Search

Explainable artificial intelligence is a research field that tries to provide more transparency for autonomous intelligent systems. Explainability has been used, particularly in reinforcement learning and robotic scenarios, to better understand the robot decision-making process. Previous work, however, has been widely focused on providing technical explanations that can be better understood by AI practitioners than non-expert end-users. In this work, we make use of human-like explanations built from the probability of success to complete the goal that an autonomous robot shows after performing an action. These explanations are intended to be understood by people who have no or very little experience with artificial intelligence methods. This paper presents a user trial to study whether these explanations that focus on the probability an action has of succeeding in its goal constitute a suitable explanation for non-expert end-users. The results obtained show that non-expert participants rate robot explanations that focus on the probability of success higher and with less variance than technical explanations generated from Q-values, and also favor counterfactual explanations over standalone explanations., 8 pages, 8 figures

Appendix D – Rubrics Appendix D1 – Focus Group Part 1 Rubric Appendix D2 – Focus Group Part 2 Rubric As part of the thesis: Evaluating Explanations of Artificial Intelligence Decisions: the Explanation Quality Rubric and Survey by Charlotte Young (Federation University)

Appendix E – Surveys Appendix E1 – Focus Group Surveys Appendix E1a – Focus Group Round 2 Appendix E1b – Focus Group Round 3 Appendix E2 – MTurk #1 Survey Appendix E3 – MTurk #2 Survey Appendix E4 – MTurk #3 Surveys Appendix E4a – MTurk #3 – Independent Validation Round 1 Appendix E4b – MTurk #3 – Independent Validation Round 2 Appendix E4c – MTurk #3 – Independent Validation Debriefing Survey As part of the thesis: Evaluating Explanations of Artificial Intelligence Decisions: the Explanation Quality Rubric and Survey by Charlotte Young (Federation University)

Appendix F – Scripts Appendix F1 – MTurk #3 – Independent Validation Round 1 Randomisation Script Appendix F2 – MTurk #3 – Independent Validation Round 2 Randomisation Script As part of the thesis: Evaluating Explanations of Artificial Intelligence Decisions: the Explanation Quality Rubric and Survey by Charlotte Young (Federation University)

Appendix C - Case Studies Appendix C1 – Focus Group Case Studies Appendix C1a – Case Studies from Part 1 Appendix C1b – Case Study from Part 2 Appendix C1c – Case Study from Part 3 Appendix C2 – Initial Validation Case Studies Appendix C2a – Initial Validation Case Studies Centrelink Letter Appendix C2b – Initial Validation Case Studies IBM Loan Appendix C3 – Constructive Feedback Case Study Appendix C4 – Independent Validation Case Studies Appendix C4a – Independent Validation Case Studies - Scenarios 1-6 Appendix C4b – Updated Independent Validation Case Studies - Scenarios 1-6 As part of the thesis: Evaluating Explanations of Artificial Intelligence Decisions: the Explanation Quality Rubric and Survey by Charlotte Young (Federation University)

Appendix A - Results Appendix A1 – Focus Group Results from Part 1 Appendix A2 – Focus Group Results from Part 2 Appendix A3 – Focus Group Results from Part 3 Appendix A4 – Initial Validation Results Appendix A5 – Constructive Feedback Results Appendix A6 – Independent Validation Results Appendix A6a – Independent Validation Results - Part 1 Appendix A6b – Independent Validation Results - Part 2 Appendix A6c – Independent Validation Debriefing Survey As part of the thesis: Evaluating Explanations of Artificial Intelligence Decisions: the Explanation Quality Rubric and Survey by Charlotte Young (Federation University)

Accurate assessment of SARS-CoV-2 immunity is critical in evaluating vaccine efficacy and devising public health policies. Whilst the exact nature of effective immunity remains incompletely defined, SARS-CoV-2-specific T-cell responses are a critical feature that will likely form a key correlate of protection against COVID-19. Here, we developed and optimized a high-throughput whole blood-based assay to determine the T-cell response associated with prior SARS-CoV-2 infection and/or vaccination amongst 231 healthy donors and 68 cancer patients. Following overnight in vitro stimulation with SARS-CoV-2-specific peptides, blood plasma samples were analysed for TH 1-type cytokines. Highly significant differential IFN-γ+ /IL-2+ SARS-CoV-2-specific T-cell responses were seen amongst previously infected COVID-19-positive healthy donors in comparison with unknown / naïve individuals (p < 0·0001). IFN-γ production was more effective at identifying asymptomatic donors, demonstrating higher sensitivity (96·0% vs. 83·3%) but lower specificity (84·4% vs. 92·5%) than measurement of IL-2. A single COVID-19 vaccine dose induced IFN-γ and/or IL-2 SARS-CoV-2-specific T-cell responses in 116 of 128 (90·6%) healthy donors, reducing significantly to 27 of 56 (48·2%) when measured in cancer patients (p < 0·0001). A second dose was sufficient to boost T-cell responses in the majority (90·6%) of cancer patients, albeit IFN-γ+ responses were still significantly lower overall than those induced in healthy donors (p = 0·034). Three-month post-vaccination T-cell responses also declined at a faster rate in cancer patients. Overall, this cost-effective standardizable test ensures accurate and comparable assessments of SARS-CoV-2-specific T-cell responses amenable to widespread population immunity testing, and identifies individuals at greater need of booster vaccinations.

Global migration poses unique challenges for migrant health and wellbeing. Within Australia, African Australians are a new and emerging population, and research concerning the health and wellbeing of the diverse people grouped under this socio-analytic category is limited. The extant literature emerging from the field of public health reveals several barriers to health and wellbeing for African Australians. Explanations for poor health outcomes among some African Australians tend to rely upon individual behavioural approaches that focus on individual attributes, behaviours, culture, and single-axis social determinants, with little attention paid to intersecting multiple dimensions of inequality. Without accounting for the interrelated nature of African Australians’ social locations and intersecting systems of oppression/privilege, practice and policy responses may have limited impact on redressing health inequalities. This research comprises two phases, where Phase 1 aimed to drive the focus of Phase 2 to an area of expressed importance to participants. Phase 1 of the research asked: From the perspectives of African Australians and non-African Australians working with and for African Australians, what are the priority concerns regarding African Australian health and wellbeing in Greater Melbourne? Phase 2 of the research, driven by the findings of Phase 1, sought to understand how members of African Australian immigrant organisations perceive their influence over, and work to improve, health and wellbeing. Phase 1 was a qualitative study that utilised intersectionality to analyse interview notes from two group interviews and 22 ‘slow interviews.’ An Issues Paper produced by 50 African Australians, was also reviewed as a secondary data source because it captured additional voices, therefore, widening the scope of the study. Interview participants included 35 African Australians and nine people of non-African backgrounds working with, and for, African Australians in the community sector. Phase 1 findings reveal that systems of oppression/privilege negatively influence health and wellbeing for specific African Australians at the intersections of race/ethnicity, migration pathway, age, and gender. Health outcomes are found to be influenced by segregation and Othering in education, labour market discrimination, and gendered racism in health-care provision. Furthermore, racism, ageism, and sexism are found to challenge the identity construction of young African Australian women, increasing their risk of homelessness. Interpersonal and structural racism at the intersections of migration pathway, gender, and age also produce particular vulnerabilities for poor mental health. Phase 1 findings also highlight the existence of, and structural barriers facing, African Australian immigrant organisations in an under-resourced and competitive community welfare sector. Participants expressed a strong desire for African Australians to drive and implement solutions to these health concerns, and immigrant organisations presented as a setting for this to occur. These findings shaped the focus of Phase 2. Despite the common tendency across different countries for immigrants to form organisations, there is a dearth of literature about immigrant organisations broadly, with little known of their efforts to reduce health inequalities. Therefore, driven by the concerns raised in Phase 1, Phase 2 of the research bridges literature from third sector studies, migration studies, and public health, with an empirical dual-case study and intersectional analysis of two African Australian immigrant organisations operating in Greater Melbourne. The study explores the potential of these immigrant organisations as health settings to reduce health inequalities identified in Phase 1. Data collected from observation, in-depth interviews, reflexive interviews, and document review were analysed through the lens of intersectionality. Phase 2 findings identify key mechanisms for influencing African Australian health and wellbeing, including the acquisition of resources, growing networks, professional and personal development, generating solidarity, community capacity building, and advocacy. The activities of the participating African Australian immigrant organisations are found to contribute to a culture of health, empowering communities, and health advocacy. Findings also reveal that while the organisations resisted hierarchical forms of oppression/privilege through some of their activities, working within and through systems of oppression/privilege both constrained, and created opportunities for, their work. In some ways, systems of oppression/privilege were also reinforced and reproduced. This thesis extends current literature concerning immigrant organisations by highlighting their potential as valuable health settings, where activities for resisting, reinforcing, and reproducing systems of oppression/privilege can influence health and wellbeing.

Background/Aims: Chronic renal failure (CRF) is often accompanied by increased oxidative stress and euthyroid sick syndrome (ESS). The cause of ESS is unknown, and it is unknown whether there exists a link between oxidant stress and ESS in CRF patients. Therefore, we aim to investigate oxidative stress and type 1 deiodinase (DIO1) expression, which plays the key role in the ESS in CRF patients. Methods: In-patients with CRF were divided into the two group: Group 1 is ESS patients consisting of 60 patients with low free triiodothyronine (FT3) and Group 2 consisting of 60 patients with normal FT3. Group 3 consisted of 60 healthy volunteers recruited as controls. The baseline clinical parameters of patients were evaluated with standard routine methods in a clinical laboratory. Serum levels of 8-isoprostane and DIO1 were measured by enzyme-linked immunosorbent assay (ELISA). Multiple regression analysis was used to analyze the relationship between oxidative stress, DIO1 and FT3. Results: The concentrations of serum 8-Isoprostane in Group 1 and Group 2 were substantially higher than that of Group 3 (p< 0.05), however there was no significant difference between Group 1 and Group 2 (p=0.516). The serum DIO1 level was higher in Group 2 than in Group 1 and Group 3 (p< 0.001). Multivariate linear regression analysis revealed that the DIO1 concentration and FT3 level were not associated with the concentration of serum 8-Isoprostane. Conclusions: CRF patients showed elevated oxidative stress. The CRF patients without ESS showed higher expression of DIO1 than patients with ESS and the control group. The concentration of serum 8-Isoprostane was not correlated with FT3 and DIO1 levels.

Background:Clinical practice guidelines provide an evidence-based approach to managing single chronic conditions, but their applicability to multiple conditions has been actively debated. Incorporating patient-preference recommendations and involving consumers in guideline development may enhance their applicability, but further understanding is needed.Objectives:To assess guidelines that include recommendations for comorbid conditions to determine the extent to which they incorporate patient-preference recommendations; use consumer-engagement processes during development, and, if so, whether these processes produce more patient-preference recommendations; and meet standard quality criteria, particularly in relation to stakeholder involvement.Design:A review of Australian guidelines published from 2006 to 2014 that incorporated recommendations for managing comorbid conditions in primary care. Document analysis of guidelines examined the presence of patient-preference recommendations and the consumer-engagement processes used. The Appraisal of Guidelines for Research and Evaluation instrument was used to assess guideline quality.Results:Thirteen guidelines were reviewed. Twelve included at least one core patient-preference recommendation. Ten used consumer-engagement processes, including participation in development groups (seven guidelines) and reviewing drafts (ten guidelines). More extensive consumer engagement was generally linked to greater incorporation of patient-preference recommendations. Overall quality of guidelines was mixed, particularly in relation to stakeholder involvement.Conclusions:Guidelines do incorporate some patient-preference recommendations, but more explicit acknowledgement is required. Consumer-engagement processes used during guideline development have the potential to assist in identifying patient preferences, but further research is needed. Clarification of the consumer role and investment in consumer training may strengthen these processes. Journal of Comorbidity 2015;5(1):122–131