A survey of skin tone assessment in prospective research.

Increasing evidence supports reduced accuracy of noninvasive assessment tools, such as pulse oximetry, temperature probes, and AI skin diagnosis benchmarks, in patients with darker skin tones. The FDA is exploring potential strategies for device regulation to improve performance across diverse skin tones by including skin tone criteria. However, there is no consensus about how prospective studies should perform skin tone assessment in order to take this bias into account. There are several tools available to conduct skin tone assessments including administered visual scales (e.g., Fitzpatrick Skin Type, Pantone, Monk Skin Tone) and color measurement tools (e.g., reflectance colorimeters, reflectance spectrophotometers, cameras), although none are consistently used or validated across multiple medical domains. Accurate and consistent skin tone measurement depends on many factors including standardized environments, lighting, body parts assessed, patient conditions, and choice of skin tone assessment tool(s). As race and ethnicity are inadequate proxies for skin tone, these considerations can be helpful in standardizing the effect of skin tone on studies such as AI dermatology diagnoses, pulse oximetry, and temporal thermometers. Skin tone bias in medical devices is likely due to systemic factors that lead to inadequate validation across diverse skin tones. There is an opportunity for researchers to use skin tone assessment methods with standardized considerations in prospective studies of noninvasive tools that may be affected by skin tone. We propose considerations that researchers must take in order to improve device robustness to skin tone bias.

A Responsible Framework for Applying Artificial Intelligence on Medical Images and Signals at the Point of Care: The PACS-AI Platform.

The potential of artificial intelligence (AI) in medicine lies in its ability to enhance clinicians' capacity to analyse medical images, thereby improving diagnostic precision and accuracy and thus enhancing current tests. However, the integration of AI within health care is fraught with difficulties. Heterogeneity among health care system applications, reliance on proprietary closed-source software, and rising cybersecurity threats pose significant challenges. Moreover, before their deployment in clinical settings, AI models must demonstrate their effectiveness across a wide range of scenarios and must be validated by prospective studies, but doing so requires testing in an environment mirroring the clinical workflow, which is difficult to achieve without dedicated software. Finally, the use of AI techniques in health care raises significant legal and ethical issues, such as the protection of patient privacy, the prevention of bias, and the monitoring of the device's safety and effectiveness for regulatory compliance. This review describes challenges to AI integration in health care and provides guidelines on how to move forward. We describe an open-source solution that we developed that integrates AI models into the Picture Archives Communication System (PACS), called PACS-AI. This approach aims to increase the evaluation of AI models by facilitating their integration and validation with existing medical imaging databases. PACS-AI may overcome many current barriers to AI deployment and offer a pathway toward responsible, fair, and effective deployment of AI models in health care. In addition, we propose a list of criteria and guidelines that AI researchers should adopt when publishing a medical AI model to enhance standardisation and reproducibility.

Research ethics and artificial intelligence for global health: perspectives from the global forum on bioethics in research.

BACKGROUND: The ethical governance of Artificial Intelligence (AI) in health care and public health continues to be an urgent issue for attention in policy, research, and practice. In this paper we report on central themes related to challenges and strategies for promoting ethics in research involving AI in global health, arising from the Global Forum on Bioethics in Research (GFBR), held in Cape Town, South Africa in November 2022. METHODS: The GFBR is an annual meeting organized by the World Health Organization and supported by the Wellcome Trust, the US National Institutes of Health, the UK Medical Research Council (MRC) and the South African MRC. The forum aims to bring together ethicists, researchers, policymakers, research ethics committee members and other actors to engage with challenges and opportunities specifically related to research ethics. In 2022 the focus of the GFBR was "Ethics of AI in Global Health Research". The forum consisted of 6 case study presentations, 16 governance presentations, and a series of small group and large group discussions. A total of 87 participants attended the forum from 31 countries around the world, representing disciplines of bioethics, AI, health policy, health professional practice, research funding, and bioinformatics. In this paper, we highlight central insights arising from GFBR 2022. RESULTS: We describe the significance of four thematic insights arising from the forum: (1) Appropriateness of building AI, (2) Transferability of AI systems, (3) Accountability for AI decision-making and outcomes, and (4) Individual consent. We then describe eight recommendations for governance leaders to enhance the ethical governance of AI in global health research, addressing issues such as AI impact assessments, environmental values, and fair partnerships. CONCLUSIONS: The 2022 Global Forum on Bioethics in Research illustrated several innovations in ethical governance of AI for global health research, as well as several areas in need of urgent attention internationally. This summary is intended to inform international and domestic efforts to strengthen research ethics and support the evolution of governance leadership to meet the demands of AI in global health research.

Drop the shortcuts: image augmentation improves fairness and decreases AI detection of race and other demographics from medical images.

BACKGROUND: It has been shown that AI models can learn race on medical images, leading to algorithmic bias. Our aim in this study was to enhance the fairness of medical image models by eliminating bias related to race, age, and sex. We hypothesise models may be learning demographics via shortcut learning and combat this using image augmentation. METHODS: This study included 44,953 patients who identified as Asian, Black, or White (mean age, 60.68 years ±18.21; 23,499 women) for a total of 194,359 chest X-rays (CXRs) from MIMIC-CXR database. The included CheXpert images comprised 45,095 patients (mean age 63.10 years ±18.14; 20,437 women) for a total of 134,300 CXRs were used for external validation. We also collected 1195 3D brain magnetic resonance imaging (MRI) data from the ADNI database, which included 273 participants with an average age of 76.97 years ±14.22, and 142 females. DL models were trained on either non-augmented or augmented images and assessed using disparity metrics. The features learned by the models were analysed using task transfer experiments and model visualisation techniques. FINDINGS: In the detection of radiological findings, training a model using augmented CXR images was shown to reduce disparities in error rate among racial groups (-5.45%), age groups (-13.94%), and sex (-22.22%). For AD detection, the model trained with augmented MRI images was shown 53.11% and 31.01% reduction of disparities in error rate among age and sex groups, respectively. Image augmentation led to a reduction in the model's ability to identify demographic attributes and resulted in the model trained for clinical purposes incorporating fewer demographic features. INTERPRETATION: The model trained using the augmented images was less likely to be influenced by demographic information in detecting image labels. These results demonstrate that the proposed augmentation scheme could enhance the fairness of interpretations by DL models when dealing with data from patients with different demographic backgrounds. FUNDING: National Science and Technology Council (Taiwan), National Institutes of Health.

An Introductory Guide to Artificial Intelligence in Interventional Radiology: Part 1 Foundational Knowledge.

Artificial intelligence (AI) is rapidly evolving and has transformative potential for interventional radiology (IR) clinical practice. However, formal training in AI may be limited for many clinicians and therefore presents a challenge for initial implementation and trust in AI. An understanding of the foundational concepts in AI may help familiarize the interventional radiologist with the field of AI, thus facilitating understanding and participation in the development and deployment of AI. A pragmatic classification system of AI based on the complexity of the model may guide clinicians in the assessment of AI. Finally, the current state of AI in IR and the patterns of implementation are explored (pre-procedural, intra-procedural, and post-procedural).

An Introductory Guide to Artificial Intelligence in Interventional Radiology: Part 2: Implementation Considerations and Harms.

The introduction of artificial intelligence (AI) in interventional radiology (IR) will bring about new challenges and opportunities for patients and clinicians. AI may comprise software as a medical device or AI-integrated hardware and will require a rigorous evaluation that should be guided based on the level of risk of the implementation. A hierarchy of risk of harm and possible harms are described herein. A checklist to guide deployment of an AI in a clinical IR environment is provided. As AI continues to evolve, regulation and evaluation of the AI medical devices will need to continue to evolve to keep pace and ensure patient safety.

Theory of radiologist interaction with instant messaging decision support tools: A sequential-explanatory study.

Radiology specific clinical decision support systems (CDSS) and artificial intelligence are poorly integrated into the radiologist workflow. Current research and development efforts of radiology CDSS focus on 4 main interventions, based around exam centric time points-after image acquisition, intra-report support, post-report analysis, and radiology workflow adjacent. We review the literature surrounding CDSS tools in these time points, requirements for CDSS workflow augmentation, and technologies that support clinician to computer workflow augmentation. We develop a theory of radiologist-decision tool interaction using a sequential explanatory study design. The study consists of 2 phases, the first a quantitative survey and the second a qualitative interview study. The phase 1 survey identifies differences between average users and radiologist users in software interventions using the User Acceptance of Information Technology: Toward a Unified View (UTAUT) framework. Phase 2 semi-structured interviews provide narratives on why these differences are found. To build this theory, we propose a novel solution called Radibot-a conversational agent capable of engaging clinicians with CDSS as an assistant using existing instant messaging systems supporting hospital communications. This work contributes an understanding of how radiologist-users differ from the average user and can be utilized by software developers to increase satisfaction of CDSS tools within radiology.

Environmental Sustainability and AI in Radiology: A Double-Edged Sword.

According to the World Health Organization, climate change is the single biggest health threat facing humanity. The global health care system, including medical imaging, must manage the health effects of climate change while at the same time addressing the large amount of greenhouse gas (GHG) emissions generated in the delivery of care. Data centers and computational efforts are increasingly large contributors to GHG emissions in radiology. This is due to the explosive increase in big data and artificial intelligence (AI) applications that have resulted in large energy requirements for developing and deploying AI models. However, AI also has the potential to improve environmental sustainability in medical imaging. For example, use of AI can shorten MRI scan times with accelerated acquisition times, improve the scheduling efficiency of scanners, and optimize the use of decision-support tools to reduce low-value imaging. The purpose of this Radiology in Focus article is to discuss this duality at the intersection of environmental sustainability and AI in radiology. Further discussed are strategies and opportunities to decrease AI-related emissions and to leverage AI to improve sustainability in radiology, with a focus on health equity. Co-benefits of these strategies are explored, including lower cost and improved patient outcomes. Finally, knowledge gaps and areas for future research are highlighted.

Peer review of GPT-4 technical report and systems card.

The study provides a comprehensive review of OpenAI's Generative Pre-trained Transformer 4 (GPT-4) technical report, with an emphasis on applications in high-risk settings like healthcare. A diverse team, including experts in artificial intelligence (AI), natural language processing, public health, law, policy, social science, healthcare research, and bioethics, analyzed the report against established peer review guidelines. The GPT-4 report shows a significant commitment to transparent AI research, particularly in creating a systems card for risk assessment and mitigation. However, it reveals limitations such as restricted access to training data, inadequate confidence and uncertainty estimations, and concerns over privacy and intellectual property rights. Key strengths identified include the considerable time and economic investment in transparent AI research and the creation of a comprehensive systems card. On the other hand, the lack of clarity in training processes and data raises concerns about encoded biases and interests in GPT-4. The report also lacks confidence and uncertainty estimations, crucial in high-risk areas like healthcare, and fails to address potential privacy and intellectual property issues. Furthermore, this study emphasizes the need for diverse, global involvement in developing and evaluating large language models (LLMs) to ensure broad societal benefits and mitigate risks. The paper presents recommendations such as improving data transparency, developing accountability frameworks, establishing confidence standards for LLM outputs in high-risk settings, and enhancing industry research review processes. It concludes that while GPT-4's report is a step towards open discussions on LLMs, more extensive interdisciplinary reviews are essential for addressing bias, harm, and risk concerns, especially in high-risk domains. The review aims to expand the understanding of LLMs in general and highlights the need for new reflection forms on how LLMs are reviewed, the data required for effective evaluation, and addressing critical issues like bias and risk.

Value Creation Through Artificial Intelligence and Cardiovascular Imaging: A Scientific Statement From the American Heart Association.

Multiple applications for machine learning and artificial intelligence (AI) in cardiovascular imaging are being proposed and developed. However, the processes involved in implementing AI in cardiovascular imaging are highly diverse, varying by imaging modality, patient subtype, features to be extracted and analyzed, and clinical application. This article establishes a framework that defines value from an organizational perspective, followed by value chain analysis to identify the activities in which AI might produce the greatest incremental value creation. The various perspectives that should be considered are highlighted, including clinicians, imagers, hospitals, patients, and payers. Integrating the perspectives of all health care stakeholders is critical for creating value and ensuring the successful deployment of AI tools in a real-world setting. Different AI tools are summarized, along with the unique aspects of AI applications to various cardiac imaging modalities, including cardiac computed tomography, magnetic resonance imaging, and positron emission tomography. AI is applicable and has the potential to add value to cardiovascular imaging at every step along the patient journey, from selecting the more appropriate test to optimizing image acquisition and analysis, interpreting the results for classification and diagnosis, and predicting the risk for major adverse cardiac events.

Fairness as an afterthought: An American perspective on fairness in model developer-clinician user collaborations.

Numerous ethics guidelines have been handed down over the last few years on the ethical applications of machine learning models. Virtually every one of them mentions the importance of "fairness" in the development and use of these models. Unfortunately, though, these ethics documents omit providing a consensually adopted definition or characterization of fairness. As one group of authors observed, these documents treat fairness as an "afterthought" whose importance is undeniable but whose essence seems strikingly elusive. In this essay, which offers a distinctly American treatment of "fairness," we comment on a number of fairness formulations and on qualitative or statistical methods that have been encouraged to achieve fairness. We argue that none of them, at least from an American moral perspective, provides a one-size-fits-all definition of or methodology for securing fairness that could inform or standardize fairness over the universe of use cases witnessing machine learning applications. Instead, we argue that because fairness comprehensions and applications reflect a vast range of use contexts, model developers and clinician users will need to engage in thoughtful collaborations that examine how fairness should be conceived and operationalized in the use case at issue. Part II of this paper illustrates key moments in these collaborations, especially when inter and intra disagreement occurs among model developer and clinician user groups over whether a model is fair or unfair. We conclude by noting that these collaborations will likely occur over the lifetime of a model if its claim to fairness is to advance beyond "afterthought" status.

AI pitfalls and what not to do: mitigating bias in AI.

Various forms of artificial intelligence (AI) applications are being deployed and used in many healthcare systems. As the use of these applications increases, we are learning the failures of these models and how they can perpetuate bias. With these new lessons, we need to prioritize bias evaluation and mitigation for radiology applications; all the while not ignoring the impact of changes in the larger enterprise AI deployment which may have downstream impact on performance of AI models. In this paper, we provide an updated review of known pitfalls causing AI bias and discuss strategies for mitigating these biases within the context of AI deployment in the larger healthcare enterprise. We describe these pitfalls by framing them in the larger AI lifecycle from problem definition, data set selection and curation, model training and deployment emphasizing that bias exists across a spectrum and is a sequela of a combination of both human and machine factors.

Ability of artificial intelligence to identify self-reported race in chest x-ray using pixel intensity counts.

Purpose: Prior studies show convolutional neural networks predicting self-reported race using x-rays of chest, hand and spine, chest computed tomography, and mammogram. We seek an understanding of the mechanism that reveals race within x-ray images, investigating the possibility that race is not predicted using the physical structure in x-ray images but is embedded in the grayscale pixel intensities. Approach: Retrospective full year 2021, 298,827 AP/PA chest x-ray images from 3 academic health centers across the United States and MIMIC-CXR, labeled by self-reported race, were used in this study. The image structure is removed by summing the number of each grayscale value and scaling to percent per image (PPI). The resulting data are tested using multivariate analysis of variance (MANOVA) with Bonferroni multiple-comparison adjustment and class-balanced MANOVA. Machine learning (ML) feed-forward networks (FFN) and decision trees were built to predict race (binary Black or White and binary Black or other) using only grayscale value counts. Stratified analysis by body mass index, age, sex, gender, patient type, make/model of scanner, exposure, and kilovoltage peak setting was run to study the impact of these factors on race prediction following the same methodology. Results: MANOVA rejects the null hypothesis that classes are the same with 95% confidence (F 7.38, P<0.0001) and balanced MANOVA (F 2.02, P<0.0001). The best FFN performance is limited [area under the receiver operating characteristic (AUROC) of 69.18%]. Gradient boosted trees predict self-reported race using grayscale PPI (AUROC 77.24%). Conclusions: Within chest x-rays, pixel intensity value counts alone are statistically significant indicators and enough for ML classification tasks of patient self-reported race.

Opportunistic detection of type 2 diabetes using deep learning from frontal chest radiographs.

Deep learning (DL) models can harness electronic health records (EHRs) to predict diseases and extract radiologic findings for diagnosis. With ambulatory chest radiographs (CXRs) frequently ordered, we investigated detecting type 2 diabetes (T2D) by combining radiographic and EHR data using a DL model. Our model, developed from 271,065 CXRs and 160,244 patients, was tested on a prospective dataset of 9,943 CXRs. Here we show the model effectively detected T2D with a ROC AUC of 0.84 and a 16% prevalence. The algorithm flagged 1,381 cases (14%) as suspicious for T2D. External validation at a distinct institution yielded a ROC AUC of 0.77, with 5% of patients subsequently diagnosed with T2D. Explainable AI techniques revealed correlations between specific adiposity measures and high predictivity, suggesting CXRs' potential for enhanced T2D screening.

Implications of predicting race variables from medical images.

AI-predicted race variables pose risks and opportunities for studying health disparities.

Fusion of imaging and non-imaging data for disease trajectory prediction for coronavirus disease 2019 patients.

Purpose: Our study investigates whether graph-based fusion of imaging data with non-imaging electronic health records (EHR) data can improve the prediction of the disease trajectories for patients with coronavirus disease 2019 (COVID-19) beyond the prediction performance of only imaging or non-imaging EHR data. Approach: We present a fusion framework for fine-grained clinical outcome prediction [discharge, intensive care unit (ICU) admission, or death] that fuses imaging and non-imaging information using a similarity-based graph structure. Node features are represented by image embedding, and edges are encoded with clinical or demographic similarity. Results: Experiments on data collected from the Emory Healthcare Network indicate that our fusion modeling scheme performs consistently better than predictive models developed using only imaging or non-imaging features, with area under the receiver operating characteristics curve of 0.76, 0.90, and 0.75 for discharge from hospital, mortality, and ICU admission, respectively. External validation was performed on data collected from the Mayo Clinic. Our scheme highlights known biases in the model prediction, such as bias against patients with alcohol abuse history and bias based on insurance status. Conclusions: Our study signifies the importance of the fusion of multiple data modalities for the accurate prediction of clinical trajectories. The proposed graph structure can model relationships between patients based on non-imaging EHR data, and graph convolutional networks can fuse this relationship information with imaging data to effectively predict future disease trajectory more effectively than models employing only imaging or non-imaging data. Our graph-based fusion modeling frameworks can be easily extended to other prediction tasks to efficiently combine imaging data with non-imaging clinical data.

MedShift: Automated Identification of Shift Data for Medical Image Dataset Curation.

Automated curation of noisy external data in the medical domain has long been in high demand, as AI technologies need to be validated using various sources with clean, annotated data. Identifying the variance between internal and external sources is a fundamental step in curating a high-quality dataset, as the data distributions from different sources can vary significantly and subsequently affect the performance of AI models. The primary challenges for detecting data shifts are - (1) accessing private data across healthcare institutions for manual detection and (2) the lack of automated approaches to learn efficient shift-data representation without training samples. To overcome these problems, we propose an automated pipeline called MedShift to detect top-level shift samples and evaluate the significance of shift data without sharing data between internal and external organizations. MedShift employs unsupervised anomaly detectors to learn the internal distribution and identify samples showing significant shiftness for external datasets, and then compares their performance. To quantify the effects of detected shift data, we train a multi-class classifier that learns internal domain knowledge and evaluates the classification performance for each class in external domains after dropping the shift data. We also propose a data quality metric to quantify the dissimilarity between internal and external datasets. We verify the efficacy of MedShift using musculoskeletal radiographs (MURA) and chest X-ray datasets from multiple external sources. Our experiments show that our proposed shift data detection pipeline can be beneficial for medical centers to curate high-quality datasets more efficiently.

Sociodemographic Variables Reporting in Human Radiology Artificial Intelligence Research.

PURPOSE: Artificial intelligence (AI) is rapidly reshaping how radiology is practiced. Its susceptibility to biases, however, is a primary concern as more AI algorithms become available for widespread use. So far, there has been limited evaluation of how sociodemographic variables are reported in radiology AI research. This study aims to evaluate the presence and extent of sociodemographic reporting in human subjects radiology AI original research. METHODS: All human subjects original radiology AI articles published from January to December 2020 in the top six US radiology journals, as determined by impact factor, were reviewed. Reporting of any sociodemographic variables (age, gender, and race or ethnicity) as well as any sociodemographic-based results were extracted. RESULTS: Of the 160 included articles, 54% reported at least one sociodemographic variable, 53% reported age, 47% gender, and 4% race or ethnicity. Six percent reported any sociodemographic-based results. There was significant variation in reporting of at least one sociodemographic variable by journal, ranging from 33% to 100%. CONCLUSIONS: Reporting of sociodemographic variables in human subjects original radiology AI research remains poor, putting the results and subsequent algorithms at increased risk of biases.