The optimisation of sales practices in consumer markets through machine learning not only harbours the potential to better match consumer preferences with products, but also risks to facilitate the ...exploitation of consumer weaknesses discovered via data analysis. More specifically, recent technological advances have brought us to the edge of mind-reading technologies, which automatically analyse mental states and adapt offers accordingly, in potentially manipulative ways. This article shows that, in market contexts, the challenges of manipulation by algorithm necessitate an integrated understanding of unfair commercial practice, data protection, and privacy law. It maps the interactions between these contiguous yet distinct fields of law, and draws on economics and computer science to develop a novel framework to deal with algorithmic influence. Furthermore, it critically discusses the Commission proposals for the Digital Services Act and the Artificial Intelligence Act, and suggests to complement them with more broadly applicable measures to mitigate algorithmic manipulation.
This paper shows that the law, in subtle ways, may set hitherto unrecognized incentives for the adoption of explainable machine learning applications. In doing so, we make two novel contributions. ...First, on the legal side, we show that to avoid liability, professional actors, such as doctors and managers, may soon be legally compelled to use explainable ML models. We argue that the importance of explainability reaches far beyond data protection law, and crucially influences questions of contractual and tort liability for the use of ML models. To this effect, we conduct two legal case studies, in medical and corporate merger applications of ML. As a second contribution, we discuss the (legally required) trade-off between accuracy and explainability and demonstrate the effect in a technical case study in the context of spam classification.
Das neue unionale Datenschutzrecht ist, entgegen mancher Befürchtung, kein law of everything. Vielmehr müssen unterschiedliche Rechtsmaterien ineinandergreifen, um eine sachgerechte Regelungsstruktur ...im Schnittbereich von Datenschutzrecht und Privatrecht aufzubauen. Philipp Hacker bestimmt das Verhältnis dieser Rechtsmaterien, insbesondere von DS-GVO und BGB. Denn die Verschränkung unterschiedlicher Technologieformen fordert mehr denn je ein rechtsbereichsübergreifendes Verständnis von juristischer Dogmatik und ein interdisziplinär fundiertes Konzept von Regulierung. Auf Basis des geltenden Rechts entwirft er ein integriertes Marktordnungsrecht für digitale Austauschverhältnisse. Die Untersuchung schließt mit Reformperspektiven, die aufzeigen, wie die informierte Einwilligung durch eine technologische ersetzt werden kann, um eine privatautonome Gestaltung von Rechtsverhältnissen unter den Bedingungen der digitalen Wirtschaft zu ermöglichen. Die Arbeit wurde mit dem Wissenschaftspreis der Deutschen Stiftung für Recht und Informatik 2020 ausgezeichnet.
Increasingly, discrimination by algorithms is perceived as a societal and legal problem. As a response, a number of criteria for implementing algorithmic fairness in machine learning have been ...developed in the literature. This paper proposes the continuous fairness algorithm
(
CFA
θ
)
which enables a continuous interpolation between different fairness definitions. More specifically, we make three main contributions to the existing literature. First, our approach allows the decision maker to continuously vary between specific concepts of individual and group fairness. As a consequence, the algorithm enables the decision maker to adopt intermediate “worldviews” on the degree of discrimination encoded in algorithmic processes, adding nuance to the extreme cases of “we’re all equal” and “what you see is what you get” proposed so far in the literature. Second, we use optimal transport theory, and specifically the concept of the barycenter, to maximize decision maker utility under the chosen fairness constraints. Third, the algorithm is able to handle cases of intersectionality, i.e., of multi-dimensional discrimination of certain groups on grounds of several criteria. We discuss three main examples (credit applications; college admissions; insurance contracts) and map out the legal and policy implications of our approach. The explicit formalization of the trade-off between individual and group fairness allows this post-processing approach to be tailored to different situational contexts in which one or the other fairness criterion may take precedence. Finally, we evaluate our model experimentally.
Smart city applications are increasingly deployed in urban spaces around the world. We contend that they only merit the attribute 'smart' if they embody what we term 'substantial smartness'. To ...develop this concept, we draw on both political and legal theories to show that citizen participation and activation, as well as respect for human and fundamental rights, are two essential dimensions of substantial smartness. Both dimensions, however, need to accommodate temporality, i.e., rapid changes in deployed technologies, their purposes and citizens' use of public infrastructure. By highlighting three examples and discussing smart city challenges to the GDPR, non-discrimination law and the proposed EU AI Act, we demonstrate that politics needs the law - and vice versa - to unlock the potential of substantively smart cities.
Lung transplantation for idiopathic pulmonary arterial hypertension has the highest reported postoperative mortality of all indications. Reasons lie in the complexity of treatment of these patients ...and the frequent occurrence of postoperative left ventricular failure. Transplantation on intraoperative extracorporeal membrane oxygenation support instead of cardiopulmonary bypass and even more the prolongation of extracorporeal membrane oxygenation into the postoperative period helps to overcome these problems. We reviewed our experience with this concept.
All patients undergoing bilateral lung transplantation for idiopathic pulmonary arterial hypertension on intraoperative extracorporeal membrane oxygenation with or without prophylactic extracorporeal membrane oxygenation prolongation into the postoperative period between January 2000 and December 2014 were retrospectively analysed.
Forty-one patients entered the study. Venoarterial extracorporeal membrane oxygenation support was prolonged into the postoperative period for a median of 2.5 days (range 1-40). Ninety-day, 1-, 3- and 5-year survival rates for the patient collective were 92.7%, 90.2%, 87.4% and 87.4%, respectively. When compared with 31 patients with idiopathic pulmonary arterial hypertension transplanted in the same period of time without prolongation of extracorporeal membrane oxygenation into the postoperative period, the results compared favourably (83.9%, 77.4%, 77.4%, and 77.4%; P = 0.189). Furthermore, these results are among the best results ever reported for this particularly difficult patient population.
Bilateral lung transplantation for idiopathic pulmonary arterial hypertension with intraoperative venoarterial extracorporeal membrane oxygenation support seems to provide superior outcome compared with the results reported about the use of cardiopulmonary bypass. Prophylactic prolongation of venoarterial extracorporeal membrane oxygenation into the early postoperative period provides stable postoperative conditions and seems to further improve the results.
In response to recent regulatory initiatives at the EU level, this article shows that training data for AI do not only play a key role in the development of AI applications, but are currently only ...inadequately captured by EU law. In this, I focus on three central risks of AI training data: risks of data quality, discrimination and innovation. Existing EU law, with the new copyright exception for text and data mining, only addresses a part of this risk profile adequately. Therefore, the article develops the foundations for a discrimination-sensitive quality regime for data sets and AI training, which emancipates itself from the controversial question of the applicability of data protection law to AI training data. Furthermore, it spells out concrete guidelines for the re-use of personal data for AI training purposes under the GDPR. Ultimately, the legislative and interpretive task rests in striking an appropriate balance between individual protection and the promotion of innovation. The article finishes with an assessment of the proposal for an Artificial Intelligence Act in this respect.
Pulmonary hypertension (PH) is characterized by increased pulmonary arterial pressure caused by the accumulation of mesenchymal-like cells in the pulmonary vasculature. PH can lead to right ...ventricular hypertrophy (RVH) and, ultimately, heart failure and death. In PH etiology, endothelial-to-mesenchymal transition (EndMT) has emerged as a critical process governing the conversion of endothelial cells into mesenchymal cells, and S100A4, EGF, and EGFR are implicated in EndMT. However, a potential role of S100A4, EGF, and EGFR in PH has to date not been elucidated. We therefore quantified S100A4, EGF, and EGFR in patients suffering from chronic thromboembolic pulmonary hypertension (CTEPH) and idiopathic pulmonary arterial hypertension (iPAH). To determine specificity for unilateral heart disease, the EndMT biomarker signature was further compared between PH patients presenting with RVH and patients suffering from aortic valve stenosis (AVS) with left ventricular hypertrophy. Reduced S100A4 concentrations were found in CTEPH and iPAH patients with RVH. Systemic EGF was increased in CTEPH but not in iPAH, while AVS patients displayed slightly diminished EGF levels. EGFR was downregulated in all patient groups when compared to healthy controls. Longitudinal data analysis revealed no effect of surgical therapies on EndMT markers. Pulmonary thrombo-endarterectomized samples were devoid of S100A4, while S100A4 tissue expression positively correlated with higher grades of Heath-Edwards histopathological lesions of iPAH-derived lung tissue. Histologically, EGFR was not detectable in CTEPH lungs or in iPAH lesions. Together, our data suggest an intricate role for S100A4 and EGF/EGFR in PH with right heart pathology.
Pulmonary hypertension (PH) is a vasoconstrictive disease characterized by elevated mean pulmonary arterial pressure (mPAP) at rest. Idiopathic pulmonary arterial hypertension (iPAH) and chronic ...thromboembolic pulmonary hypertension (CTEPH) represent two distinct subtypes of PH. Persisting PH leads to right ventricular (RV) hypertrophy, heart failure, and death. RV performance predicts survival and surgical interventions re-establishing physiological mPAP reverse cardiac remodeling. Nonetheless, a considerable number of PH patients are deemed inoperable. The underlying mechanism(s) governing cardiac regeneration, however, remain largely elusive.
In a longitudinal approach, we profiled the transcriptional landscapes of hypertrophic RVs and recovered hearts 3 months after surgery of iPAH and CTEPH patients.
Genes associated with cellular responses to inflammatory stimuli and metal ions were downregulated, and cardiac muscle tissue development was induced in iPAH after recovery. In CTEPH patients, genes related to muscle cell development were decreased, and genes governing cardiac conduction were upregulated in RVs following regeneration. Intriguingly, early growth response 1 (
), a profibrotic regulator, was identified as a major transcription factor of hypertrophic RVs in iPAH and CTEPH. A histological assessment confirmed our biocomputational results, and suggested a pivotal role for EGR1 in RV vasculopathy.
Our findings improved our understanding of the molecular events driving reverse cardiac remodeling following surgery. EGR1 might represent a promising candidate for targeted therapy of PH patients not eligible for surgical treatment.
Capacitively coupled discharges with a radio-frequency operated voltage (ccrf) are important for plasma assisted material processing. Experiments with electronegative oxygen ccrf discharges show a ...high-energy peak in the energy distribution of negative ions arriving at the anode, depending on the cathode material used. One possible explanation is ionization at or close to the surface of the cathode for the production of negative ions. By introducing an additional surface ionization model into a Particle-In-Cell (PIC) simulation with Monte Carlo Collisions (MCC) the experimental result is reproduced qualitatively. Comparison of one dimensional and two dimensional simulation results allows an improved understanding of the microscopic processes determining the dynamics of negative ions.
Graphical abstract