The Gaussian mechanism (GM) represents a universally employed tool for achieving differential privacy (DP), and a large body of work has been devoted to its analysis. We argue that the three prevailing interpretations of the GM, namely epsilon/delta-DP, f-DP and Rényi DP can be expressed by using a single parameter psi, which we term the sensitivity index. Psi uniquely characterises the GM and its properties by encapsulating its two fundamental quantities: the sensitivity of the query and the magnitude of the noise perturbation. With strong links to the ROC curve and the hypothesis-testing interpretation of DP, psi offers the practitioner a powerful method for interpreting, comparing and communicating the privacy guarantees of Gaussian mechanisms.
We study the applications of differential privacy (DP) in the context of graph-structured data and discuss the formulations of DP applicable to the publication of graphs and their associated statistics as well as machine learning on graph-based data, including graph neural networks (GNNs). Interpreting DP guarantees in the context of graph-structured data can be challenging, as individual data points are interconnected (often non-linearly or sparsely). This connectivity complicates the computation of individual privacy loss in differentially private learning. The problem is exacerbated by an absence of a single, well-established formulation of DP in graph settings. This issue extends to the domain of GNNs, rendering private machine learning on graph-structured data a challenging task. A lack of prior systematisation work motivated us to study graph-based learning from a privacy perspective. In this work, we systematise different formulations of DP on graphs, discuss challenges and promising applications, including the GNN domain. We compare and separate works into graph analytics tasks and graph learning tasks with GNNs. We conclude our work with a discussion of open questions and potential directions for further research in this area.
Background: Computed tomography (CT) is the most common imaging modality in traumatic brain injury (TBI). However, its conventional use requires expert clinical interpretation and does not provide detailed quantitative outputs, which may have prognostic importance. Deep learning could reliably and efficiently detect, distinguish, and quantify different lesion types, providing opportunities for personalised treatment strategies and clinical research. Methods: An initial convolutional neural network (CNN) was trained and validated on expert manual segmentations (97 scans). This CNN was then used to automatically segment a new set of 839 scans, which were then manually corrected by experts. From these, a subset of 184 scans was used to train a final CNN for multi-class, voxel-wise segmentation of lesion types. The performance of this CNN was evaluated on a held-out test set with 655 scans. External validation was performed on a large, independent set of 500 patients from a different continent. Findings: When compared to manual reference, CNN-derived lesion volumes showed a mean error of 0·86mL (95% CI -5·23 to 6·94) for intraparenchymal haemorrhage (IPH), 1·83mL (-12·01 to 15·66) for extra-axial haemorrhage (EAH), 2·09mL (-9·38 to 13·56) for perilesional oedema and 0·07mL (-1·00 to 1·13) for intraventricular haemorrhage (IVH). Further, the CNN detected lesions with AUCs of 0·90 (0·86-0·94) for IPH, 0·80 (0·75-0·85) for EAH, 0·95 (0·89-1·00) for IVH on the external, independent patient dataset. Interpretation: We demonstrate the ability of a CNN to separately segment, detect and quantify multi-class haemorrhagic lesions and importantly, perilesional oedema. These volumetric lesion estimates allow clinically relevant quantification of lesion burden and progression, with potential applications in clinical care and research in TBI. Funding: European Union 7th Framework Programme, Hannelore Kohl Stiftung; OneMind; Integra Neurosciences; European Research Council Horizon 2020; Engineering and Physical Sciences Research Council (UK); Academy of Medical Sciences/Health Foundation (UK); National Institute for Health Research (UK). ; CENTER-TBI study was supported by the European Union 7th Framework program (EC grant 602150). Additional funding sources: Hannelore Kohl Stiftung; NeuroTrauma Sciences; Integra Neurosciences; European Research Council (ERC) Horizon 2020 (EC grant 757173); Engineering and Physical Sciences Research Council (EPSRC) (EP/R511547/1); Academy of Medical Sciences/The Health Foundation (UK); National Institute for Health Research (UK).
The results leading to this publication have received funding from the Innovative Medicines Initiative 2 Joint Undertaking under grant agreement No 777394 for the project AIMS-2-TRIALS. This Joint Undertaking receives support from the European Union's Horizon 2020 research and innovation programme and EFPIA and AUTISM SPEAKS, Autistica, SFARI. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results. Any views expressed are those of the author(s) and not necessarily those of the funders.
Abstract: Background: The European Union (EU) aims to optimize patient protection and efficiency of health-care research by harmonizing procedures across Member States. Nonetheless, further improvements are required to increase multicenter research efficiency. We investigated IRB procedures in a large prospective European multicenter study on traumatic brain injury (TBI), aiming to inform and stimulate initiatives to improve efficiency. Methods: We reviewed relevant documents regarding IRB submission and IRB approval from European neurotrauma centers participating in the Collaborative European NeuroTrauma Effectiveness Research in Traumatic Brain Injury (CENTER-TBI). Documents included detailed information on IRB procedures and the duration from IRB submission until approval(s). They were translated and analyzed to determine the level of harmonization of IRB procedures within Europe. Results: From 18 countries, 66 centers provided the requested documents. The primary IRB review was conducted centrally (N = 11, 61%) or locally (N = 7, 39%) and primary IRB approval was obtained after one (N = 8, 44%), two (N = 6, 33%) or three (N = 4, 23%) review rounds with a median duration of respectively 50 and 98 days until primary IRB approval. Additional IRB approval was required in 55% of countries and could increase duration to 535 days. Total duration from submission until required IRB approval was obtained was 114 days (IQR 75–224) and appeared to be shorter after submission to local IRBs compared to central IRBs (50 vs. 138 days, p = 0.0074). Conclusion: We found variation in IRB procedures between and within European countries. There were differences in submission and approval requirements, number of review rounds and total duration. Research collaborations could benefit from the implementation of more uniform legislation and regulation while acknowledging local cultural habits and moral values between countries.