Effects of a comprehensive brain computed tomography deep learning model on radiologist detection accuracy.

OBJECTIVES: Non-contrast computed tomography of the brain (NCCTB) is commonly used to detect intracranial pathology but is subject to interpretation errors. Machine learning can augment clinical decision-making and improve NCCTB scan interpretation. This retrospective detection accuracy study assessed the performance of radiologists assisted by a deep learning model and compared the standalone performance of the model with that of unassisted radiologists. METHODS: A deep learning model was trained on 212,484 NCCTB scans drawn from a private radiology group in Australia. Scans from inpatient, outpatient, and emergency settings were included. Scan inclusion criteria were age ≥ 18 years and series slice thickness ≤ 1.5 mm. Thirty-two radiologists reviewed 2848 scans with and without the assistance of the deep learning system and rated their confidence in the presence of each finding using a 7-point scale. Differences in AUC and Matthews correlation coefficient (MCC) were calculated using a ground-truth gold standard. RESULTS: The model demonstrated an average area under the receiver operating characteristic curve (AUC) of 0.93 across 144 NCCTB findings and significantly improved radiologist interpretation performance. Assisted and unassisted radiologists demonstrated an average AUC of 0.79 and 0.73 across 22 grouped parent findings and 0.72 and 0.68 across 189 child findings, respectively. When assisted by the model, radiologist AUC was significantly improved for 91 findings (158 findings were non-inferior), and reading time was significantly reduced. CONCLUSIONS: The assistance of a comprehensive deep learning model significantly improved radiologist detection accuracy across a wide range of clinical findings and demonstrated the potential to improve NCCTB interpretation. CLINICAL RELEVANCE STATEMENT: This study evaluated a comprehensive CT brain deep learning model, which performed strongly, improved the performance of radiologists, and reduced interpretation time. The model may reduce errors, improve efficiency, facilitate triage, and better enable the delivery of timely patient care. KEY POINTS: • This study demonstrated that the use of a comprehensive deep learning system assisted radiologists in the detection of a wide range of abnormalities on non-contrast brain computed tomography scans. • The deep learning model demonstrated an average area under the receiver operating characteristic curve of 0.93 across 144 findings and significantly improved radiologist interpretation performance. • The assistance of the comprehensive deep learning model significantly reduced the time required for radiologists to interpret computed tomography scans of the brain.

Governance of health research in four eastern and southern African countries.

BACKGROUND: Health research governance is an essential function of national health research systems. Yet many African countries have not developed strong health research governance structures and processes. This paper presents a comparative analysis of national health research governance in Botswana, Kenya, Uganda and Zambia, where health sciences research production is well established relative to some others in the region and continues to grow. The paper aims to examine progress made and challenges faced in strengthening health research governance in these countries. METHODS: We collected data through document review and key informant interviews with a total of 80 participants including decision-makers, researchers and funders across stakeholder institutions in the four countries. Data on health research governance were thematically coded for policies, legislation, regulation and institutions and analysed comparatively across the four national health research systems. RESULTS: All countries were found to be moving from using a research governance framework set by national science, technology and innovation policies to one that is more anchored in health research structures and policies within the health sectors. Kenya and Zambia have adopted health research legislation and policies, while Botswana and Uganda are in the process of developing the same. National-level health research coordination and regulation is hampered by inadequate financial and human resource capacities, which present challenges for building strong health research governance institutions. CONCLUSION: Building health research governance as a key pillar of national health research systems involves developing stronger governance institutions, strengthening health research legislation, increasing financing for governance processes and improving human resource capacity in health research governance and management.

Reflexivity in Health Promotion: A Typology for Training.

Reflexivity has emerged as a key concept in the field of health promotion (HP). Yet it remains unclear how diverse forms of reflexivity are specifically relevant to HP concerns, and how these "reflexivities" are interconnected. We argue that frameworks are needed to support more systematic integration of reflexivity in HP training and practice. In this article, we propose a typology of reflexivity in HP to facilitate the understanding of reflexivity in professional training. Drawing from key theories and models of reflexivity, this typology proposes three reflexive positions (ideal-types) with specific purposes for HP: reflexivity in, on, and underlying action. This article illustrates our typology's ideal-types with vignettes collected from HP actors working with reflexivity in North America and Europe. We suggest that our typology constitutes a conceptual device to organize and discuss a variety of experiences of engaging with reflexivity for HP. We propose the typology may support integrating reflexivity as a key feature in training a future cadre of health promoters and as a means for building a responsible HP practice.

Parasites and genetic diversity in an invasive bumblebee.

Biological invasions are facilitated by the global transportation of species and climate change. Given that invasions may cause ecological and economic damage and pose a major threat to biodiversity, understanding the mechanisms behind invasion success is essential. Both the release of non-native populations from natural enemies, such as parasites, and the genetic diversity of these populations may play key roles in their invasion success. We investigated the roles of parasite communities, through enemy release and parasite acquisition, and genetic diversity in the invasion success of the non-native bumblebee, Bombus hypnorum, in the United Kingdom. The invasive B. hypnorum had higher parasite prevalence than most, or all native congeners for two high-impact parasites, probably due to higher susceptibility and parasite acquisition. Consequently parasites had a higher impact on B. hypnorum queens' survival and colony-founding success than on native species. Bombus hypnorum also had lower functional genetic diversity at the sex-determining locus than native species. Higher parasite prevalence and lower genetic diversity have not prevented the rapid invasion of the United Kingdom by B. hypnorum. These data may inform our understanding of similar invasions by commercial bumblebees around the world. This study suggests that concerns about parasite impacts on the small founding populations common to re-introduction and translocation programs may be less important than currently believed.

Analysis of Line and Tube Detection Performance of a Chest X-ray Deep Learning Model to Evaluate Hidden Stratification.

This retrospective case-control study evaluated the diagnostic performance of a commercially available chest radiography deep convolutional neural network (DCNN) in identifying the presence and position of central venous catheters, enteric tubes, and endotracheal tubes, in addition to a subgroup analysis of different types of lines/tubes. A held-out test dataset of 2568 studies was sourced from community radiology clinics and hospitals in Australia and the USA, and was then ground-truth labelled for the presence, position, and type of line or tube from the consensus of a thoracic specialist radiologist and an intensive care clinician. DCNN model performance for identifying and assessing the positioning of central venous catheters, enteric tubes, and endotracheal tubes over the entire dataset, as well as within each subgroup, was evaluated. The area under the receiver operating characteristic curve (AUC) was assessed. The DCNN algorithm displayed high performance in detecting the presence of lines and tubes in the test dataset with AUCs > 0.99, and good position classification performance over a subpopulation of ground truth positive cases with AUCs of 0.86-0.91. The subgroup analysis showed that model performance was robust across the various subtypes of lines or tubes, although position classification performance of peripherally inserted central catheters was relatively lower. Our findings indicated that the DCNN algorithm performed well in the detection and position classification of lines and tubes, supporting its use as an assistant for clinicians. Further work is required to evaluate performance in rarer scenarios, as well as in less common subgroups.

Increasing Effectiveness and Equity in Strengthening Health Research Capacity Using Data and Metrics: Recent Advances of the ESSENCE Mechanism.

Background: The ESSENCE on Health Research initiative established a Working Group on Review of Investments in 2018 to improve coordination and collaboration among funders of health research capacity strengthening. The Working Group comprises more than a dozen ESSENCE members, including diverse representation by geography, country income level, the public sector, and philanthropy. Objective: The overall goal of the Working Group is increased research on national health priorities as well as improved pandemic preparedness, and, ultimately, fewer countries with very limited research capacity. Methods: We developed a basic set of metrics for national health research capacity, assessed different models of coordination and collaboration, took a deeper dive into eight countries to characterize their national research capacity, and began to identify opportunities to better coordinate our investments. In this article, we summarize the presentations, discussions, and outcomes of our second annual (virtual) meeting, which had more than 100 participants representing funders, researchers, and other stakeholders from higher- and lower-income countries worldwide. Findings and conclusions: Presentations on the first day included the keynote speaker, Dr. Soumya Swaminathan, chief scientist of the World Health Organization (WHO), and updates on data and metrics for research capacity, which are critical to establish targets, road maps, and budgets. The second day focused on improving collaboration and coordination among funders and other stakeholders, the potential return on investment for health research, ongoing work to increase coordination at the country level, and examples of research capacity strengthening efforts in diverse health research areas from around the world. We concluded that an intentional data- and metric-driven approach to health research capacity strengthening, emphasizing coordination among funders, local leadership, and equitable partnerships and allocation of resources, will enhance the health systems of resource-poor countries as well as the world's pandemic preparedness.

The Roles of Regional Organisations in Strengthening Health Research Systems in Africa: Activities, Gaps, and Future Perspectives.

BACKGROUND: Regional cooperation on health in Africa is not new. The institutional landscape of regional cooperation for health and health research, however, has seen important changes. Recent health emergencies have focussed regional bodies' attention on supporting aspects of national health preparedness and response. The state of national health research systems is a key element of capacity to plan and respond to health needs - raising questions about the roles African regional bodies can or should play in strengthening health research systems. METHODS: We mapped regional organisations involved in health research across Africa and conducted 18 interviews with informants from 15 regional organisations. We investigated the roles, challenges, and opportunities of these bodies in strengthening health research. We deductively coded interview data using themes from established pillars of health research systems - governance, creating resources, research production and use, and financing. We analysed organisations' relevant activities in these areas, how they do this work, and where they perceive impact. RESULTS: Regional organisations with technical foci on health or higher education (versus economic or political remits) were involved in all four areas. Most organisations reported activities in governance and research use. Involvement in governance centred mainly around agenda-setting and policy harmonisation. For organisations involved in creating resources, activities focused on strengthening human resources, but few reported developing research institutions, networks, or infrastructure. Organisations reported more involvement in disseminating than producing research. Generally, few have directly contributed to financing health research. Informants reported gaps in research coordination, infrastructure, and advocacy at regional level. Finally, we found regional bodies' mandates, authority, and collaborations influence their activities in supporting national health research systems. CONCLUSION: Continued strengthening of health research on the African continent requires strategic thinking about the roles, comparative advantages, and capability of regional organisations to facilitate capacity and growth of health research systems.

Mapping regional cooperation of state actors for health research systems in Africa: A social network analysis.

Regional bodies can potentially play an important role in improving health research in Africa. This study analyses the network of African state-based regional organisations for health research and assesses their potential relationship with national health research performance metrics. After cataloguing organisations and their membership, we conducted a social network analysis to determine key network attributes of national governments' connections via regional organisations supporting functions of health research systems. This data was used to test the hypothesis that state actors with more connections to other actors via regional organisations would have higher levels of health research performance across indicators. With 21 unique regional organisations, the African continent is densely networked around health research systems issues. In general, the regional network for health research is inclusive. No single actor serves as a nexus. However, when statistics are grouped by African Union regions, influential poles emerge, with the most predominate spheres of influence in Eastern and Western Africa. Further, when connectivity data was analysed against national health research performance, there were no statistically significant relationships between increased connectivity and higher performance of key health research metrics. The inclusive and dense network dynamics of African regional organisations for health research strengthening present key opportunities for knowledge diffusion and cooperation to improve research capacity on the continent. Further reflection is needed on appropriate and meaningful ways to assess the role of regionalism and evaluate the influence of regional organisations in strengthening health research systems in Africa.

Evaluation of an Artificial Intelligence Model for Detection of Pneumothorax and Tension Pneumothorax in Chest Radiographs.

Importance: Early detection of pneumothorax, most often via chest radiography, can help determine need for emergent clinical intervention. The ability to accurately detect and rapidly triage pneumothorax with an artificial intelligence (AI) model could assist with earlier identification and improve care. Objective: To compare the accuracy of an AI model vs consensus thoracic radiologist interpretations in detecting any pneumothorax (incorporating both nontension and tension pneumothorax) and tension pneumothorax. Design, Setting, and Participants: This diagnostic study was a retrospective standalone performance assessment using a data set of 1000 chest radiographs captured between June 1, 2015, and May 31, 2021. The radiographs were obtained from patients aged at least 18 years at 4 hospitals in the Mass General Brigham hospital network in the United States. Included radiographs were selected using 2 strategies from all chest radiography performed at the hospitals, including inpatient and outpatient. The first strategy identified consecutive radiographs with pneumothorax through a manual review of radiology reports, and the second strategy identified consecutive radiographs with tension pneumothorax using natural language processing. For both strategies, negative radiographs were selected by taking the next negative radiograph acquired from the same radiography machine as each positive radiograph. The final data set was an amalgamation of these processes. Each radiograph was interpreted independently by up to 3 radiologists to establish consensus ground-truth interpretations. Each radiograph was then interpreted by the AI model for the presence of pneumothorax and tension pneumothorax. This study was conducted between July and October 2021, with the primary analysis performed between October and November 2021. Main Outcomes and Measures: The primary end points were the areas under the receiver operating characteristic curves (AUCs) for the detection of pneumothorax and tension pneumothorax. The secondary end points were the sensitivities and specificities for the detection of pneumothorax and tension pneumothorax. Results: The final analysis included radiographs from 985 patients (mean [SD] age, 60.8 [19.0] years; 436 [44.3%] female patients), including 307 patients with nontension pneumothorax, 128 patients with tension pneumothorax, and 550 patients without pneumothorax. The AI model detected any pneumothorax with an AUC of 0.979 (95% CI, 0.970-0.987), sensitivity of 94.3% (95% CI, 92.0%-96.3%), and specificity of 92.0% (95% CI, 89.6%-94.2%) and tension pneumothorax with an AUC of 0.987 (95% CI, 0.980-0.992), sensitivity of 94.5% (95% CI, 90.6%-97.7%), and specificity of 95.3% (95% CI, 93.9%-96.6%). Conclusions and Relevance: These findings suggest that the assessed AI model accurately detected pneumothorax and tension pneumothorax in this chest radiograph data set. The model's use in the clinical workflow could lead to earlier identification and improved care for patients with pneumothorax.

Nuclear localization of factor inhibitor hypoxia-inducible factor in prostate cancer is associated with poor prognosis.

PURPOSE: We determined the role of factor inhibiting hypoxia-inducible factor-1 in prostate cancer specimens. MATERIALS AND METHODS: A tissue microarray of 152 prostate cancers was constructed and stained for factor inhibiting hypoxia-inducible factor-1, hypoxia-inducible factor-1α and 2α, and glucose transporter 1 as a prototypical downstream target of hypoxia-inducible factor-1α. Correlation analysis was done between these variables, and between factor inhibiting hypoxia-inducible factor-1, and clinical and pathological variables, including prostate specific antigen as a surrogate of recurrence. RESULTS: Factor inhibiting hypoxia-inducible factor-1 was expressed in the cytoplasm and/or the nucleus in 86.5% of tumors, including exclusive cytoplasmic expression in 51.3% and exclusive nuclear expression in 5.3%. Any nuclear and exclusive expression of factor inhibiting hypoxia-inducible factor was associated with poor prognosis on univariate analysis (p = 0.007 and 0.042, respectively). On multivariate analysis men with nuclear expression in tumors were twice as likely to experience recurrence (p = 0.034). CONCLUSIONS: Factor inhibiting hypoxia-inducible factor-1 is widely expressed in prostate tumors. Its differential subcellular expression suggests that regulation of its expression is an important factor in the activity of the hypoxia-inducible factor pathway. Its modulation may help treat hypoxia-inducible factor driven aggressive prostate cancer.

Twenty-five years after the Ottawa Charter: the critical role of health promotion for public health.

After a quarter of a century, the Ottawa Charter for Health Promotion, often recognized as a foundational document of health promotion, continues to be relevant for public health. Inspired by the WHO Constitution, the Alma Ata Declaration, and the Lalonde Report, the Ottawa Charter endorses a positive definition of health, situates health as a product of daily life, proposes core values and principles for public health action, and outlines three strategies and five action areas reaching beyond the boundaries of the health care sector. The Charter established a radical agenda for public health, specifically to expressly convey the values public health pursues, thereby increasing the potential for the reflexivity of the field and opportunities to consider complementary values in actions that promote population health. In this paper, we examine how public health has integrated health promotion by exploring examples of changes in public health systems and practice at international and national levels of governance. Nevertheless, an important challenge remains for health promotion: better use of research to understand how the values, principles and processes of health promotion can help to achieve public health mandates. A three-pronged action plan is proposed.

A scoping review of theories and conceptual frameworks used to analyse health financing policy processes in sub-Saharan Africa.

Health financing policies are critical policy instruments to achieve Universal Health Coverage, and they constitute a key area in policy analysis literature for the health policy and systems research (HPSR) field. Previous reviews have shown that analyses of policy change in low- and middle-income countries are under-theorised. This study aims to explore which theories and conceptual frameworks have been used in research on policy processes of health financing policy in sub-Saharan Africa and to identify challenges and lessons learned from their use. We conducted a scoping review of literature published in English and French between 2000 and 2017. We analysed 23 papers selected as studies of health financing policies in sub-Saharan African countries using policy process or health policy-related theory or conceptual framework ex ante. Theories and frameworks used alone were from political science (35%), economics (9%) and HPSR field (17%). Thirty-five per cent of authors adopted a 'do-it-yourself' (bricolage) approach combining theories and frameworks from within political science or between political science and HPSR. Kingdon's multiple streams theory (22%), Grindle and Thomas' arenas of conflict (26%) and Walt and Gilson's policy triangle (30%) were the most used. Authors select theories for their empirical relevance, methodological rational (e.g. comparison), availability of examples in literature, accessibility and consensus. Authors cite few operational and analytical challenges in using theory. The hybridisation, diversification and expansion of mid-range policy theories and conceptual frameworks used deductively in health financing policy reform research are issues for HPSR to consider. We make three recommendations for researchers in the HPSR field. Future research on health financing policy change processes in sub-Saharan Africa should include reflection on learning and challenges for using policy theories and frameworks in the context of HPSR.

Chest radiographs and machine learning - Past, present and future.

Despite its simple acquisition technique, the chest X-ray remains the most common first-line imaging tool for chest assessment globally. Recent evidence for image analysis using modern machine learning points to possible improvements in both the efficiency and the accuracy of chest X-ray interpretation. While promising, these machine learning algorithms have not provided comprehensive assessment of findings in an image and do not account for clinical history or other relevant clinical information. However, the rapid evolution in technology and evidence base for its use suggests that the next generation of comprehensive, well-tested machine learning algorithms will be a revolution akin to early advances in X-ray technology. Current use cases, strengths, limitations and applications of chest X-ray machine learning systems are discussed.

Assessment of the effect of a comprehensive chest radiograph deep learning model on radiologist reports and patient outcomes: a real-world observational study.

OBJECTIVES: Artificial intelligence (AI) algorithms have been developed to detect imaging features on chest X-ray (CXR) with a comprehensive AI model capable of detecting 124 CXR findings being recently developed. The aim of this study was to evaluate the real-world usefulness of the model as a diagnostic assistance device for radiologists. DESIGN: This prospective real-world multicentre study involved a group of radiologists using the model in their daily reporting workflow to report consecutive CXRs and recording their feedback on level of agreement with the model findings and whether this significantly affected their reporting. SETTING: The study took place at radiology clinics and hospitals within a large radiology network in Australia between November and December 2020. PARTICIPANTS: Eleven consultant diagnostic radiologists of varying levels of experience participated in this study. PRIMARY AND SECONDARY OUTCOME MEASURES: Proportion of CXR cases where use of the AI model led to significant material changes to the radiologist report, to patient management, or to imaging recommendations. Additionally, level of agreement between radiologists and the model findings, and radiologist attitudes towards the model were assessed. RESULTS: Of 2972 cases reviewed with the model, 92 cases (3.1%) had significant report changes, 43 cases (1.4%) had changed patient management and 29 cases (1.0%) had further imaging recommendations. In terms of agreement with the model, 2569 cases showed complete agreement (86.5%). 390 (13%) cases had one or more findings rejected by the radiologist. There were 16 findings across 13 cases (0.5%) deemed to be missed by the model. Nine out of 10 radiologists felt their accuracy was improved with the model and were more positive towards AI poststudy. CONCLUSIONS: Use of an AI model in a real-world reporting environment significantly improved radiologist reporting and showed good agreement with radiologists, highlighting the potential for AI diagnostic support to improve clinical practice.

Do comprehensive deep learning algorithms suffer from hidden stratification? A retrospective study on pneumothorax detection in chest radiography.

OBJECTIVES: To evaluate the ability of a commercially available comprehensive chest radiography deep convolutional neural network (DCNN) to detect simple and tension pneumothorax, as stratified by the following subgroups: the presence of an intercostal drain; rib, clavicular, scapular or humeral fractures or rib resections; subcutaneous emphysema and erect versus non-erect positioning. The hypothesis was that performance would not differ significantly in each of these subgroups when compared with the overall test dataset. DESIGN: A retrospective case-control study was undertaken. SETTING: Community radiology clinics and hospitals in Australia and the USA. PARTICIPANTS: A test dataset of 2557 chest radiography studies was ground-truthed by three subspecialty thoracic radiologists for the presence of simple or tension pneumothorax as well as each subgroup other than positioning. Radiograph positioning was derived from radiographer annotations on the images. OUTCOME MEASURES: DCNN performance for detecting simple and tension pneumothorax was evaluated over the entire test set, as well as within each subgroup, using the area under the receiver operating characteristic curve (AUC). A difference in AUC of more than 0.05 was considered clinically significant. RESULTS: When compared with the overall test set, performance of the DCNN for detecting simple and tension pneumothorax was statistically non-inferior in all subgroups. The DCNN had an AUC of 0.981 (0.976-0.986) for detecting simple pneumothorax and 0.997 (0.995-0.999) for detecting tension pneumothorax. CONCLUSIONS: Hidden stratification has significant implications for potential failures of deep learning when applied in clinical practice. This study demonstrated that a comprehensively trained DCNN can be resilient to hidden stratification in several clinically meaningful subgroups in detecting pneumothorax.

Effect of a comprehensive deep-learning model on the accuracy of chest x-ray interpretation by radiologists: a retrospective, multireader multicase study.

BACKGROUND: Chest x-rays are widely used in clinical practice; however, interpretation can be hindered by human error and a lack of experienced thoracic radiologists. Deep learning has the potential to improve the accuracy of chest x-ray interpretation. We therefore aimed to assess the accuracy of radiologists with and without the assistance of a deep-learning model. METHODS: In this retrospective study, a deep-learning model was trained on 821 681 images (284 649 patients) from five data sets from Australia, Europe, and the USA. 2568 enriched chest x-ray cases from adult patients (≥16 years) who had at least one frontal chest x-ray were included in the test dataset; cases were representative of inpatient, outpatient, and emergency settings. 20 radiologists reviewed cases with and without the assistance of the deep-learning model with a 3-month washout period. We assessed the change in accuracy of chest x-ray interpretation across 127 clinical findings when the deep-learning model was used as a decision support by calculating area under the receiver operating characteristic curve (AUC) for each radiologist with and without the deep-learning model. We also compared AUCs for the model alone with those of unassisted radiologists. If the lower bound of the adjusted 95% CI of the difference in AUC between the model and the unassisted radiologists was more than -0·05, the model was considered to be non-inferior for that finding. If the lower bound exceeded 0, the model was considered to be superior. FINDINGS: Unassisted radiologists had a macroaveraged AUC of 0·713 (95% CI 0·645-0·785) across the 127 clinical findings, compared with 0·808 (0·763-0·839) when assisted by the model. The deep-learning model statistically significantly improved the classification accuracy of radiologists for 102 (80%) of 127 clinical findings, was statistically non-inferior for 19 (15%) findings, and no findings showed a decrease in accuracy when radiologists used the deep-learning model. Unassisted radiologists had a macroaveraged mean AUC of 0·713 (0·645-0·785) across all findings, compared with 0·957 (0·954-0·959) for the model alone. Model classification alone was significantly more accurate than unassisted radiologists for 117 (94%) of 124 clinical findings predicted by the model and was non-inferior to unassisted radiologists for all other clinical findings. INTERPRETATION: This study shows the potential of a comprehensive deep-learning model to improve chest x-ray interpretation across a large breadth of clinical practice. FUNDING: Annalise.ai.

Evolution of a rapid response system from voluntary to mandatory activation.

A policy requiring rapid response activation for all patients who met physiologic instability criteria, which was initiated four years after the rapid response system's inception, significantly increased the number of rapid response calls and was associated with a reduction in cardiorespiratory arrests outside of critical care areas.