Refine
Year of publication
Document Type
- Article (35)
- Conference Proceeding (9)
- Preprint (4)
Has Fulltext
- yes (48)
Is part of the Bibliography
- no (48)
Keywords
- Clinical decision support systems (2)
- Computer-assisted diagnosis (2)
- Rare diseases (2)
- artificial intelligence (2)
- healthcare (2)
- trustworthy AI (2)
- ACURATE neo (1)
- ALK gene (1)
- BCOR (1)
- Behavior (1)
Institute
Artificial Intelligence (AI) has the potential to greatly improve the delivery of healthcare and other services that advance population health and wellbeing. However, the use of AI in healthcare also brings potential risks that may cause unintended harm. To guide future developments in AI, the High-Level Expert Group on AI set up by the European Commission (EC), recently published ethics guidelines for what it terms “trustworthy” AI. These guidelines are aimed at a variety of stakeholders, especially guiding practitioners toward more ethical and more robust applications of AI. In line with efforts of the EC, AI ethics scholarship focuses increasingly on converting abstract principles into actionable recommendations. However, the interpretation, relevance, and implementation of trustworthy AI depend on the domain and the context in which the AI system is used. The main contribution of this paper is to demonstrate how to use the general AI HLEG trustworthy AI guidelines in practice in the healthcare domain. To this end, we present a best practice of assessing the use of machine learning as a supportive tool to recognize cardiac arrest in emergency calls. The AI system under assessment is currently in use in the city of Copenhagen in Denmark. The assessment is accomplished by an independent team composed of philosophers, policy makers, social scientists, technical, legal, and medical experts. By leveraging an interdisciplinary team, we aim to expose the complex trade-offs and the necessity for such thorough human review when tackling socio-technical applications of AI in healthcare. For the assessment, we use a process to assess trustworthy AI, called 1Z-Inspection® to identify specific challenges and potential ethical trade-offs when we consider AI in practice.
At the Institute for Applied Physics (IAP), University of Frankfurt, a s.c. 325 MHz CH-Cavity is under development for future beam tests at GSI UNILAC, Darmstadt. The cavity with 7 accelerating cells has a geometrical beta of 0.15 corresponding to 11.4 AMeV. The design gradient is 5 MV/m. The geometry of this resonator was optimized with respect to a compact design, low peak fields, surface processing, power coupling and tuning. Furthermore a new tuning system based on bellow tuners inside the resonator will control the frequency during operation. After rf tests in Frankfurt the cavity will be tested with a 10 mA, 11.4 AMeV beam delivered by the GSI UNILAC. In this paper rf simulations, multipacting analysis as well as thermal calculations will be presented.
At GSI a new, superconducting (sc) continuous wave (cw) LINAC is under design in cooperation with the Institute for Applied Physics (IAP) of Frankfurt University and the Helmholtz Institut Mainz (HIM). This proposed LINAC is highly requested by a broad community of future users to fulfill the requirements of nuclear chemistry, nuclear physics, and especially in the research field of Super Heavy Elements (SHE). In this context the preliminary layout of the LINAC has been carried out by IAP. The main acceleration of up to 7.3 AMeV will be provided by nine sc Crossbar-H-mode (CH) cavities operated at 217 MHz. Currently, a prototype of the cw LINAC as a demonstrator is under development. The demonstrator comprises a sc CH-cavity embedded between two sc solenoids mounted in a horizontal cryomodule. A full performance test of the demonstrator in 2013/14 by injecting and accelerating a beam from the GSI High Charge Injector (HLI) is one important milestone of the project. The status of the demonstrator is presented.
The superconducting CH-structure (Crossbar-H-mode) is a multi-cell drift tube cavity for the low and medium energy range operated in the H21-mode, which has been developed at the Institute for Applied Physics (IAP) of Frankfurt University. With respect to different high power applications two types of superconducting CH-structures (f = 325 MHz, β = 0.16, seven cells and f = 217 MHz, β = 0.059, 15 cells) are presently under construction and accordingly under development. The structural mechanical simulation is a very important aspect of the cavity design. Furthermore, several simulations with ANSYS Workbench have been performed to predict the deformation of the cavity walls due to the cavity cool-down, pressure effects and mechanical vibrations. To readjust the fast frequency changes in consequence of the cavity shape deformation, a new concept for the dynamic frequency tuning has been investigated, including a novel type of bellow-tuner.
The ARMADILLO bunch compressor currently being designed at IAP is capable of reaching a longitudinal pulse compression ratio of 45 for proton beams of 150 mA at 2 MeV. It will provide one nanosecond proton pulses with a peak current of 7.7 A. The system guides nine linacμbunches deflected by a 5 MHz rf kicker and uses four dipole magnets - two homogeneous and two with field gradients - to merge them on the target. For longitudinal focusing and an energy variation of ±200 keV two multitrack rf cavities are included. ARMADILLO will be installed at the end of the Frankfurt Neutron Source FRANZ making use of the unique 250 kHz time structure. This contribution will provide an overview of the layout of the system as well as recent advances in component design and beam dynamics of the compressor.
This novel kind of neutron beam facility will provide 1 ns short neutron pulses with an approximately thermal energy distribution around 30 keV. The pulse repetition rate will be up to 250 kHz, the total proton number per pulse will be up to 6×1010 in the final stage, starting with a p – source current of 200 mA. A second target station will allow n – activation experiments by cw beam operation. An intense 2 MeV proton beam will drive a neutron source by the 7 Li (p,n) 7 Be reaction. The facility is under construction at the physics experimental hall of the J.W. Goethe – University. The 1m thick concrete tunnel was installed in 2009. In 2011 all rf amplifiers will be delivered and installed. Successful 200 mA proton source experiments in 2010 at a test stand will be followed by experiments on the 120 kV FRANZ terminal in 2011. The 250 kHz, 100 ns chopper in front of the rf linac is under construction, while the 2 MeV bunch compressor design was finished and the technical design of all components has started. The main accelerator cavity is under construction. First 2 MeV beam tests are expected for end of 2012.
For the research program with cooled antiprotons at FAIR a dedicated 70 MeV, 70 mA proton injector is required. The main acceleration of this room temperature linac will be provided by six CH cavities operated at 325 MHz. Each cavity will be powered by a 2.5 MW Klystron. For the second acceleration unit from 11.5 MeV to 24.2 MeV a 1:2 scaled model has been built. Low level RF measurements have been performed to determine the main parameters and to prove the concept of coupled CH cavities. For this second tank technical and mechanical investigations have been performed in 2010 to develop a complete technical concept for the manufacturing. In Spring 2011, the construction of the first power prototype has started. The main components of this cavity will be ready for measurements in summer 2011. At that time, the cavity will be tested with a preliminary aluminum drift tube structure, which will allow precise frequency and field tuning. This paper will report on the recent technical development and achievements. It will outline the main fabrication steps towards that novel type of proton DTL. Also first low level RF measurements are expected.
Background: Body dysmorphic disorder (BDD) is characterized by an excessive preoccupation with one or more perceived flaws in one’s own appearance. Previous studies provided evidence for deficits in configural and holistic processing in BDD. Preliminary evidence suggests abnormalities at an early stage of visual processing. The present study is the first examining early neurocognitive perception of the own face in BDD by using electroencephalography (EEG). We investigated the face inversion effect, in which inverted (upside-down) faces are disproportionately poorly processed compared to upright faces. This effect reflects a disruption of configural and holistic processing, and in consequence a preponderance of featural face processing.
Methods: We recorded face-sensitive event-related potentials (ERPs) in 16 BDD patients and 16 healthy controls, all unmedicated. Participants viewed upright and inverted (upside-down) images of their own face and an unfamiliar other face, each in two facial emotional expressions (neutral vs. smiling). We calculated the early ERP components P100, N170, P200, N250, and the late positive component (LPC), and compared amplitudes among both groups.
Results: In the early P100, no face inversion effects were found in both groups. In the N170, both groups exhibited the common face inversion effects, with significantly larger N170 amplitudes for inverted than upright faces. In the P200, both groups exhibited larger inversion effects to other (relative to own) faces, with larger P200 amplitudes for other upright than inverted faces. In the N250, no significant group differences were found in face processing. In the LPC, both groups exhibited larger inversion effects to other (relative to own) faces, with larger LPC amplitudes for other inverted than upright faces. These overall patterns appeared to be comparable for both groups. Smaller inversion effects to own (relative to other) faces were observed in none of these components in BDD, relative to controls.
Conclusions: The findings suggest no evidence for abnormalities at all levels of early face processing in our observed sample of BDD patients. Further research should investigate the neural substrates underlying BDD symptomatology.
Superconducting, energy variable heavy ion linac with constant
β, multicell cavities of CH-type
(2009)
An energy variable ion linac consisting of multigap, constant-β cavities was developed. The effect of phase sliding, unavoidable in any constant-β section, is leading to a coherent rf phase motion, which fits well to the H-type structures with their long π-mode sections and separated lenses. The exact periodicity of the cell lengths within each cavity results in technical advantages, such as higher calculation accuracy when only one single period can be simulated, simpler manufacturing, and tuning. This is most important in the case of superconducting cavities. By using this concept, an improved design for a 217 MHz cw superconducting heavy ion linac with energy variation has been worked out. The small output energy spread of ± 3 AkeV is provided over the whole range of energy variation from 3.5 to 7.3 AMeV. These capabilities would allow for a competitive research in the field of radiochemistry and for a production of super heavy elements (SHE), especially. A first 19-cell cavity of that type was designed, built, and rf tested successfully at the Institute for Applied Physics (IAP) Frankfurt. A 325.224 MHz, seven-cell cavity with constant β = 0.16 is under development and will be operated in a frequency controlled mode. It will be equipped with a power coupler and beam tests with Unilac beams at GSI are foreseen.
Objective: Acute kidney injury (AKI) after cardiac surgery procedures is associated with poor patient outcomes. Cystatin C as a marker for renal failure has been shown to be of prognostic value; however, a wide range of its predictive accuracy has been reported. The aim of the study was to evaluate whether the measurement of pre- and postoperative serum cystatin C improves the prediction of AKI.
Methods: In a single-centre, prospective study of 70 patients (74 ± 9ys; range 47-85ys; 77% male), cystatin C was measured six times: (T1 = preoperative, T2 = start cardiopulmonary bypass (CPB), T3 = 20 min after CPB, T4 = end of operation; T5 = 24 h postoperatively; T6 = 7d postoperatively). Predictive property, in terms of the need for renal replacement therapy (RRT), was analysed by receiver operating characteristics (ROC) statistics and described by the area under the curve (AUC).
Results: With respect to RRT (n = 8), serum cystatin C was significantly higher at the end of the operation (T4), 24 h postoperatively at T5 and at T6. The AUCs for preoperative T1 and intraoperative T2/3 cystatin C were <0.7 (95% CI, 0.47-0.85). The earliest significant predictive AUCs were found at the end of the operation (T4: p = 0.03 95% CI 0.58-0.88 AUC 0.73) and 24 h postoperatively (T5: p = 0.003 95% CI 0.74-0.96 AUC 0.85).
Conclusions: Early postoperative serum cystatin C increase appears to be a moderate biomarker in the prediction of AKI, whereas a preoperative and intraoperative cystatin C increase has only a limited diagnostic and predictive value.