of care, partnership with patients, families, and communities, support for patients in improving self-management skills, attention to preventive measures, decision-making support for healthcare providers, and availability of clinical information systems [3, 5-7]. Empirical evidence underlines the effectiveness of chronic illness management [8-10]. Of these, models that incorporate patient self-management support show the most improved outcomes [3, 7, 11]. Patient self-management refers to actions performed by patients for themselves in daily life to manage their illness and treatment, and to avoid health deterioration [5, 12]. Related support consists of two components: the training of disease specific knowledge and technical skills, and the training of non-disease specific problem solving and other skills to assist behavior change [13]. A growing body of evidence in patients with chronic illness demonstrates that supporting patient self-management positively impacts outcomes [10, 14-19]. An essential component of patient self-management is managing the medical regimen, including adherence, i.e., “the extent to which a person’s behavior (taking medications, following a recommended diet, and/or executing lifestyle changes) corresponds with the agreed recommendations of a healthcare provider” [8]. The scale and impact of medication adherence regarding patient outcomes have been widely demonstrated in chronic patient populations [20-22]. Recent literature reviews regarding kidney transplantation [23-25] demonstrate that non-adherence to immunosuppressive therapy is a major contributor to poor clinical outcomes. Given that inadequate medication adherence has critical implications on health outcomes, focusing prominently on adherence as an essential part of patient self-management is crucial to improve outcomes in the kidney transplant population.
The gaps in the literature guiding this research program were as follow: 1) as no conceptualization was available for patient self-management in the kidney transplant population, it was necessary to define one; 2) little information was available on the diagnostic accuracy of measurement methods to identify medication non-adherence in the kidney transplant population, 3) there was a need to test medication adherence enhancing interventions, as very little information was available on this patient group; and 4) there was a need to evaluate a technological intervention designed for patient use, as such information was lacking.
The work and studies incorporated in this research program to address these gaps used a variety of methods, including both quantitative and qualitative approaches. The studies are summarized as follows. First, a comprehensive definition of kidney transplant recipient self-management has been developed, summarizing evidence from the transplant literature. This definition provides both detailed kidney transplant specific self-management activities and core skills that patients may acquire or further develop for successful self-management. It also provides a conceptual model using a care paradigm that regards the patient as a worker having expertise at managing the illness in daily life. This is a crucial aspect of chronic illness management. The model outlined here can be used as a basis for the development of systematic and comprehensive kidney transplant recipient self-management support. It furthermore constitutes a crucial first step to allow transplant clinics to shift from an acute to a chronic care model for long-term transplant recipient management. Second, the literature summarized current understanding about medication nonadherence, and provided an overview of current knowledge regarding correlates of medication non-adherence, as well as of medication adherence enhancing interventions in the kidney transplant population. Further, to offer a concrete example on how to implement theory based adherence enhancing strategies into an individual situation, it reports on a case study [26]. Third, we used a cross sectional study to test the diagnostic accuracy of immunosuppression assay, patients’ self-reports, clinicians’ collateral reports, and constructed composite adherence scores using electronic monitoring as a reference standard for a convenience sample of 249 kidney transplant recipients (female: 43.4%; mean age 53.6 (SD: 12.7), median 7 years (IQR: 9 years) post-transplantation). Medication non-adherence prevalence, as assessed by electronic monitoring, was 17.3%. Across the measurement methods, prevalence rates varied from 12.4% for self-reports to 38.9% for composite adherence scores. Of all the measures, the composite adherence score yielded both the highest sensitivity (72.1%) and the highest likelihood ratio of a positive test (2.74), while collateral reports of at least three clinicians showed the highest specificity (93.1%). While no measures showed high sensitivity alongside high specificity, combining measures increased diagnostic accuracy, indicating the relevance of combined measures for clinical and research purposes [27]. Fourth, we tested the efficacy of an educational/behavioral intervention and enhanced social support intervention to increase medication adherence in 18 non-adherent renal transplant recipients (age: 45.6±1.2 yr; 78.6% male). Using a pilot randomized controlled trial, the study showed a remarkable decrease in non-adherence in the intervention group (IG, n=6) and in the enhanced standard care group (EUCG, n=12) over the first three months (IG, χ2 =3.97, df=1, p=.04; EUCG, χ2=3.40, df=1, p=.06). The interventions appeared to add further benefit to medication adherence levels in the IG, as the greatest decrease in non-adherence was observed there. This result was not, however, statistically significant (at 90 days:, χ2=1.05, df=1, p=.31), owing to insufficient sample size [28]. Fifth, we tested the content validity and usability of a computer based patient information and education tool (OTISTM), from the perspectives of clinicians and patients. Using qualitative methods and a purposive sample of 8 clinicians and 14 patients, the study identified deviations from current medical practice regarding the content, language, and information structure of OTISTM. Seven of the eight clinicians rated OTISTM as nonrelevant for implementation in clinical practice and all patients encountered usability problems, mostly regarding the program’s interface. Emerging categories from the patients’ perspectives vis à vis content were knowledge acquisition, illness management, and partnership forming. The study demonstrated the need to establish the presented material’s content validity and usability by involving clinicians and patients well before its clinical implementation phase [29]. The results of our research program contribute in five main ways to the evidence base regarding kidney transplant recipients’ self-management, and, more specifically, adherence to post-transplantation medication taking. First, it described, for the first time, a comprehensive kidney transplant recipient self-management model, outlining disease specific activities and non-disease specific patient core skills. Second, it summarized knowledge on current understanding, correlates of medication adherence, and posttransplant adherence enhancing interventions. Third, it added detailed knowledge on diagnostic accuracy of state-of-the art measures to identify medication non-adherence in renal transplant recipients. Fourth, it provided evidence and thus added to the very limited amount of available information, supporting the feasibility of enhancing medication adherence in non-adherent renal transplant recipients using a package of educationalbehavioral interventions and social support. Finally, it suggested that in order to ensure and maximize benefits to its intended users, technological interventions for patient use need to be evaluated with regard to usability and content validity. Future research should focus on further development and testing of the conceptual model presented here, with attention to relationships between the model variables, to develop and evaluate valid kidney transplant recipient self-management measures, and to test whether supporting such self-management results in improved long-term health outcomes
Morphology-assisted galaxy mass-to-light predictions using deep learning
No full textThis record is for a(n) offprint of an article published in Astronomy & Astrophysics on 2019-04-18; the version of record is available at https://doi.org/10.1051/0004-6361/201834575.Context. One of the most important properties of a galaxy is the total stellar mass, or equivalently the stellar mass-to-light ratio (M/L). It is not directly observable, but can be estimated from stellar population synthesis. Currently, a galaxy’s M/L is typically estimated from global fluxes. For example, a single global g − i colour correlates well with the stellar M/L. Spectral energy distribution (SED) fitting can make use of all available fluxes and their errors to make a Bayesian estimate of the M/L. Aims. We want to investigate the possibility of using morphology information to assist predictions of M/L. Our first goal is to develop and train a method that only requires a g-band image and redshift as input. This will allows us to study the correlation between M/L and morphology. Next, we can also include the i-band flux, and determine if morphology provides additional constraints compared to a method that only uses g- and i-band fluxes. Methods. We used a machine learning pipeline that can be split in two steps. First, we detected morphology features with a convolutional neural network. These are then combined with redshift, pixel size and g-band luminosity features in a gradient boosting machine. Our training target was the M/L acquired from the GALEX-SDSS-WISE Legacy Catalog, which uses global SED fitting and contains galaxies with z ∼ 0.1. Results. Morphology is a useful attribute when no colour information is available, but can not outperform colour methods on its own. When we combine the morphology features with global g- and i-band luminosities, we find an improved estimate compared to a model which does not make use of morphology. Conclusions. While our method was trained to reproduce global SED fitted M/L, galaxy morphology gives us an important additional constraint when using one or two bands. Our framework can be extended to other problems to make use of morphological information.offprin
Dynamische grids in Monte Carlo stralingsoverdracht. /
No full textMaster of Science in de fysica en de sterrenkund
Estimating metallicity using machine learning /
No full textMaster of Science in de fysica en de sterrenkund
