Interpolated Joint Space Adversarial Training for Robust and Generalizable Defenses.

Adversarial training (AT) is considered to be one of the most reliable defenses against adversarial attacks. However, models trained with AT sacrifice standard accuracy and do not generalize well to unseen attacks. Recent works show generalization improvement with adversarial samples under unseen threat models such as on-manifold threat model or neural perceptual threat model. However, the former requires exact manifold information while the latter requires algorithm relaxation. Motivated by these considerations, we propose a novel threat model called Joint Space Threat Model (JSTM), which exploits the underlying manifold information with Normalizing Flow, ensuring that the exact manifold assumption holds. Under JSTM, we develop novel adversarial attacks and defenses. Specifically, we propose the Robust Mixup strategy in which we maximize the adversity of the interpolated images and gain robustness and prevent overfitting. Our experiments show that Interpolated Joint Space Adversarial Training (IJSAT) achieves good performance in standard accuracy, robustness, and generalization. IJSAT is also flexible and can be used as a data augmentation method to improve standard accuracy and combined with many existing AT approaches to improve robustness. We demonstrate the effectiveness of our approach on three benchmark datasets, CIFAR-10/100, OM-ImageNet and CIFAR-10-C.

Factors related to good death in the Eastern Mediterranean Region: a systematic review.

Background: Considerations for palliative care and quality of death has significantly increased over the past 10 years in the Eastern Mediterranean Region (EMR). Recent trends in ageing and increasing chronic disease burden have drawn attention to the need to pay attention to the concept of good death and related factors from the perspective of the local population. Aims: To assess the factors related to good death in the EMR. Methods: We searched PubMed, Embase, Scopus, Web of Science, and ProQuest on 22 October 2021 for English language articles, with no time limit, using keywords "quality of death", "good death", "quality of dying", "good dying", "Middle Eastern", and countries in the Region. The quality of articles was evaluated using the Hawker criterion and based on the PRISMA guidelines. From the thematic analysis, the factors influencing good death were extracted. EndNote X8 software was used for data management. Results: The search yielded 55 articles, and 14 were included in the study, with a total of 3589 participants. Factors related to good death were classified into 2 main categories: patient preferences and end-of-life care. The former was divided into 4 groups: symptom management, psychological support, social support, and spiritual care. The second category included 2 subcategories: death control and patient autonomy, and end-of-life care. Conclusion: Although patients' beliefs about good death are personal, unique, and different, perception about good death in the EMR depends on the extent to which patients' preferences are met and end-of-life care is provided. More research on good death is recommended in the context of Islamic countries in EMR, and to empower patients and their families to better manage the dying process and create educational programmes.

Psychometric Properties of the Persian Version of Palliative Care Outcome Scale (POS) in Adult Patients With Cancer.

Background: Measuring the outcomes of palliative care plays an important role to improve the quality, efficiency, and availability of these services in patients with cancer. Using valid, reliable, and culturally appropriate tools has a considerable role to measure these outcomes. This study aimed to assess the psychometric properties of the translated version of the Palliative care Outcome Scale (POS). Methods: This methodological study was conducted in two outpatient clinics related to Shohada Tajrish and Baqiyatallah hospitals in Tehran in 2019-2020. The translation was done using the Forward-Backward approach after gaining permission from the developer. Face validity was tested with 10 patients with cancer through cognitive interviewing, as well as content validity with four experts. Construct validity was performed by (n = 203) exploratory factor analysis and confirmation (N = 150). To assess the reliability, internal consistency was assessed by using Cronbach's alpha coefficient, and relative stability was assessed using the interclass correlation coefficient (ICC). Furthermore, interpretability and ceiling and floor effects were assessed. Results: A total of 353 patients with cancer under palliative care were included in the study. Then, three psychological (30%), physical (12.25%), and social factors (12.08%) with a cumulative variance of 54.34% were extracted in exploratory factor analysis. Confirmatory factor analysis showed that the model has a good fit of information. Cronbach's alpha coefficient for scale was 0.719. Furthermore, the ICC was 0.812. The scale was interpretable, and ceiling and floor effects were 0%. Conclusion: Persian version of the POS was evaluated as a valid and reliable tool. Therefore, it can be used by the clinician to monitor the consequences of palliative care in Iranian cancer patients.