A survey of recent methods for addressing AI fairness and bias in biomedicine

OBJECTIVES

Artificial intelligence (AI) systems have the potential to revolutionize clinical practices, including improving diagnostic accuracy and surgical decision-making, while also reducing costs and manpower. However, it is important to recognize that these systems may perpetuate social inequities or demonstrate biases, such as those based on race or gender. Such biases can occur before, during, or after the development of AI models, making it critical to understand and address potential biases to enable the accurate and reliable application of AI models in clinical settings. To mitigate bias concerns during model development, we surveyed recent publications on different debiasing methods in the fields of biomedical natural language processing (NLP) or computer vision (CV). Then we discussed the methods, such as data perturbation and adversarial learning, that have been applied in the biomedical domain to address bias.

METHODS

We performed our literature search on PubMed, ACM digital library, and IEEE Xplore of relevant articles published between January 2018 and December 2023 using multiple combinations of keywords. We then filtered the result of 10,041 articles automatically with loose constraints, and manually inspected the abstracts of the remaining 890 articles to identify the 55 articles included in this review. Additional articles in the references are also included in this review. We discuss each method and compare its strengths and weaknesses. Finally, we review other potential methods from the general domain that could be applied to biomedicine to address bias and improve fairness.

RESULTS

The bias of AIs in biomedicine can originate from multiple sources such as insufficient data, sampling bias and the use of health-irrelevant features or race-adjusted algorithms. Existing debiasing methods that focus on algorithms can be categorized into distributional or algorithmic. Distributional methods include data augmentation, data perturbation, data reweighting methods, and federated learning. Algorithmic approaches include unsupervised representation learning, adversarial learning, disentangled representation learning, loss-based methods and causality-based methods.

Hidden Flaws Behind Expert-Level Accuracy of GPT-4 Vision in Medicine

Recent studies indicate that Generative Pre-trained Transformer 4 with Vision (GPT-4V) outperforms human physicians in medical challenge tasks. However, these evaluations primarily focused on the accuracy of multi-choice questions alone. Our study extends the current scope by conducting a comprehensive analysis of GPT-4V's rationales of image comprehension, recall of medical knowledge, and step-by-step multimodal reasoning when solving New England Journal of Medicine (NEJM) Image Challenges - an imaging quiz designed to test the knowledge and diagnostic capabilities of medical professionals. Evaluation results confirmed that GPT-4V performs comparatively to human physicians regarding multi-choice accuracy (81.6% vs. 77.8%). GPT-4V also performs well in cases where physicians incorrectly answer, with over 78% accuracy. However, we discovered that GPT-4V frequently presents flawed rationales in cases where it makes the correct final choices (35.5%), most prominent in image comprehension (27.2%). Regardless of GPT-4V's high accuracy in multi-choice questions, our findings emphasize the necessity for further in-depth evaluations of its rationales before integrating such multimodal AI models into clinical workflows.

A span-based model for extracting overlapping PICO entities from randomized controlled trial publications

OBJECTIVES

Extracting PICO (Populations, Interventions, Comparison, and Outcomes) entities is fundamental to evidence retrieval. We present a novel method, PICOX, to extract overlapping PICO entities.

MATERIALS AND METHODS

PICOX first identifies entities by assessing whether a word marks the beginning or conclusion of an entity. Then, it uses a multi-label classifier to assign one or more PICO labels to a span candidate. PICOX was evaluated using 1 of the best-performing baselines, EBM-NLP, and 3 more datasets, ie, PICO-Corpus and randomized controlled trial publications on Alzheimer's Disease (AD) or COVID-19, using entity-level precision, recall, and F1 scores.

RESULTS

PICOX achieved superior precision, recall, and F1 scores across the board, with the micro F1 score improving from 45.05 to 50.87 (P ≪.01). On the PICO-Corpus, PICOX obtained higher recall and F1 scores than the baseline and improved the micro recall score from 56.66 to 67.33. On the COVID-19 dataset, PICOX also outperformed the baseline and improved the micro F1 score from 77.10 to 80.32. On the AD dataset, PICOX demonstrated comparable F1 scores with higher precision when compared to the baseline.

CONCLUSION

PICOX excels in identifying overlapping entities and consistently surpasses a leading baseline across multiple datasets. Ablation studies reveal that its data augmentation strategy effectively minimizes false positives and improves precision.

Identifying social determinants of health from clinical narratives: A study of performance, documentation ratio, and potential bias

OBJECTIVE

To develop a natural language processing (NLP) package to extract social determinants of health (SDoH) from clinical narratives, examine the bias among race and gender groups, test the generalizability of extracting SDoH for different disease groups, and examine population-level extraction ratio.

METHODS

We developed SDoH corpora using clinical notes identified at the University of Florida (UF) Health. We systematically compared 7 transformer-based large language models (LLMs) and developed an open-source package - SODA (i.e., SOcial DeterminAnts) to facilitate SDoH extraction from clinical narratives. We examined the performance and potential bias of SODA for different race and gender groups, tested the generalizability of SODA using two disease domains including cancer and opioid use, and explored strategies for improvement. We applied SODA to extract 19 categories of SDoH from the breast (n = 7,971), lung (n = 11,804), and colorectal cancer (n = 6,240) cohorts to assess patient-level extraction ratio and examine the differences among race and gender groups.

RESULTS

We developed an SDoH corpus using 629 clinical notes of cancer patients with annotations of 13,193 SDoH concepts/attributes from 19 categories of SDoH, and another cross-disease validation corpus using 200 notes from opioid use patients with 4,342 SDoH concepts/attributes. We compared 7 transformer models and the GatorTron model achieved the best mean average strict/lenient F1 scores of 0.9122 and 0.9367 for SDoH concept extraction and 0.9584 and 0.9593 for linking attributes to SDoH concepts. There is a small performance gap (∼4%) between Males and Females, but a large performance gap (>16 %) among race groups. The performance dropped when we applied the cancer SDoH model to the opioid cohort; fine-tuning using a smaller opioid SDoH corpus improved the performance. The extraction ratio varied in the three cancer cohorts, in which 10 SDoH could be extracted from over 70 % of cancer patients, but 9 SDoH could be extracted from less than 70 % of cancer patients. Individuals from the White and Black groups have a higher extraction ratio than other minority race groups.

CONCLUSIONS

Our SODA package achieved good performance in extracting 19 categories of SDoH from clinical narratives. The SODA package with pre-trained transformer models is available at https://github.com/uf-hobi-informatics-lab/SODA_Docker.

Leveraging generative AI for clinical evidence synthesis needs to ensure trustworthiness

Evidence-based medicine promises to improve the quality of healthcare by empowering medical decisions and practices with the best available evidence. The rapid growth of medical evidence, which can be obtained from various sources, poses a challenge in collecting, appraising, and synthesizing the evidential information. Recent advancements in generative AI, exemplified by large language models, hold promise in facilitating the arduous task. However, developing accountable, fair, and inclusive models remains a complicated undertaking. In this perspective, we discuss the trustworthiness of generative AI in the context of automated summarization of medical evidence.

Confidence score: a data-driven measure for inclusive systematic reviews considering unpublished preprints

OBJECTIVES

COVID-19, since its emergence in December 2019, has globally impacted research. Over 360 000 COVID-19-related manuscripts have been published on PubMed and preprint servers like medRxiv and bioRxiv, with preprints comprising about 15% of all manuscripts. Yet, the role and impact of preprints on COVID-19 research and evidence synthesis remain uncertain.

MATERIALS AND METHODS

We propose a novel data-driven method for assigning weights to individual preprints in systematic reviews and meta-analyses. This weight termed the "confidence score" is obtained using the survival cure model, also known as the survival mixture model, which takes into account the time elapsed between posting and publication of a preprint, as well as metadata such as the number of first 2-week citations, sample size, and study type.

RESULTS

Using 146 preprints on COVID-19 therapeutics posted from the beginning of the pandemic through April 30, 2021, we validated the confidence scores, showing an area under the curve of 0.95 (95% CI, 0.92-0.98). Through a use case on the effectiveness of hydroxychloroquine, we demonstrated how these scores can be incorporated practically into meta-analyses to properly weigh preprints.

DISCUSSION

It is important to note that our method does not aim to replace existing measures of study quality but rather serves as a supplementary measure that overcomes some limitations of current approaches.

CONCLUSION

Our proposed confidence score has the potential to improve systematic reviews of evidence related to COVID-19 and other clinical conditions by providing a data-driven approach to including unpublished manuscripts.

Towards long-tailed, multi-label disease classification from chest X-ray: Overview of the CXR-LT challenge

Many real-world image recognition problems, such as diagnostic medical imaging exams, are "long-tailed" - there are a few common findings followed by many more relatively rare conditions. In chest radiography, diagnosis is both a long-tailed and multi-label problem, as patients often present with multiple findings simultaneously. While researchers have begun to study the problem of long-tailed learning in medical image recognition, few have studied the interaction of label imbalance and label co-occurrence posed by long-tailed, multi-label disease classification. To engage with the research community on this emerging topic, we conducted an open challenge, CXR-LT, on long-tailed, multi-label thorax disease classification from chest X-rays (CXRs). We publicly release a large-scale benchmark dataset of over 350,000 CXRs, each labeled with at least one of 26 clinical findings following a long-tailed distribution. We synthesize common themes of top-performing solutions, providing practical recommendations for long-tailed, multi-label medical image classification. Finally, we use these insights to propose a path forward involving vision-language foundation models for few- and zero-shot disease classification.

The incidence of surgical site infections in China

Surgical site infections (SSIs) are a common type of healthcare-associated infection. We performed a literature review to demonstrate the incidence of SSIs in mainland China based on studies since 2010. We included 231 eligible studies with ≥30 postoperative patients, comprising 14 providing overall SSI data regardless of surgical sites and 217 reporting SSIs for a specific site. We found that the overall SSI incidence was 2.91% (median; interquartile range: 1.05%, 4.57%) or 3.18% (pooled; 95% confidence interval: 1.85%, 4.51%) and the SSI incidence varied remarkably according to the surgical site between the lowest (median, 1.00%; pooled, 1.69%) in thyroid surgeries and the highest (median, 14.89%; pooled, 12.54%) in colorectal procedures. We uncovered that Enterobacterales and staphylococci were the most common types of micro-organisms associated with SSIs after various abdominal surgeries and cardiac or neurological procedures, respectively. We identified two, nine, and five studies addressing the impact of SSIs on mortality, the length of stay (LOS) in hospital, and additional healthcare-related economic burden, respectively, all of which demonstrated increased mortality, prolonged LOS, and elevated medical costs associated with SSIs among affected patients. Our findings illustrate that SSIs remain a relatively common, serious threat to patient safety in China, requiring more action. To tackle SSIs, we propose to establish a nationwide network for SSI surveillance using unified criteria with the aid of informatic techniques and to tailor and implement countermeasures based on local data and observation. We highlight that the impact of SSIs in China warrants further study.

Retrieval augmented scientific claim verification

Objective

To automate scientific claim verification using PubMed abstracts.

Materials and Methods

We developed CliVER, an end-to-end scientific Claim VERification system that leverages retrieval-augmented techniques to automatically retrieve relevant clinical trial abstracts, extract pertinent sentences, and use the PICO framework to support or refute a scientific claim. We also created an ensemble of three state-of-the-art deep learning models to classify rationale of support, refute, and neutral. We then constructed CoVERt, a new COVID VERification dataset comprising 15 PICO-encoded drug claims accompanied by 96 manually selected and labeled clinical trial abstracts that either support or refute each claim. We used CoVERt and SciFact (a public scientific claim verification dataset) to assess CliVER's performance in predicting labels. Finally, we compared CliVER to clinicians in the verification of 19 claims from 6 disease domains, using 189 648 PubMed abstracts extracted from January 2010 to October 2021.

Results

In the evaluation of label prediction accuracy on CoVERt, CliVER achieved a notable F1 score of 0.92, highlighting the efficacy of the retrieval-augmented models. The ensemble model outperforms each individual state-of-the-art model by an absolute increase from 3% to 11% in the F1 score. Moreover, when compared with four clinicians, CliVER achieved a precision of 79.0% for abstract retrieval, 67.4% for sentence selection, and 63.2% for label prediction, respectively.

Conclusion

CliVER demonstrates its early potential to automate scientific claim verification using retrieval-augmented strategies to harness the wealth of clinical trial abstracts in PubMed. Future studies are warranted to further test its clinical utility.

Decoding Suicide Decedent Profiles and Signs of Suicidal Intent Using Latent Class Analysis

Importance

Suicide rates in the US increased by 35.6% from 2001 to 2021. Given that most individuals die on their first attempt, earlier detection and intervention are crucial. Understanding modifiable risk factors is key to effective prevention strategies.

Objective

To identify distinct suicide profiles or classes, associated signs of suicidal intent, and patterns of modifiable risks for targeted prevention efforts.

Design, Setting, and Participants

This cross-sectional study used data from the 2003-2020 National Violent Death Reporting System Restricted Access Database for 306 800 suicide decedents. Statistical analysis was performed from July 2022 to June 2023.

Exposures

Suicide decedent profiles were determined using latent class analyses of available data on suicide circumstances, toxicology, and methods.

Main Outcomes and Measures

Disclosure of recent intent, suicide note presence, and known psychotropic usage.

Results

Among 306 800 suicide decedents (mean [SD] age, 46.3 [18.4] years; 239 627 males [78.1%] and 67 108 females [21.9%]), 5 profiles or classes were identified. The largest class, class 4 (97 175 [31.7%]), predominantly faced physical health challenges, followed by polysubstance problems in class 5 (58 803 [19.2%]), and crisis, alcohol-related, and intimate partner problems in class 3 (55 367 [18.0%]), mental health problems (class 2, 53 928 [17.6%]), and comorbid mental health and substance use disorders (class 1, 41 527 [13.5%]). Class 4 had the lowest rates of disclosing suicidal intent (13 952 [14.4%]) and leaving a suicide note (24 351 [25.1%]). Adjusting for covariates, compared with class 1, class 4 had the highest odds of not disclosing suicide intent (odds ratio [OR], 2.58; 95% CI, 2.51-2.66) and not leaving a suicide note (OR, 1.45; 95% CI, 1.41-1.49). Class 4 also had the lowest rates of all known psychiatric illnesses and psychotropic medications among all suicide profiles. Class 4 had more older adults (23 794 were aged 55-70 years [24.5%]; 20 100 aged ≥71 years [20.7%]), veterans (22 220 [22.9%]), widows (8633 [8.9%]), individuals with less than high school education (15 690 [16.1%]), and rural residents (23 966 [24.7%]).

Conclusions and Relevance

This study identified 5 distinct suicide profiles, highlighting a need for tailored prevention strategies. Improving the detection and treatment of coexisting mental health conditions, substance and alcohol use disorders, and physical illnesses is paramount. The implementation of means restriction strategies plays a vital role in reducing suicide risks across most of the profiles, reinforcing the need for a multifaceted approach to suicide prevention.

A survey of recent methods for addressing AI fairness and bias in biomedicine

Objectives

Artificial intelligence (AI) systems have the potential to revolutionize clinical practices, including improving diagnostic accuracy and surgical decision-making, while also reducing costs and manpower. However, it is important to recognize that these systems may perpetuate social inequities or demonstrate biases, such as those based on race or gender. Such biases can occur before, during, or after the development of AI models, making it critical to understand and address potential biases to enable the accurate and reliable application of AI models in clinical settings. To mitigate bias concerns during model development, we surveyed recent publications on different debiasing methods in the fields of biomedical natural language processing (NLP) or computer vision (CV). Then we discussed the methods, such as data perturbation and adversarial learning, that have been applied in the biomedical domain to address bias.

Methods

We performed our literature search on PubMed, ACM digital library, and IEEE Xplore of relevant articles published between January 2018 and December 2023 using multiple combinations of keywords. We then filtered the result of 10,041 articles automatically with loose constraints, and manually inspected the abstracts of the remaining 890 articles to identify the 55 articles included in this review. Additional articles in the references are also included in this review. We discuss each method and compare its strengths and weaknesses. Finally, we review other potential methods from the general domain that could be applied to biomedicine to address bias and improve fairness.

Results

The bias of AIs in biomedicine can originate from multiple sources such as insufficient data, sampling bias and the use of health-irrelevant features or race-adjusted algorithms. Existing debiasing methods that focus on algorithms can be categorized into distributional or algorithmic. Distributional methods include data augmentation, data perturbation, data reweighting methods, and federated learning. Algorithmic approaches include unsupervised representation learning, adversarial learning, disentangled representation learning, loss-based methods and causality-based methods.

Pan-cancer transcriptomic data of ABI1 transcript variants and molecular constitutive elements identifies novel cancer metastatic and prognostic biomarkers

BACKGROUND

Abelson interactor 1 (ABI1) is associated with the metastasis and prognosis of many malignancies. The association between ABI1 transcript spliced variants, their molecular constitutive exons and exon-exon junctions (EEJs) in 14 cancer types and clinical outcomes remains unsolved.

OBJECTIVE

To identify novel cancer metastatic and prognostic biomarkers from ABI1 total mRNA, TSVs, and molecular constitutive elements.

METHODS

Using data from TCGA and TSVdb database, the standard median of ABI1 total mRNA, TSV, exon, and EEJ expression was used as a cut-off value. Kaplan-Meier analysis, Chi-squared test (X2) and Kendall's tau statistic were used to identify novel metastatic and prognostic biomarkers, and Cox regression analysis was performed to screen and identify independent prognostic factors.

RESULTS

A total of 35 ABI1-related factors were found to be closely related to the prognosis of eight candidate cancer types. A total of 14 ABI1 TSVs and molecular constitutive elements were identified as novel metastatic and prognostic biomarkers in four cancer types. A total of 13 ABI1 molecular constitutive elements were identified as independent prognostic biomarkers in six cancer types.

CONCLUSIONS

In this study, we identified 14 ABI1-related novel metastatic and prognostic markers and 21 independent prognostic factors in total 8 candidate cancer types.

Total iron binding capacity: an independent predictor of prognosis for pulmonary arterial hypertension in systemic lupus erythematosus

OBJECTIVE

To investigate the role of parameters of iron metabolism in systemic lupus erythematosus (SLE) patients with pulmonary arterial hypertension (PAH).

METHOD

This was a prospective observational study recruiting patients diagnosed with systemic lupus erythematosus-associated pulmonary arterial hypertension (SLE-PAH). Patients with other factors that might lead to PAH were excluded from the study. All patients were assessed for PAH every 1-3 months and were followed up for 6 months. The primary outcome was considered improved if the grade of risk stratification declined at the endpoint; otherwise, it was considered unimproved.

RESULTS

In total, 29 patients with SLE-PAH were included in this study. The mean of serum ferritin was higher than normal, and total iron binding capacity (TIBC) decreased in 48% of patients. Correlation analyses showed that serum iron (SI) was negatively correlated with World Health Organization functional class (WHO-FC) (r = -0.409, p = 0.028), and positively correlated with Six-Minute Walk Test distance (6MWD) (r = 0.427, p = 0.021) and tricuspid annular plane systolic excursion (TAPSE) (r = 0.388, p = 0.037). Primary outcomes improved in 12 patients at the endpoint, and univariate logistic regression analyses indicated that TIBC was associated with improved primary outcomes in patients with SLE-PAH (odds ratio 12.00, 95% confidence interval 1.90-75.72).

CONCLUSION

SI was negatively correlated with WHO-FC, and positively correlated with 6MWD and TAPSE. Furthermore, TIBC was associated with improved outcomes of patients with SLE-PAH, which could be an independent predictor of prognosis. Further research is needed to verify the findings.

Faithful AI in Medicine: A Systematic Review with Large Language Models and Beyond

Objective

While artificial intelligence (AI), particularly large language models (LLMs), offers significant potential for medicine, it raises critical concerns due to the possibility of generating factually incorrect information, leading to potential long-term risks and ethical issues. This review aims to provide a comprehensive overview of the faithfulness problem in existing research on AI in healthcare and medicine, with a focus on the analysis of the causes of unfaithful results, evaluation metrics, and mitigation methods.

Materials and Methods

Using PRISMA methodology, we sourced 5,061 records from five databases (PubMed, Scopus, IEEE Xplore, ACM Digital Library, Google Scholar) published between January 2018 to March 2023. We removed duplicates and screened records based on exclusion criteria.

Results

With 40 leaving articles, we conducted a systematic review of recent developments aimed at optimizing and evaluating factuality across a variety of generative medical AI approaches. These include knowledge-grounded LLMs, text-to-text generation, multimodality-to-text generation, and automatic medical fact-checking tasks.

Discussion

Current research investigating the factuality problem in medical AI is in its early stages. There are significant challenges related to data resources, backbone models, mitigation methods, and evaluation metrics. Promising opportunities exist for novel faithful medical AI research involving the adaptation of LLMs and prompt engineering.

Conclusion

This comprehensive review highlights the need for further research to address the issues of reliability and factuality in medical AI, serving as both a reference and inspiration for future research into the safe, ethical use of AI in medicine and healthcare.

Adopting and expanding ethical principles for generative artificial intelligence from military to healthcare

In 2020, the U.S. Department of Defense officially disclosed a set of ethical principles to guide the use of Artificial Intelligence (AI) technologies on future battlefields. Despite stark differences, there are core similarities between the military and medical service. Warriors on battlefields often face life-altering circumstances that require quick decision-making. Medical providers experience similar challenges in a rapidly changing healthcare environment, such as in the emergency department or during surgery treating a life-threatening condition. Generative AI, an emerging technology designed to efficiently generate valuable information, holds great promise. As computing power becomes more accessible and the abundance of health data, such as electronic health records, electrocardiograms, and medical images, increases, it is inevitable that healthcare will be revolutionized by this technology. Recently, generative AI has garnered a lot of attention in the medical research community, leading to debates about its application in the healthcare sector, mainly due to concerns about transparency and related issues. Meanwhile, questions around the potential exacerbation of health disparities due to modeling biases have raised notable ethical concerns regarding the use of this technology in healthcare. However, the ethical principles for generative AI in healthcare have been understudied. As a result, there are no clear solutions to address ethical concerns, and decision-makers often neglect to consider the significance of ethical principles before implementing generative AI in clinical practice. In an attempt to address these issues, we explore ethical principles from the military perspective and propose the "GREAT PLEA" ethical principles, namely Governability, Reliability, Equity, Accountability, Traceability, Privacy, Lawfulness, Empathy, and Eutonomy, for generative AI in healthcare. Furthermore, we introduce a framework for adopting and expanding these ethical principles in a practical way that has been useful in the military and can be applied to healthcare for generative AI, based on contrasting their ethical concerns and risks. Ultimately, we aim to proactively address the ethical dilemmas and challenges posed by the integration of generative AI into healthcare practice.

Automated classification of lay health articles using natural language processing: a case study on pregnancy health and postpartum depression

Objective

Evidence suggests that high-quality health education and effective communication within the framework of social support hold significant potential in preventing postpartum depression. Yet, developing trustworthy and engaging health education and communication materials requires extensive expertise and substantial resources. In light of this, we propose an innovative approach that involves leveraging natural language processing (NLP) to classify publicly accessible lay articles based on their relevance and subject matter to pregnancy and mental health.

Materials and methods

We manually reviewed online lay articles from credible and medically validated sources to create a gold standard corpus. This manual review process categorized the articles based on their pertinence to pregnancy and related subtopics. To streamline and expand the classification procedure for relevance and topics, we employed advanced NLP models such as Random Forest, Bidirectional Encoder Representations from Transformers (BERT), and Generative Pre-trained Transformer model (gpt-3.5-turbo).

Results

The gold standard corpus included 392 pregnancy-related articles. Our manual review process categorized the reading materials according to lifestyle factors associated with postpartum depression: diet, exercise, mental health, and health literacy. A BERT-based model performed best (F1 = 0.974) in an end-to-end classification of relevance and topics. In a two-step approach, given articles already classified as pregnancy-related, gpt-3.5-turbo performed best (F1 = 0.972) in classifying the above topics.

Discussion

Utilizing NLP, we can guide patients to high-quality lay reading materials as cost-effective, readily available health education and communication sources. This approach allows us to scale the information delivery specifically to individuals, enhancing the relevance and impact of the materials provided.

A scoping review on multimodal deep learning in biomedical images and texts

Computer-assisted diagnostic and prognostic systems of the future should be capable of simultaneously processing multimodal data. Multimodal deep learning (MDL), which involves the integration of multiple sources of data, such as images and text, has the potential to revolutionize the analysis and interpretation of biomedical data. However, it only caught researchers' attention recently. To this end, there is a critical need to conduct a systematic review on this topic, identify the limitations of current work, and explore future directions. In this scoping review, we aim to provide a comprehensive overview of the current state of the field and identify key concepts, types of studies, and research gaps with a focus on biomedical images and texts joint learning, mainly because these two were the most commonly available data types in MDL research. This study reviewed the current uses of multimodal deep learning on five tasks: (1) Report generation, (2) Visual question answering, (3) Cross-modal retrieval, (4) Computer-aided diagnosis, and (5) Semantic segmentation. Our results highlight the diverse applications and potential of MDL and suggest directions for future research in the field. We hope our review will facilitate the collaboration of natural language processing (NLP) and medical imaging communities and support the next generation of decision-making and computer-assisted diagnostic system development.

Utilizing Longitudinal Chest X-Rays and Reports to Pre-Fill Radiology Reports

Despite the reduction in turn-around times in radiology reporting with the use of speech recognition software, persistent communication errors can significantly impact the interpretation of radiology reports. Pre-filling a radiology report holds promise in mitigating reporting errors, and despite multiple efforts in literature to generate comprehensive medical reports, there lacks approaches that exploit the longitudinal nature of patient visit records in the MIMIC-CXR dataset. To address this gap, we propose to use longitudinal multi-modal data, i.e., previous patient visit CXR, current visit CXR, and the previous visit report, to pre-fill the "findings" section of the patient's current visit. We first gathered the longitudinal visit information for 26,625 patients from the MIMIC-CXR dataset, and created a new dataset called Longitudinal-MIMIC. With this new dataset, a transformer-based model was trained to capture the multi-modal longitudinal information from patient visit records (CXR images + reports) via a cross-attention-based multi-modal fusion module and a hierarchical memory-driven decoder. In contrast to previous works that only uses current visit data as input to train a model, our work exploits the longitudinal information available to pre-fill the "findings" section of radiology reports. Experiments show that our approach outperforms several recent approaches. Code will be published at https://github.com/CelestialShine/Longitudinal-Chest-X-Ray.

Improving model fairness in image-based computer-aided diagnosis

Deep learning has become a popular tool for computer-aided diagnosis using medical images, sometimes matching or exceeding the performance of clinicians. However, these models can also reflect and amplify human bias, potentially resulting inaccurate missed diagnoses. Despite this concern, the problem of improving model fairness in medical image classification by deep learning has yet to be fully studied. To address this issue, we propose an algorithm that leverages the marginal pairwise equal opportunity to reduce bias in medical image classification. Our evaluations across four tasks using four independent large-scale cohorts demonstrate that our proposed algorithm not only improves fairness in individual and intersectional subgroups but also maintains overall performance. Specifically, the relative change in pairwise fairness difference between our proposed model and the baseline model was reduced by over 35%, while the relative change in AUC value was typically within 1%. By reducing the bias generated by deep learning models, our proposed approach can potentially alleviate concerns about the fairness and reliability of image-based computer-aided diagnosis.

Impacts of pregnancy and menopause on COVID-19 severity: a systematic review and meta-analysis of 4.6 million women

BACKGROUND

Corona Virus Disease 2019 (COVID-19) pandemic is still a public health emergency of international concern. However, whether pregnancy and menopause impact the severity of COVID-19 remain unclear.

AIM

This study is performed to investigate the truth.

DESIGN

Study appraisal and synthesis follows PRISMA guideline. Meta-analysis is performed in random-effects model.

METHODS

PubMed, Embase, Cochrane database, Central, CINAHL, ClinicalTrials.gov, WHO COVID-19 database and WHO-ICTRP are searched until 28 March 2023.

RESULTS

In total, 57 studies (4 640 275 COVID-19 women) were analyzed. Pregnant women were at a lower risk of severe COVID-19, intensive care unit (ICU) admission and disease mortality compared to those nonpregnant women with comparable comorbidities. In contrast, pregnant women with more prepregnancy comorbidities were at a higher risk of severe COVID-19, ICU admission and invasive mechanical ventilation (IMV). In addition, pregnant women with pregnancy complications had a significantly increased risk of severe COVID-19 and ICU admission. Menopause increased COVID-19 severity, IMV requirement and disease mortality. Hormone replacement therapy inhibited COVID-19 severity in postmenopausal women. Premenopausal and postmenopausal women had a lower chance of severe illness than age-matched men. The impact of pregnancy on COVID-19 severity was significant in Americans and Caucasians, whereas the effect of menopause on COVID-19 severity was only significant in Chinese.

CONCLUSIONS

Pregnancy and menopause are protective and risk factors for severe COVID-19, respectively. The protective role of pregnancy on COVID-19 is minimal and could be counteracted or masked by prepregnancy or pregnancy comorbidities. The administration of estrogen and progesterone may prevent severe COVID-19.

Impact of PRaG Therapy on Immune Microenvironment of Bilateral Subcutaneous Tumor Model of Colon Cancer

PURPOSE/OBJECTIVE(S)

Immune microenvironment is closely related to the efficacy of PD-1 inhibitors. The immune microenvironment contains a variety of immune cells, including effector T cells (cytotoxic CD8+T cells and effector CD4+T cells), dendritic cells (DC), and Myeloid-derived suppressor cells (MDSCs). The antitumor effects of PRaG therapy have been confirmed in bilateral subcutaneous transplantation tumor model of colon cancer. But the impact of PRaG therapy on immune microenvironment of such model is unclear. Therefore, the study continued to reveal the changes of immune microenvironment in mice.

MATERIALS/METHODS

80 male Balb/c mice aged 6-8 weeks were divided into five groups: control group, PD-1 inhibitor group, radiation group, radiation + PD-1 inhibitor group, and radiation + PD-1 inhibitor +GM-CSF (PRaG therapy) group. Bilateral subcutaneous tumor model of colon cancer in mice was constructed. 3×105 CT26.WT cells were inoculated subcutaneously in the right thigh root, and then the left thigh root 3 days later. Right subcutaneous tumor was selected for radiotherapy of 8 Gy×3. GM-CSF (100ng, i.p.) was given on the 1st day and PD-1 inhibitor (0.25mg/kg, i.p.) was given on the 2nd day after radiotherapy with one cycle every 3 days. On day 15, the spleen, left inguinal lymph node and left subcutaneous tumor of mice were collected. The proportion of immune cells was detected by flow cytometry.

RESULTS

Compared with other groups, PRaG therapy decreased the proportion of cDC1 in left inguinal lymph node, increased the proportion of cDC2 in left subcutaneous tumor and left inguinal lymph node. Moreover, PRaG therapy increased the proportion of CD8+ effector memory T cells and CD226+CD8+T cells in left inguinal lymph nodes. Finally, PRaG therapy increased the proportion of CD4+, CD8+ central memory T cells and CD69+CD8+T cells and reduced the proportion of M-MDSCs in spleen.

CONCLUSION

PRaG therapy can improve the immune microenvironment of spleen, unirradiated tumors and inguinal draining lymph nodes of bilateral subcutaneous tumor model of colon cancer in mice.

A scoping review on multimodal deep learning in biomedical images and texts

OBJECTIVE

Computer-assisted diagnostic and prognostic systems of the future should be capable of simultaneously processing multimodal data. Multimodal deep learning (MDL), which involves the integration of multiple sources of data, such as images and text, has the potential to revolutionize the analysis and interpretation of biomedical data. However, it only caught researchers' attention recently. To this end, there is a critical need to conduct a systematic review on this topic, identify the limitations of current work, and explore future directions.

METHODS

In this scoping review, we aim to provide a comprehensive overview of the current state of the field and identify key concepts, types of studies, and research gaps with a focus on biomedical images and texts joint learning, mainly because these two were the most commonly available data types in MDL research.

RESULT

This study reviewed the current uses of multimodal deep learning on five tasks: (1) Report generation, (2) Visual question answering, (3) Cross-modal retrieval, (4) Computer-aided diagnosis, and (5) Semantic segmentation.

CONCLUSION

Our results highlight the diverse applications and potential of MDL and suggest directions for future research in the field. We hope our review will facilitate the collaboration of natural language processing (NLP) and medical imaging communities and support the next generation of decision-making and computer-assisted diagnostic system development.

A Generalized Deep Learning Method for Synthetic CT Generation

PURPOSE/OBJECTIVE(S)

The application of deep learning to generate synthetic CT (sCT) has been widely studied in radiotherapy. Existing methods generally involve data from two different image modalities, such as CBCT-CT or MRI-CT, the quality of sCT is adversely affected by source image quality. We propose a unique method of synthesizing MRI and CBCT into sCT based on single-modal CT for training, and call it SmGAN.

MATERIALS/METHODS

We used planning CT of a group of 35 head and neck cases to as training data. We then applied two different spatial transformations to the planning CT image to produce the transformed CT1 and CT2. And We used a random style enhancement technique (Shuffle Remap) to modify the image distribution of CT1 which we termed CT1+E. CT1+E was used to simulate the patient's "image of the day" while CT2 to simulate the "planning image". After feeding both CT1+E and CT2 into the generator, we obtained the sCT predicted by the generator. The generator was trained using the Mean Absolute Error (MAE) loss between sCT and CT1. In the actual clinical process, we use the patient's CBCT or MRI instead of CT1+E and the patient's planning CT instead of CT2 as the input of the generator. After processing, we get an sCT that can maintain the spatial position of the image taken on the day, while presenting features similar to the planning CT. The evaluation data we have includes 10 pairs of MRI-Def_CT and 10 pairs of CBCT-Def_CT Head and Neck patients. Def_CT is obtained from the planning CT based on the spatial position deformation of MRI and CBCT. To evaluate the accuracy of sCT based on MRI and CBCT with Def CT, we use a range of metrics, including Hounsfield Unit (HU) difference, peak signal-to-noise ratio (PSNR), structural similarity (SSIM) and gamma pass rate. All results will be benchmarks against the advanced method RegGAN for comparison.

RESULTS

Compared to RegGAN, the results of SmGAN were significantly better. The mean absolute errors within the body were (44.7±216.2 HU vs. 36.7±131.4 HU) and (64.9±123.7 HU vs. 58.2±152.8 HU) for the CBCT-SCT and MRI-SCT, respectively (Table 1). In addition, experimental results show that SmGAN also outperforms RegGAN in dose calculation accuracy. For example, under the 10% threshold, SmGAN's gamma pass rate of 1mm and 1% is 0.926±0.02, compared with gamma rate of 0.896±0.02 for RegGAN.

CONCLUSION

We proposed a generalized deep learning model for synthetic CT generation, based on CBCT or MRI images. The proposed algorithm achieved high accuracy of dosimetric metrics, as well as excellent IMRT QA verification results. Compared to other existing synthetic CT generation methods, the proposed SmGAN required a single-modal image for training, which is considered as a major breakthrough in the industry, and is expected to have wide spread of clinical applications.

How Does Pruning Impact Long-Tailed Multi-label Medical Image Classifiers?

Pruning has emerged as a powerful technique for compressing deep neural networks, reducing memory usage and inference time without significantly affecting overall performance. However, the nuanced ways in which pruning impacts model behavior are not well understood, particularly for long-tailed, multi-label datasets commonly found in clinical settings. This knowledge gap could have dangerous implications when deploying a pruned model for diagnosis, where unexpected model behavior could impact patient well-being. To fill this gap, we perform the first analysis of pruning's effect on neural networks trained to diagnose thorax diseases from chest X-rays (CXRs). On two large CXR datasets, we examine which diseases are most affected by pruning and characterize class "forgettability" based on disease frequency and co-occurrence behavior. Further, we identify individual CXRs where uncompressed and heavily pruned models disagree, known as pruning-identified exemplars (PIEs), and conduct a human reader study to evaluate their unifying qualities. We find that radiologists perceive PIEs as having more label noise, lower image quality, and higher diagnosis difficulty. This work represents a first step toward understanding the impact of pruning on model behavior in deep long-tailed, multi-label medical image classification. All code, model weights, and data access instructions can be found at https://github.com/VITA-Group/PruneCXR.

Mitigation of Dosimetric Uncertainty in MRI-Based Proton Planning Using Spot-Scanning Proton Arc (SPArc) Technique

PURPOSE/OBJECTIVE(S)

MRI-based synthetic CT (SCT) images created using generative adversarial network (GAN) have been demonstrated to be feasible for intensity-modulated proton therapy (IMPT) planning. However, dose calculation accuracy can be uncertain in some regions within/near the target of head and neck patients due to the local CT number estimation error or sharp dose fall-off. This study investigated the feasibility of using the SPArc technique to mitigate such dosimetric uncertainty.

MATERIALS/METHODS

A GAN using a 3D U-net as the generator and a 6-layer 3D convolutional neural network as the discriminator was trained with T1-weighted MR-CT image pairs from 162 nasopharyngeal carcinoma patients (14 for validation). The generator was used to generate SCT images from MR images for 7 test patients. For each test patient, the CT image was used to create a SPArc plan and an IMPT plan with the same clinical objectives. The SPArc plans (control point frequency sampling, arc trajectory, etc.) were optimized using a previously developed iterative approach. The dose distributions of both SPArc plans and IMPT plans were re-calculated on the SCT images and compared to the one calculated on the CT images. The dosimetric uncertainty was quantified using the gamma index.

RESULTS

The 2%/2mm and 3%/3mm passing rates for SPArc plans were (96.9¡À2.7) % and (98.6¡À1.5) %, while the passing rates for IMPT plans were (94.0¡À3.9) % and (96.4+2.9) %. A significant reduction in dosimetric uncertainty was identified for SPArc plans (p ¡Ü0.021). Table 1 shows the passing rates for the 7 test individuals.

CONCLUSION

SPArc can mitigate the uncertainty of dose calculation in MRI-based proton planning. Further research needs to validate these findings on a larger patient cohort. The study paves the road map for using MRI for SPArc planning.

Multidetector CT-derived tricuspid annulus measurements predict tricuspid regurgitation reduction after transcatheter aortic valve replacement

AIM

To use multidetector row computed tomography (MDCT)-derived tricuspid annulus (TA) measurements to identify predictors for tricuspid regurgitation (TR) reduction after transcatheter aortic valve replacement (TAVR), and to investigate the impact of TR change on prognosis.

MATERIALS AND METHODS

A retrospective, single-centre study was conducted on consecutive patients who underwent TAVR with concomitant baseline mild or more severe TR from April 2012 to April 2022. TA parameters were measured using MDCT.

RESULTS

The study comprised 266 patients (mean age 74.2 ± 7.6 years, 147 men) and 45.1% had more than one grade of TR reduction at follow-up. Independent predictors of TR reduction at follow-up were distance between TA centroid and antero-septal commissure (odd ratio [OR] 0.776; 95% confidence interval [CI]: 0.672-0.896, p=0.001), baseline TR of moderate or worse (OR 4.599; 95% CI: 2.193-9.648, p<0.001), systolic pulmonary artery pressure (OR 1.018; 95% CI: 1.002-1.035, p=0.027), age (OR 0.955; 95% CI: 0.920-0.993, p=0.019), and pre-existing atrial fibrillation (OR 0.209; 95% CI: 0.101-0.433, p<0.001). Patients without TR reduction had higher rates of rehospitalisation (hazard ratio [HR] 0.642; 95% CI: 0.413-0.998, p=0.049).

CONCLUSIONS

The MDCT-derived TA parameter was predictive of TR reduction after TAVR. Persistent TR after TAVR was associated with higher rates of rehospitalisation.

Utilizing Longitudinal Chest X-Rays and Reports to Pre-fill Radiology Reports

Despite the reduction in turn-around times in radiology reporting with the use of speech recognition software, persistent communication errors can significantly impact the interpretation of radiology reports. Pre-filling a radiology report holds promise in mitigating reporting errors, and despite multiple efforts in literature to generate comprehensive medical reports, there lacks approaches that exploit the longitudinal nature of patient visit records in the MIMIC-CXR dataset. To address this gap, we propose to use longitudinal multi-modal data, i.e., previous patient visit CXR, current visit CXR, and the previous visit report, to pre-fill the "findings" section of the patient's current visit. We first gathered the longitudinal visit information for 26,625 patients from the MIMIC-CXR dataset, and created a new dataset called Longitudinal-MIMIC. With this new dataset, a transformer-based model was trained to capture the multi-modal longitudinal information from patient visit records (CXR images + reports) via a cross-attention-based multi-modal fusion module and a hierarchical memory-driven decoder. In contrast to previous works that only uses current visit data as input to train a model, our work exploits the longitudinal information available to pre-fill the "findings" section of radiology reports. Experiments show that our approach outperforms several recent approaches by ≥3% on F1 score, and ≥2% for BLEU-4, METEOR and ROUGE-L respectively. Code will be published at https://github.com/CelestialShine/Longitudinal-Chest-X-Ray.

Impact of PRaG Therapy on Peripheral Immune Cells of Subcutaneous Tumor Peritoneal Metastasis Model of Colon Cancer

PURPOSE/OBJECTIVE(S)

Immune cells in peripheral blood may be closely related to the efficacy of immune checkpoint inhibitors. T cells originally present in tumors may have limited antitumor effects, and T cells that respond to immune checkpoint inhibitors may be derived from peripheral blood. Therefore, in this study, subcutaneous tumor peritoneal metastasis model of colon cancer was constructed to reveal the changes of T cells and their subsets (CD4+T cells, CD8+T cells, CD226+T cells), MDSCs and their subsets (G-MDSCs, M-MDSCs) in peripheral blood of mice after PRaG therapy.

MATERIALS/METHODS

A total of 90 male Balb/c mice aged 6-8 weeks were divided into five groups: control group, PD-1 inhibitor group, radiation group, radiation + PD-1 inhibitor group, and radiation + PD-1 inhibitor +GM-CSF (PRaG therapy) group. The subcutaneous tumor peritoneal metastasis model of colon cancer was constructed. 3×105 CT26.WT cells was inoculated subcutaneously at the right thigh root, and 5 days later, 1×105 CT26.WT cells was inoculated on the left side at the junction of the anterior superior iliac spine and the midabdominal line. The subcutaneous tumor was selected for radiotherapy of 8 Gy×3. GM-CSF (100ng, i.p.) was given on the 1st day and PD-1 inhibitor (0.25mg/kg, i.p.) was given on the 2nd day after radiotherapy with one cycle every 3 days. On day 22, the peripheral blood of mice was collected. The proportion of immune cells was detected by flow cytometry.

RESULTS

Compared with other groups, PRaG therapy decreased the proportion of CD4+T cells and increased the proportion of CD8+T cells. Moreover, PRaG therapy increased the proportion of CD226+CD4+T cells and CD226+CD8+T cells. Finally, PRaG therapy increased the proportion of M-MDSCs and decreased the proportion of G-MDSCs.

CONCLUSION

PRaG therapy can improve the immune microenvironment of peripheral blood of subcutaneous tumor peritoneal metastasis model of colon cancer in mice.

Radiation-Induced Nasopharyngeal Necrosis in Locally-Recurrent Nasopharyngeal Carcinoma Patients after Re-Radiotherapy

PURPOSE/OBJECTIVE(S)

Re-radiotherapy (re-RT) is the main treatment for locally recurrent nasopharyngeal carcinoma (lrNPC) patients, and commonly led to radiation-induced nasopharyngeal (NP) necrosis, which was lethal but rare study has focused on it. The aim of this study was to evaluate the cause and impact of radiation-induced NP necrosis in lrNPC patients who received re-RT.

MATERIALS/METHODS

Totally 252 lrNPC patients who received re-RT between January 2013 and December 2020 were retrospectively collected. The inclusion criteria were as follows: (1) no NP necrosis before re-RT; (2) complete medical records, including treatment, clinical and dosimetric information; (3) conventional fractionated radiotherapy. All patients received intensity-modulated radiotherapy ± chemotherapy. Radiation-induced NP necrosis was diagnosed by magnetic resonance imaging and/or electronic nasopharyngoscopy. Dosimetric factors of the planning target volume of primary tumor (PTVp) were extracted from the dose-volume histogram (DVH), which was rescaled to an equivalent dose of 2 Gy per fraction (EQD 2 Gy) using a linear quadratic model. Logistic regression was used to identify the independent prognostic factors for generating the nomogram.

RESULTS

With a median follow-up of 44.63 months (inter-quartile range [IQR], 27.70 - 69.20 months), 47.6% of patients (120/252) occurred radiation-induced NP necrosis, which mostly happened within 1 year post re-RT (median [IQR], 5.83 [3.37 - 11.57] months). The 3-year overall survival was 83.0% vs 39.7% (P<0.001) in lrNPC patients with or without radiation-induced NP necrosis. Except for the fractionated dose, other dosimetric factors of PTVp were not significantly different between two groups, including D98 (dose to 98% of PTVp), D50, D2 and homogeneity index (Table 1). Furthermore, multivariate analysis showed that continuous variable age (HR [95% CI]: 1.04 [1.02 - 1.07], P = 0.003) and tumor volume (HR [95% CI]: 1.02 [1.01 - 1.03], P<0.001), and fractionated dose > 2.22 Gy (HR [95% CI]: 2.36 [1.32 - 4.21], P = 0.004) were independent factors in predicting radiation-induced NP necrosis, which yielded a C-index of 0.742 (95% CI, 0.682 - 0.803) for OS in the nomogram.

CONCLUSION

The incidence of radiation-induced NP necrosis was high in lrNPC patients who received re-RT. Patients with older age, larger tumor volume or receiving fractionated dose over 2.22 Gy were more easily to suffer NP necrosis, which need to explore novel treatment strategies to improve patients' survivals.

Exploring angular-steering illumination-based eyebox expansion for holographic displays

Holography represents an enabling technology for next-generation virtual and augmented reality systems. However, it remains challenging to achieve both wide field of view and large eyebox at the same time for holographic near-eye displays, mainly due to the essential étendue limitation of existing hardware. In this work, we present an approach to expanding the eyebox for holographic displays without compromising their underlying field of view. This is achieved by utilizing a compact 2D steering mirror to deliver angular-steering illumination beams onto the spatial light modulator in alignment with the viewer's eye movements. To facilitate the same image for the virtual objects perceived by the viewer when the eye moves, we explore an off-axis computational hologram generation scheme. Two bench-top holographic near-eye display prototypes with the proposed angular-steering scheme are developed, and they successfully showcase an expanded eyebox up to 8 mm × 8 mm for both VR- and AR-modes, as well as the capability of representing multi-depth holographic images.

Evaluating large language models on medical evidence summarization

Recent advances in large language models (LLMs) have demonstrated remarkable successes in zero- and few-shot performance on various downstream tasks, paving the way for applications in high-stakes domains. In this study, we systematically examine the capabilities and limitations of LLMs, specifically GPT-3.5 and ChatGPT, in performing zero-shot medical evidence summarization across six clinical domains. We conduct both automatic and human evaluations, covering several dimensions of summary quality. Our study demonstrates that automatic metrics often do not strongly correlate with the quality of summaries. Furthermore, informed by our human evaluations, we define a terminology of error types for medical evidence summarization. Our findings reveal that LLMs could be susceptible to generating factually inconsistent summaries and making overly convincing or uncertain statements, leading to potential harm due to misinformation. Moreover, we find that models struggle to identify the salient information and are more error-prone when summarizing over longer textual contexts.

[Preliminary experience of transcatheter pulmonary valve replacement using domestic balloon-expandable valve]

Objectives: To evaluate the feasibility and preliminary clinical results of transcatheter pulmonary valve replacement (TPVR) with the domestically-produced balloon-expandable Prizvalve system. Methods: This is a prospective single-center observational study. Patients with postoperative right ventricular outflow tract (RVOT) dysfunction, who were admitted to West China Hospital of Sichuan University from September 2021 to March 2023 and deemed anatomically suitable for TPVR with balloon-expandable valve, were included. Clinical, imaging, procedural and follow-up data were analyzed. The immediate procedural results were evaluated by clinical implant success rate, which is defined as successful valve implantation with echocardiography-assessed pulmonary regurgitation Collapse

Key Words Collapse

MESH Headings

• Male
• Humans
• Pulmonary Valve/surgery
• Heart Valve Prosthesis/adverse effects
• Heart Valve Prosthesis Implantation
• Constriction, Pathologic/complications
• Constriction, Pathologic/surgery
• Prospective Studies
• Ventricular Outflow Obstruction/etiology
• Ventricular Outflow Obstruction/surgery
• Treatment Outcome
• Cardiac Catheterization/methods
• Transcatheter Aortic Valve Replacement

Collapse

Grants

• 2022ZDZX0030 Sichuan Science and Technology Program

Collapse

Affiliation(s)

Z G Zhao Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China
R T Li Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China
X Wei Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China
Y Peng Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China
J F Wei Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China
S He Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China
Q Li Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China
X Li Department of Cardiovascular Surgery, West China Hospital, Sichuan University, Chengdu 610041, China
Y J Li Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China
X Li Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China
X Zhou Department of Radiology, West China Hospital, Sichuan University, Chengdu 610041, China
M X Zheng Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China
G Chen Department of Anesthesiology, West China Hospital, Sichuan University, Chengdu 610041, China
Q An Department of Cardiovascular Surgery, West China Hospital, Sichuan University, Chengdu 610041, China
M Chen Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China
Y Feng Department of Cardiology, West China Hospital, Sichuan University, Chengdu 610041, China

Collapse

In-plane reorientation induced single laser pulse magnetization reversal

Single Pulse All Optical Switching represents the ability to reverse the magnetization of a nanostructure using a femtosecond single laser pulse without any applied field. Since the first switching experiments carried out on GdFeCo ferrimagnets, this phenomena has been only recently extended to a few other materials, MnRuGa alloys and Tb/Co multilayers with a very specific range of thickness and composition. Here, we demonstrate that single pulse switching can be obtained for a large range of rare earth-transition metal multilayers, making this phenomenon much more general. Surprisingly, the threshold fluence for switching is observed to be independent of the laser pulse duration. Moreover, at high laser intensities, concentric ring domain structures are induced. These striking features contrast to those observed in Gd based materials pointing towards a different reversal mechanism. Concomitant with the demonstration of an in-plane magnetization reorientation, a precessional reversal mechanism explains all the observed features.

How Does Pruning Impact Long-Tailed Multi-Label Medical Image Classifiers?

Pruning has emerged as a powerful technique for compressing deep neural networks, reducing memory usage and inference time without significantly affecting overall performance. However, the nuanced ways in which pruning impacts model behavior are not well understood, particularly for long-tailed, multi-label datasets commonly found in clinical settings. This knowledge gap could have dangerous implications when deploying a pruned model for diagnosis, where unexpected model behavior could impact patient well-being. To fill this gap, we perform the first analysis of pruning's effect on neural networks trained to diagnose thorax diseases from chest X-rays (CXRs). On two large CXR datasets, we examine which diseases are most affected by pruning and characterize class "forgettability" based on disease frequency and co-occurrence behavior. Further, we identify individual CXRs where uncompressed and heavily pruned models disagree, known as pruning-identified exemplars (PIEs), and conduct a human reader study to evaluate their unifying qualities. We find that radiologists perceive PIEs as having more label noise, lower image quality, and higher diagnosis difficulty. This work represents a first step toward understanding the impact of pruning on model behavior in deep long-tailed, multi-label medical image classification. All code, model weights, and data access instructions can be found at https://github.com/VITA-Group/PruneCXR.

Internet use predicts Chinese character spelling performance of junior high school students: multiple mediating roles of pinyin input proficiency and net-speak experience

To examine the complex relationship between Internet use experience and character spelling performance among Chinese junior high school students, the study explored the multiple mediating roles of Pinyin input proficiency and net-speak experience. A total of 447 Chinese junior high school students aged 12-15 years old completed the Internet Use Experience and Pinyin Input Proficiency Assessment, the Net-speak Experience Questionnaire and the Chinese Spelling Test. The results showed that: (1) All investigated variables were significantly correlated with each other, but there was no direct relationship between Internet use and Chinese spelling performance. (2) Pinyin input proficiency and net-speak experience play a chain mediating role in the relationship between Internet use and Chinese character spelling performance. Teens' Internet use experience indirectly and positively predicted Chinese character spelling performance through the mediation of Pinyin input method use and net-speak experience. The implication of this study is that Chinese children should be guided to engage in Internet activities that require Pinyin typing and use net-speak creatively in order to promote the traditional Chinese character spelling skills when instructing teenagers to engage in Internet activities.

An NLP approach to identify SDoH-related circumstance and suicide crisis from death investigation narratives

OBJECTIVES

Suicide presents a major public health challenge worldwide, affecting people across the lifespan. While previous studies revealed strong associations between Social Determinants of Health (SDoH) and suicide deaths, existing evidence is limited by the reliance on structured data. To resolve this, we aim to adapt a suicide-specific SDoH ontology (Suicide-SDoHO) and use natural language processing (NLP) to effectively identify individual-level SDoH-related social risks from death investigation narratives.

MATERIALS AND METHODS

We used the latest National Violent Death Report System (NVDRS), which contains 267 804 victim suicide data from 2003 to 2019. After adapting the Suicide-SDoHO, we developed a transformer-based model to identify SDoH-related circumstances and crises in death investigation narratives. We applied our model retrospectively to annotate narratives whose crisis variables were not coded in NVDRS. The crisis rates were calculated as the percentage of the group's total suicide population with the crisis present.

RESULTS

The Suicide-SDoHO contains 57 fine-grained circumstances in a hierarchical structure. Our classifier achieves AUCs of 0.966 and 0.942 for classifying circumstances and crises, respectively. Through the crisis trend analysis, we observed that not everyone is equally affected by SDoH-related social risks. For the economic stability crisis, our result showed a significant increase in crisis rate in 2007-2009, parallel with the Great Recession.

CONCLUSIONS

This is the first study curating a Suicide-SDoHO using death investigation narratives. We showcased that our model can effectively classify SDoH-related social risks through NLP approaches. We hope our study will facilitate the understanding of suicide crises and inform effective prevention strategies.

[Construction of a dual fluorescent reporter system for tracing horizontal transfer of mcr-1-carrying plasmid]

The green fluorescent reporter gene was inserted into the gene interval of polymyxin resistant mcr-1-carrying plasmid (pSH13G841) by homologous recombination of suicide plasmid. At the same time, E. coli J53 with red fluorescent reporter gene was constructed. Using the ability of spontaneous conjugation of drug resistant plasmid (pSH13G841), pSH13G841-GFP plasmid was transferred into J53 RFP bacteria to construct a double fluorescent labeled donor bacterium. The two light-emitting systems could stably and spontaneously express fluorescence without mutual interference. The dual fluorescence report system constructed can be used for visual tracing horizontal transfer of mcr-1-carrying plasmid, the subsequent model can study the colonization, transfer and prognosis of drug-resistant bacteria/drug-resistant genes mcr-1 by using mouse in vivo imaging technology.

Faithful AI in Medicine: A Systematic Review with Large Language Models and Beyond

Artificial intelligence (AI), especially the most recent large language models (LLMs), holds great promise in healthcare and medicine, with applications spanning from biological scientific discovery and clinical patient care to public health policymaking. However, AI methods have the critical concern for generating factually incorrect or unfaithful information, posing potential long-term risks, ethical issues, and other serious consequences. This review aims to provide a comprehensive overview of the faithfulness problem in existing research on AI in healthcare and medicine, with a focus on the analysis of the causes of unfaithful results, evaluation metrics, and mitigation methods. We systematically reviewed the recent progress in optimizing the factuality across various generative medical AI methods, including knowledge-grounded LLMs, text-to-text generation, multimodality-to-text generation, and automatic medical fact-checking tasks. We further discussed the challenges and opportunities of ensuring the faithfulness of AI-generated information in these applications. We expect that this review will assist researchers and practitioners in understanding the faithfulness problem in AI-generated information in healthcare and medicine, as well as the recent progress and challenges in related research. Our review can also serve as a guide for researchers and practitioners who are interested in applying AI in medicine and healthcare.

Less Likely Brainstorming: Using Language Models to Generate Alternative Hypotheses

A human decision-maker benefits the most from an AI assistant that corrects for their biases. For problems such as generating interpretation of a radiology report given findings, a system predicting only highly likely outcomes may be less useful, where such outcomes are already obvious to the user. To alleviate biases in human decision-making, it is worth considering a broad differential diagnosis, going beyond the most likely options. We introduce a new task, "less likely brainstorming," that asks a model to generate outputs that humans think are relevant but less likely to happen. We explore the task in two settings: a brain MRI interpretation generation setting and an everyday commonsense reasoning setting. We found that a baseline approach of training with less likely hypotheses as targets generates outputs that humans evaluate as either likely or irrelevant nearly half of the time; standard MLE training is not effective. To tackle this problem, we propose a controlled text generation method that uses a novel contrastive learning strategy to encourage models to differentiate between generating likely and less likely outputs according to humans. We compare our method with several state-of-the-art controlled text generation models via automatic and human evaluations and show that our models' capability of generating less likely outputs is improved.

[Feasibility and safety of transseptal transcatheter mitral valve replacement for severe mitral regurgitation]

A prospective, single-center, single-arm, and open-design study was performed to evaluate the feasibility and safety of transseptal transcatheter mitral valve replacement in the treatment of severe mitral regurgitation. Patients with symptomatic moderate-severe or severe mitral regurgitation at high-surgical risk and anatomically appropriate for the HighLife transseptal mitral valve replacement (TSMVR) system in West China Hospital, Sichuan University from December 2021 to August 2022 were enrolled. Four patients (1 male and 3 females) with severe mitral regurgitation were included, with a median age of 68.5 (64.0-77.0) years and a median Society of Thoracic Surgeons (STS) score of 8.1% (6.4%-8.9%). Technical success was achieved in all the patients. There was no residual mitral regurgitation, paravalvular leakage, or left ventricular outflow tract obstruction. Three major cardiovascular and cerebrovascular adverse events occurred within 30 days after the procedure, including ventricular tachycardia, iatrogenic atrial septal defect closure, and heart failure readmission. The current study preliminarily demonstrates that transcatheter mitral valve replacement using the HighLife system via the transseptal approach for severe mitral regurgitation is feasible and relatively safe.

A new deep learning framework to process Matrix-assisted Laser Desorption/Ionisation Mass Spectrometry Imaging (MALDI-MSI) data of Tissue Microarrays (TMAs)

Matrix-Assisted Laser Desorption Ionization mass spectrometry imaging (MALDI-MSI) is a mass spectrometry ionization technique that can be used to directly analyze tissues and has led the way in the development of biological and clinical applications for imaging mass spectrometry. One of its advantages is measuring the distribution of a large number of analytes at one time without destroying the sample, making it a useful method in tissue-based studies. However, analysis of the MALDI-MSI images from tissue microarrays (TMAs) remains less studied. While several automated systems have been developed for tissue classification (e.g., cancer vs non-cancer), they process the MALDI data at the measuring point level, which ignores spatial relationships among individual points within the tissue sample. In this work, we propose mNet, a new deep learning framework to analyze MALDI-MSI data of TMAs at the tissue-needle-core level to ensure that the samples maintain their original spatial context. In addition, we introduced data augmentation techniques to increase data size which is often limited in biomedical data. We applied our framework to analyzing TMAs from breast and lung cancer. We found that our framework outperforms conventional machine learning methods in the challenging race detection task. The results highlight the potential of deep learning to assist pathologists in analyzing tissue specimens in a label-free, high-throughput manner.

Evaluate underdiagnosis and overdiagnosis bias of deep learning model on primary open-angle glaucoma diagnosis in under-served populations

In the United States, primary open-angle glaucoma (POAG) is the leading cause of blindness, especially among African American and Hispanic individuals. Deep learning has been widely used to detect POAG using fundus images as its performance is comparable to or even surpasses diagnosis by clinicians. However, human bias in clinical diagnosis may be reflected and amplified in the widely-used deep learning models, thus impacting their performance. Biases may cause (1) underdiagnosis, increasing the risks of delayed or inadequate treatment, and (2) overdiagnosis, which may increase individuals' stress, fear, well-being, and unnecessary/costly treatment. In this study, we examined the underdiagnosis and overdiagnosis when applying deep learning in POAG detection based on the Ocular Hypertension Treatment Study (OHTS) from 22 centers across 16 states in the United States. Our results show that the widely-used deep learning model can underdiagnose or overdiagnose under-served populations. The most underdiagnosed group is female younger (< 60 yrs) group, and the most overdiagnosed group is Black older (≥ 60 yrs) group. Biased diagnosis through traditional deep learning methods may delay disease detection, treatment and create burdens among under-served populations, thereby, raising ethical concerns about using deep learning models in ophthalmology clinics.

Microplastic formation and simultaneous release of phthalic acid esters from residual mulch film in soil through mechanical abrasion

The application of plastic mulch film could effectively enhance the productivity of facility agriculture. However, releasing microplastic and phthalate from mulch films in soil has attracted increasing concerns, and releasing characters of microplastic and phthalate from mulch films during their mechanical abrasion remains unclear. This study elucidated the dynamics and impact factors of microplastic generation, including the thickness, polymer types and ageing of mulch film during mechanical abrasion. Releasing characters of the di(2-Ethylhexyl) phthalate (DEHP), a common type of phthalate in soil, from mulch film during mechanical abrasion were also explored. Results showed that 2 pieces of mulch film debris increased to 1291 pieces of microplastic after five days of mechanical abrasion, with exponential growth in the microplastic generation. After mechanical abrasion, the thinnest (0.008 mm) mulch film completely transformed into microplastics. However, the thicker mulch (>0.01 mm) suffered slight disintegration, making it feasible to be recycled. The biodegradable mulch film discharged the most microplastics (906 pieces) compared with the HDPE (359 pieces) and LDPE (703 pieces) mulch film after three days of mechanical abrasion. In addition, the mild thermal and oxidative ageing could result in 3047 and 4532 pieces of microplastic emissions from mulch film after three days of mechanical abrasion, which were ten times more than the original mulch film (359 pieces). Furthermore, negligible DEHP was released from mulch film without mechanical abrasion, while the released DEHP significantly correlated with generated microplastics during mechanical abrasion. These results demonstrated the crucial role of mulch film disintegration in phthalate emissions.

[Corneal melt after intracorneal ring segment implantations: a case report]

A 50-year-old female patient presented to the Beijing Tongren Eye Center, Beijing Tongren Hospital, Capital Medical University, Beijing Institute of Ophthalmology, with complaints of right eye pain, tearing, and difficulty opening the eye for over a month after intrastromal corneal ring segment (ICRS) implantation 18 years prior in both eyes. Slit lamp examination revealed corneal stromal melting around the ICRS at the 3 to 4 o'clock position of the right eye, with fluorescein staining. Optical coherence tomography showed epithelial and superficial stromal layer defects in the area of the lesion. The patient was diagnosed with corneal melting after ICRS implantation in the right eye. Under general anesthesia, the corneal stromal ring was removed, and deep lamellar keratoplasty was performed. The patient had no discomfort and the corneal graft remained transparent after the surgery.

[Prevalence of parasitic infections in human stool samples from a hospital in Chenzhou City of Hunan Province]

OBJECTIVE

To investigate the prevalence of parasitic infections in human stool samples from a hospital in Chenzhou City, Hunan Province, so as to provide insights into the management of intestinal parasitic diseases.

METHODS

Stool samples were collected from patients admitted to a hospital in Chenzhou City from September 2020 to March 2021, subjected to physiological saline smearing and microscopy for detection of intestinal parasites. The prevalence of parasitic infections and the species of parasites were descriptively analyzed.

RESULTS

The overall prevalence of intestinal parasitic infections was 1.61% in the 10 728 stool samples, and there were 3 samples with mixed infections of two parasite species. A total of seven parasite species were identified, including Blastocystis hominis (162 cases, 1.55%), Giardia lamblia (5 cases, 0.05%), Dientamoeba fragilis (5 cases, 0.05%), Endolimax nana (one case, 0.01%), Iodamoeba bütschlii (one case, 0.01%), Strongyloides stercoralis (one case, 0.01%) and Trichomonas hominis (one case, 0.01%). The prevalence of intestinal parasitic infection was significantly higher among women than in men (2.14% vs. 1.25%; χ² = 13.01, P < 0.01), and a high prevalence rate was seen among patients at ages of 20 to 30 years (2.99%) and 80 years and older (2.86%); however, no age-specific prevalence of intestinal parasitic infection was detected (χ² = 12.45, P > 0.05).

CONCLUSIONS

The overall prevalence of intestinal parasitic infection was low among patients admitted to a hospital in Chenzhou City, and gender-specific prevalence was found. Food-borne and opportunistic parasites were predominant intestinal parasites, including B. hominis, G. lamblia and D. fragilis.

Representing and utilizing clinical textual data for real world studies: An OHDSI approach

Clinical documentation in electronic health records contains crucial narratives and details about patients and their care. Natural language processing (NLP) can unlock the information conveyed in clinical notes and reports, and thus plays a critical role in real-world studies. The NLP Working Group at the Observational Health Data Sciences and Informatics (OHDSI) consortium was established to develop methods and tools to promote the use of textual data and NLP in real-world observational studies. In this paper, we describe a framework for representing and utilizing textual data in real-world evidence generation, including representations of information from clinical text in the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM), the workflow and tools that were developed to extract, transform and load (ETL) data from clinical notes into tables in OMOP CDM, as well as current applications and specific use cases of the proposed OHDSI NLP solution at large consortia and individual institutions with English textual data. Challenges faced and lessons learned during the process are also discussed to provide valuable insights for researchers who are planning to implement NLP solutions in real-world studies.