pathCLIP: Detection of Genes and Gene Relations from Biological Pathway Figures through Image-Text Contrastive Learning.

In biomedical literature, biological pathways are commonly described through a combination of images and text. These pathways contain valuable information, including genes and their relationships, which provide insight into biological mechanisms and precision medicine. Curating pathway information across the literature enables the integration of this information to build a comprehensive knowledge base. While some studies have extracted pathway information from images and text independently, they often overlook the correspondence between the two modalities. In this paper, we present a pathway figure curation system named pathCLIP for identifying genes and gene relations from pathway figures. Our key innovation is the use of an image-text contrastive learning model to learn coordinated embeddings of image snippets and text descriptions of genes and gene relations, thereby improving curation. Our validation results, using pathway figures from PubMed, showed that our multimodal model outperforms models using only a single modality. Additionally, our system effectively curates genes and gene relations from multiple literature sources. Two case studies on extracting pathway information from literature of non-small cell lung cancer and Alzheimer's disease further demonstrate the usefulness of our curated pathway information in enhancing related pathways in the KEGG database.

An alternative processing approach to increase CD138 intensity in flow cytometric analysis of plasma cells.

BACKGROUND: Surface median immunofluorescence intensity (MFI) of plasma cells antigens, particularly CD138, by flow cytometry underestimates plasma cell populations when compared with that estimated by morphological assessment on Wright's-stained slides. CD138 MFI using traditional sample preparation methods for flow cytometric analysis is often dim and difficult to interpret due to multiple factors. This becomes critical when diagnosing and accurately classifying plasma cell dyscrasias. METHODS: In this study, we analyzed 280 flow cytometric results collected from 2016 to 2022 for CD38 and CD138 MFI on bone marrow aspirates performed by two different methods of sample processing-traditional method of lyse-wash and the alternative method of lyse-no-wash. RESULTS: Visual examination of histograms showed a clear advantage to CD138 expression intensity with the no-wash method. Although no significant difference was observed in CD38 MFI between the two techniques (p = 0.3), considerable improvement was observed in CD138 MFI with the lyse-no-wash technique of sample processing compared with the conventional method (p = 0.003). CONCLUSIONS: We concluded that the method of lyse-no-wash is superior to traditional methods especially when it comes to handling bone marrow aspirate samples for plasma cell immunophenotyping. This alternate technique increases the sensitivity of flow cytometry to detect plasma cells resulting in bright and crisp signal intensity for surface CD138. This technique may be particularly advantageous when analyzing low tumor burden such as minimal residual disease.

pathCLIP: Detection of Genes and Gene Relations from Biological Pathway Figures through Image-Text Contrastive Learning.

In biomedical literature, biological pathways are commonly described through a combination of images and text. These pathways contain valuable information, including genes and their relationships, which provide insight into biological mechanisms and precision medicine. Curating pathway information across the literature enables the integration of this information to build a comprehensive knowledge base. While some studies have extracted pathway information from images and text independently, they often overlook the correspondence between the two modalities. In this paper, we present a pathway figure curation system named pathCLIP for identifying genes and gene relations from pathway figures. Our key innovation is the use of an image-text contrastive learning model to learn coordinated embeddings of image snippets and text descriptions of genes and gene relations, thereby improving curation. Our validation results, using pathway figures from PubMed, showed that our multimodal model outperforms models using only a single modality. Additionally, our system effectively curates genes and gene relations from multiple literature sources. A case study on extracting pathway information from non-small cell lung cancer literature further demonstrates the usefulness of our curated pathway information in enhancing related pathways in the KEGG database.

Applications of cutting-edge artificial intelligence technologies in biomedical literature and document mining.

The biomedical literature is a vast and invaluable resource for biomedical research. Integrating knowledge from the literature with biomedical data can help biological studies and the clinical decision-making process. Efforts have been made to gather information from the biomedical literature and create biomedical knowledge bases, such as KEGG and Reactome. However, manual curation remains the primary method to retrieve accurate biomedical entities and relationships. Manual curation becomes increasingly challenging and costly as the volume of biomedical publications quickly grows. Fortunately, recent advancements in Artificial Intelligence (AI) technologies offer the potential to automate the process of curating, updating, and integrating knowledge from the literature. Herein, we highlight the AI capabilities to aid in mining knowledge and building the knowledge base from the biomedical literature.

Dipeptidylpeptidase 4 promotes survival and stemness of acute myeloid leukemia stem cells.

Leukemic-stem-cell-specific targeting may improve the survival of patients with acute myeloid leukemia (AML) by avoiding the ablative effects of standard regimens on normal hematopoiesis. Herein, we perform an unbiased screening of compounds targeting cell surface proteins and identify clinically used DPP4 inhibitors as strong suppressors of AML development in both murine AML models and primary human AML cells xenograft model. We find in retrovirus-induced AML mouse models that DPP4-deficient AML cell-transplanted mice exhibit delay and reversal of AML development, whereas deletion of DPP4 has no significant effect on normal hematopoiesis. DPP4 activates and sustains survival of AML stem cells that are critical for AML development in both human and animal models via binding with Src kinase and activation of nuclear factor κB (NF-κB) signaling. Thus, inhibition of DPP4 is a potential therapeutic strategy against AML development through suppression of survival and stemness of AML cells.

Adherence to guidelines-recommended diagnostic testing was associated with overall survival in patients with diffuse large B-cell lymphoma after rituximab-based treatment: an observational cohort study.

PURPOSE: This study assessed the impact of adherence to guidelines-recommended diagnostic testing on treatment selection and overall survival (OS) in patients with diffuse large B-cell lymphoma (DLBCL) initiated on rituximab-based first line of treatment (1-LOT). METHODS: This retrospective cohort study used a nationwide electronic health record-derived de-identified database, including diagnostic testing information on immunohistochemistry (IHC), fluorescence in situ hybridization (FISH) and karyotype analysis that were abstracted from pathology reports or clinical visit notes, where available. The study included patients above 18 years old who were diagnosed with DLBCL between January 2011 and December 2019 and initiated on rituximab-based 1-LOT. Patients were classified into 'non-adherence,' 'partial-adherence' and 'complete-adherence' groups according to the evidence/documentation of a confirmed known result for IHC and molecular profiling tests (FISH and karyotyping) on a selection of the markers prior to the initiation of 1-LOT. Logistic regression was used to evaluate associations of adherence to diagnostic testing with 1-LOT between R-CHOP and other rituximab-based regimens. Median OS after the start of rituximab-based 1-LOT was calculated using the Kaplan-Meier method. Multivariable-adjusted Cox proportional hazards regression was used to assess the risk of all-cause death after initiation of 1-LOT by the degrees of adherence to guidelines-recommended diagnostic testing. RESULTS: In total, 3730 patients with DLBCL who initiated on rituximab-based 1-LOT were included. No association was found between adherence to guidelines-recommended diagnostic testing and treatment selection of 1-LOT for R-CHOP versus other rituximab-based regimens. Patients with a higher degree of adherence to guidelines-recommended diagnostic testing survived longer (median OS at 5.1, 6.9 and 7.1 years for 'non-adherence,' 'partial-adherence' and 'complete-adherence' groups, respectively [log-rank p < 0.001]) and had a decreased mortality risk (multivariable-adjusted hazard ratio with 95% confidence intervals at 0.83 [0.70-0.99] for 'partial-adherence' and 0.77 [0.64-0.91] for 'complete-adherence' groups, respectively). CONCLUSION: Patients' adherence to guidelines-recommended diagnostic testing were associated with better survival benefit, reinforcing the need for adoption of diagnostic testing guidelines in routine clinical care.

Rural populations facilitated early SARS-CoV-2 evolution and transmission in Missouri, USA.

In the United States, rural populations comprise 60 million individuals and suffered from high COVID-19 disease burdens. Despite this, surveillance efforts are biased toward urban centers. Consequently, how rurally circulating SARS-CoV-2 viruses contribute toward emerging variants remains poorly understood. In this study, we aim to investigate the role of rural communities in the evolution and transmission of SARS-CoV-2 during the early pandemic. We collected 544 urban and 435 rural COVID-19-positive respiratory specimens from an overall vaccine-naïve population in Southwest Missouri between July and December 2020. Genomic analyses revealed 53 SARS-CoV-2 Pango lineages in our study samples, with 14 of these lineages identified only in rural samples. Phylodynamic analyses showed that frequent bi-directional diffusions occurred between rural and urban communities in Southwest Missouri, and that four out of seven Missouri rural-origin lineages spread globally. Further analyses revealed that the nucleocapsid protein (N):R203K/G204R paired substitutions, which were detected disproportionately across multiple Pango lineages, were more associated with urban than rural sequences. Positive selection was detected at N:204 among rural samples but was not evident in urban samples, suggesting that viruses may encounter distinct selection pressures in rural versus urban communities. This study demonstrates that rural communities may be a crucial source of SARS-CoV-2 evolution and transmission, highlighting the need to expand surveillance and resources to rural populations for COVID-19 mitigation.

Iterative Prompt Refinement for Mining Gene Relationships from ChatGPT.

ChatGPT has demonstrated its potential as a surrogate knowledge graph. Trained on extensive data sources, including open-access publications, peer-reviewed research articles and biomedical websites, ChatGPT extracted information on gene relationships and biological pathways. However, a major challenge is model hallucination, i.e., high false positive rates. To assess and address this challenge, we systematically evaluated ChatGPT's capacity for predicting gene relationships using GPT-3.5-turbo and GPT-4. Benchmarking against the KEGG Pathway Database as the ground truth, we experimented with diverse prompting strategies, targeting gene relationships of activation, inhibition, and phosphorylation. We introduced an innovative iterative prompt refinement technique. By assessing prompt efficacy using metrics like F-1 score, precision, and recall, GPT-4 was re-engaged to suggest improved prompts. A refined prompt, which combines a specialized role with explanatory text, significantly enhances the performance. Going beyond pairwise gene relationships, we also deciphered complex gene interplays, such as gene interaction chains and pathways pertinent to diseases like non-small cell lung cancer. Direct prompts showed limited success, but "least-to-most" prompting exhibited significant potentials for such network constructions. The methods in this study may be used for some other bioinformatics prediction problems.

Immune-Related Gene Signatures to Predict the Effectiveness of Chemoimmunotherapy in Triple-Negative Breast Cancer Using Exploratory Subgroup Discovery.

Triple-negative breast cancer (TNBC) is an aggressive subtype of breast cancer with limited therapeutic options. Although immunotherapy has shown potential in TNBC patients, clinical studies have only demonstrated a modest response. Therefore, the exploration of immunotherapy in combination with chemotherapy is warranted. In this project we identified immune-related gene signatures for TNBC patients that may explain differences in patients' outcomes after anti-PD-L1+chemotherapy treatment. First, we ran the exploratory subgroup discovery algorithm on the TNBC dataset comprised of 422 patients across 24 studies. Secondly, we narrowed down the search to twelve homogenous subgroups based on tumor mutational burden (TMB, low or high), relapse status (disease-free or recurred), tumor cellularity (high, low and moderate), menopausal status (pre- or post) and tumor stage (I, II and III). For each subgroup we identified a union of the top 10% of genotypic patterns. Furthermore, we employed a multinomial regression model to predict significant genotypic patterns that would be linked to partial remission after anti-PD-L1+chemotherapy treatment. Finally, we uncovered distinct immune cell populations (T-cells, B-cells, Myeloid, NK-cells) for TNBC patients with various treatment outcomes. CD4-Tn-LEF1 and CD4-CXCL13 T-cells were linked to partial remission on anti-PD-L1+chemotherapy treatment. Our informatics pipeline may help to select better responders to chemoimmunotherapy, as well as pinpoint the underlying mechanisms of drug resistance in TNBC patients at single-cell resolution.

Early expression of mature αß TCR in CD4-CD8- T cell progenitors enables MHC to drive development of T-ALL bearing NOTCH mutations.

During normal T cell development in mouse and human, a low-frequency population of immature CD4-CD8- double-negative (DN) thymocytes expresses early, mature αß T cell antigen receptor (TCR). We report that these early αß TCR+ DN (EADN) cells are DN3b-DN4 stage and require CD3δ but not major histocompatibility complex (MHC) for their generation/detection. When MHC - is present, however, EADN cells can respond to it, displaying a degree of coreceptor-independent MHC reactivity not typical of mature, conventional αß T cells. We found these data to be connected with observations that EADN cells were susceptible to T cell acute lymphoblastic leukemia (T-ALL) transformation in both humans and mice. Using the OT-1 TCR transgenic system to model EADN-stage αß TCR expression, we found that EADN leukemogenesis required MHC to induce development of T-ALL bearing NOTCH1 mutations. This leukemia-driving MHC requirement could be lost, however, upon passaging the tumors in vivo, even when matching MHC was continuously present in recipient animals and on the tumor cells themselves. These data demonstrate that MHC:TCR signaling can be required to initiate a cancer phenotype from an understudied developmental state that appears to be represented in the mouse and human disease spectrum.

Investigation of the Utility of Features in a Clinical De-identification Model: A Demonstration Using EHR Pathology Reports for Advanced NSCLC Patients.

BACKGROUND: Electronic health record (EHR) systems contain a large volume of texts, including visit notes, discharge summaries, and various reports. To protect the confidentiality of patients, these records often need to be fully de-identified before circulating for secondary use. Machine learning (ML) based named entity recognition (NER) model has emerged as a popular technique of automatic de-identification. OBJECTIVE: The performance of a machine learning model highly depends on the selection of appropriate features. The objective of this study was to investigate the usability of multiple features in building a conditional random field (CRF) based clinical de-identification NER model. METHODS: Using open-source natural language processing (NLP) toolkits, we annotated protected health information (PHI) in 1,500 pathology reports and built supervised NER models using multiple features and their combinations. We further investigated the dependency of a model's performance on the size of training data. RESULTS: Among the 10 feature extractors explored in this study, n-gram, prefix-suffix, word embedding, and word shape performed the best. A model using combination of these four feature sets yielded precision, recall, and F1-score for each PHI as follows: NAME (0.80; 0.79; 0.80), LOCATION (0.85; 0.83; 0.84), DATE (0.86; 0.79; 0.82), HOSPITAL (0.96; 0.93; 0.95), ID (0.99; 0.82; 0.90), and INITIALS (0.97; 0.49; 0.65). We also found that the model's performance becomes saturated when the training data size is beyond 200. CONCLUSION: Manual de-identification of large-scale data is an impractical procedure since it is time-consuming and subject to human errors. Analysis of the NER model's performance in this study sheds light on a semi-automatic clinical de-identification pipeline for enterprise-wide data warehousing.

SARS-CoV-2 show no infectivity at later stages in a prolonged COVID-19 patient despite positivity in RNA testing.

Inpatient coronavirus disease 2019 (COVID-19) cases present enormous costs to patients and health systems in the United States. Many hospitalized patients may continue testing COVID-19 positive even after the resolution of symptoms. Thus, a pressing concern for clinicians is the safety of discharging these asymptomatic patients if they have any remaining infectivity. This case report explores the viral viability in a patient with persistent COVID-19 over the course of a 2-month hospitalization. Positive nasopharyngeal swab samples were collected and isolated in the laboratory and analyzed by quantitative reverse-transcription polymerase chain reactions (qRT-PCR), and serology was tested for neutralizing antibodies throughout the hospitalization period. The patient experienced waning symptoms by hospital day 40 and had no viable virus growth by hospital day 41, suggesting no risk of infectivity, despite positive RT-PCR results which prolonged his hospital stay. Notably, this case showed infectivity for at least 24 days after disease onset, which is longer than the discontinuation of transmission-based precautions recommended by the Center for Disease Control and Prevention. Thus, our findings suggest that the timeline for discontinuing transmission-based precautions may need to be extended for patients with severe and prolonged COVID-19 disease. Additional large-scale studies are needed to draw definitive conclusions on the appropriate clinical management for these patients. ​.

Rapid High-Throughput Whole-Genome Sequencing of SARS-CoV-2 by Using One-Step Reverse Transcription-PCR Amplification with an Integrated Microfluidic System and Next-Generation Sequencing.

The long-lasting global COVID-19 pandemic demands timely genomic investigation of SARS-CoV-2 viruses. Here, we report a simple and efficient workflow for whole-genome sequencing utilizing one-step reverse transcription-PCR (RT-PCR) amplification on a microfluidic platform, followed by MiSeq amplicon sequencing. The method uses Fluidigm integrated fluidic circuit (IFC) and instruments to amplify 48 samples with 39 pairs of primers, including 35 custom-designed primer pairs and four additional primer pairs from the ARTIC network protocol v3. Application of this method on RNA samples from both viral isolates and clinical specimens demonstrates robustness and efficiency in obtaining the full genome sequence of SARS-CoV-2.

The Value of Artificial Intelligence in Laboratory Medicine.

OBJECTIVES: As laboratory medicine continues to undergo digitalization and automation, clinical laboratorians will likely be confronted with the challenges associated with artificial intelligence (AI). Understanding what AI is good for, how to evaluate it, what are its limitations, and how it can be implemented are not well understood. With a survey, we aimed to evaluate the thoughts of stakeholders in laboratory medicine on the value of AI in the diagnostics space and identify anticipated challenges and solutions to introducing AI. METHODS: We conducted a web-based survey on the use of AI with participants from Roche's Strategic Advisory Network that included key stakeholders in laboratory medicine. RESULTS: In total, 128 of 302 stakeholders responded to the survey. Most of the participants were medical practitioners (26%) or laboratory managers (22%). AI is currently used in the organizations of 15.6%, while 66.4% felt they might use it in the future. Most had an unsure attitude on what they would need to adopt AI in the diagnostics space. High investment costs, lack of proven clinical benefits, number of decision makers, and privacy concerns were identified as barriers to adoption. Education in the value of AI, streamlined implementation and integration into existing workflows, and research to prove clinical utility were identified as solutions needed to mainstream AI in laboratory medicine. CONCLUSIONS: This survey demonstrates that specific knowledge of AI in the medical community is poor and that AI education is much needed. One strategy could be to implement new AI tools alongside existing tools.

Digital Tumor Board Solutions Have Significant Impact on Case Preparation.

PURPOSE: Multidisciplinary tumor boards (TBs) are the gold standard for decision-making in cancer care. Variability in preparation, conduction, and impact is widely reported. The benefit of digital technologies to support TBs is unknown. This study evaluated the impact of the NAVIFY Tumor Board solution (NTB) on TB preparation time across multiple user groups in 4 cancer categories: breast, GI, head and neck (ie, ear, nose, and throat, or ENT), and hematopathology. METHODS: This prospective study evaluated TB preparation time in multiple phases pre- and post-NTB implementation at an academic health care center. TB preparation times were recorded for multiple weeks using a digital time tracker. RESULTS: Preparation times for 59 breast, 61 GI, 36 ENT, and 71 hematopathology cancer TBs comparing a pre-NTB phase to 3 phases of NTB implementation were evaluated between February 2018 and July 2019. NTB resulted in significant reductions in overall preparation time (30%) across 3 TBs pre-NTB compared with the final post-NTB implementation phase. In the breast TB, NTB reduced overall preparation time by 28%, with a 76% decrease in standard deviation (SD). In the GI TB, a 23% reduction in average preparation time was observed for all users, with a 48% decrease in SD. In the ENT TB, a 33% reduction in average preparation time was observed for all users, with a 73% decrease in SD. The hematopathology TB, which was the cocreation partner and initial adopter of the solution, showed variable results. CONCLUSION: This study showed a significant impact of a digital solution on time preparation for TBs across multiple users and different TBs, reflecting the generalizability of the NTB. Adoption of such a solution could improve the efficiency of TBs and have a direct economic impact on hospitals.

Artificial Intelligence-Driven Structurization of Diagnostic Information in Free-Text Pathology Reports.

BACKGROUND: Free-text sections of pathology reports contain the most important information from a diagnostic standpoint. However, this information is largely underutilized for computer-based analytics. The vast majority of NLP-based methods lack a capacity to accurately extract complex diagnostic entities and relationships among them as well as to provide an adequate knowledge representation for downstream data-mining applications. METHODS: In this paper, we introduce a novel informatics pipeline that extends open information extraction (openIE) techniques with artificial intelligence (AI) based modeling to extract and transform complex diagnostic entities and relationships among them into Knowledge Graphs (KGs) of relational triples (RTs). RESULTS: Evaluation studies have demonstrated that the pipeline's output significantly differs from a random process. The semantic similarity with original reports is high (Mean Weighted Overlap of 0.83). The precision and recall of extracted RTs based on experts' assessment were 0.925 and 0.841 respectively (P <0.0001). Inter-rater agreement was significant at 93.6% and inter-rated reliability was 81.8%. CONCLUSION: The results demonstrated important properties of the pipeline such as high accuracy, minimality and adequate knowledge representation. Therefore, we conclude that the pipeline can be used in various downstream data-mining applications to assist diagnostic medicine.

A clinician's perspective on co-developing and co-implementing a digital tumor board solution.

Healthcare has entered the information age. This will deliver huge opportunities for healthcare providers to deliver more individualized treatments for patients, and as such improve outcomes. Nowhere is the prospect greater than in cancer care. Healthcare providers now need to manage the challenge of how to best capture, interpret and exploit insights from real-world clinical data. A significant aspect of cancer care is the challenge of preparing and conducting tumor boards. Currently, data are distributed across multiple systems and cannot be easily aggregated or integrated. In recognition that no suitable solution existed, the University of Missouri School of Medicine, in partnership with Roche, have co-developed and co-implemented a digital tumor board solution. This article describes the development process and the enablers and barriers for adoption from a clinician's perspective. In addition, it reflects on some of the key factors for success and some of the future opportunities.

The utility of thromboelastography and tranexamic acid in plasminogen activator inhibitor deficiency during pregnancy: a rare case report.

: Complete plasminogen activator inhibitor-1 (PAI-1) deficiency is a very rare genetic disorder that is associated with an increased risk of bleeding diathesis. Patients with PAI-1 deficiency are also known to be at increased risk for massive postpartum hemorrhage. We discuss one such rare case of PAI-1 deficiency in a young pregnant patient at 22 weeks of gestation with history of prolonged bleeding. Tranexamic acid was administered for menorrhagia and resumed later for labor and continued into the postpartum period since antifibrinolytics have been the mainstay in the management of PAI-1 deficiency. The patient delivered a healthy infant at 39 weeks. As PAI-1 deficiency causes increased fibrinolysis, the patient's coagulation panel was monitored by performing serial thromboelastograms to monitor for any increase in fibrinolysis. We believe that thromboelastograms might be a useful tool in the monitoring and management of fibrinolytic conditions such as PAI-1 deficiency.

Localized accumulation of kappa restricted Russell body-containing plasma cells in tonsil.

An abnormal clonal plasma cell proliferation with Russell bodies is rare in chronic inflammatory reactions in adult patients. We describe the first case of light chain restricted Russell body accumulation within germinal centers of lymphoid follicles of the tonsil in a child. This should not be confused with a neoplastic process.