Operational encoding enhances action knowledge integration: insights from event-related potential analysis.

In this study, we conducted an examination of knowledge integration concerning action information and assessed the impact of operational on this process. Additionally, we delved into the underlying mechanisms of how operational encoding influences the processing of knowledge integration of action information, utilizing the event-related potential technique. The results of our investigation revealed that operational encoding, encompassing the observed operational encoding and the imagined operational encoding, exhibited superior performance in the integration of action knowledge compared to verbal encoding. This distinction may be attributed to the greater efficiency of operant encoding in activating motor cortical areas, thereby inducing more robust brain activity. These findings suggest the potential advantages of operational encoding in facilitating the integration of knowledge related to movement information at both cognitive and neural levels, underscoring its significant role in the processing of such information. Future studies can further explore the applications of operational encoding in domains, such as motor learning, skill training, and rehabilitation therapy. Such investigations may offer novel insights into enhancing human behavior and motor control.

A Hybrid Framework for Maritime Surveillance: Detecting Illegal Activities through Vessel Behaviors and Expert Rules Fusion.

Maritime traffic is essential for global trade but faces significant challenges, including navigation safety, environmental protection, and the prevention of illicit activities. This work presents a framework for detecting illegal activities carried out by vessels, combining navigation behavior detection models with rules based on expert knowledge. Using synthetic and real datasets based on the Automatic Identification System (AIS), we structured our framework into five levels based on the Joint Directors of Laboratories (JDL) model, efficiently integrating data from multiple sources. Activities are classified into four categories: illegal fishing, suspicious activity, anomalous activity, and normal activity. To address the issue of a lack of labels and integrate data-driven detection with expert knowledge, we employed a stack ensemble model along with active learning. The results showed that the framework was highly effective, achieving 99% accuracy in detecting illegal fishing and 92% in detecting suspicious activities. Furthermore, it drastically reduced the need for manual checks by specialists, transforming experts' tacit knowledge into explicit knowledge through the models and allowing continuous updates of maritime domain rules. This work significantly contributes to maritime surveillance, offering a scalable and efficient solution for detecting illegal activities in the maritime domain.

A systematic review of biologically-informed deep learning models for cancer: fundamental trends for encoding and interpreting oncology data.

BACKGROUND: There is an increasing interest in the use of Deep Learning (DL) based methods as a supporting analytical framework in oncology. However, most direct applications of DL will deliver models with limited transparency and explainability, which constrain their deployment in biomedical settings. METHODS: This systematic review discusses DL models used to support inference in cancer biology with a particular emphasis on multi-omics analysis. It focuses on how existing models address the need for better dialogue with prior knowledge, biological plausibility and interpretability, fundamental properties in the biomedical domain. For this, we retrieved and analyzed 42 studies focusing on emerging architectural and methodological advances, the encoding of biological domain knowledge and the integration of explainability methods. RESULTS: We discuss the recent evolutionary arch of DL models in the direction of integrating prior biological relational and network knowledge to support better generalisation (e.g. pathways or Protein-Protein-Interaction networks) and interpretability. This represents a fundamental functional shift towards models which can integrate mechanistic and statistical inference aspects. We introduce a concept of bio-centric interpretability and according to its taxonomy, we discuss representational methodologies for the integration of domain prior knowledge in such models. CONCLUSIONS: The paper provides a critical outlook into contemporary methods for explainability and interpretability used in DL for cancer. The analysis points in the direction of a convergence between encoding prior knowledge and improved interpretability. We introduce bio-centric interpretability which is an important step towards formalisation of biological interpretability of DL models and developing methods that are less problem- or application-specific.

PriPath: identifying dysregulated pathways from differential gene expression via grouping, scoring, and modeling with an embedded feature selection approach.

BACKGROUND: Cell homeostasis relies on the concerted actions of genes, and dysregulated genes can lead to diseases. In living organisms, genes or their products do not act alone but within networks. Subsets of these networks can be viewed as modules that provide specific functionality to an organism. The Kyoto encyclopedia of genes and genomes (KEGG) systematically analyzes gene functions, proteins, and molecules and combines them into pathways. Measurements of gene expression (e.g., RNA-seq data) can be mapped to KEGG pathways to determine which modules are affected or dysregulated in the disease. However, genes acting in multiple pathways and other inherent issues complicate such analyses. Many current approaches may only employ gene expression data and need to pay more attention to some of the existing knowledge stored in KEGG pathways for detecting dysregulated pathways. New methods that consider more precompiled information are required for a more holistic association between gene expression and diseases. RESULTS: PriPath is a novel approach that transfers the generic process of grouping and scoring, followed by modeling to analyze gene expression with KEGG pathways. In PriPath, KEGG pathways are utilized as the grouping function as part of a machine learning algorithm for selecting the most significant KEGG pathways. A machine learning model is trained to differentiate between diseases and controls using those groups. We have tested PriPath on 13 gene expression datasets of various cancers and other diseases. Our proposed approach successfully assigned biologically and clinically relevant KEGG terms to the samples based on the differentially expressed genes. We have comparatively evaluated the performance of PriPath against other tools, which are similar in their merit. For each dataset, we manually confirmed the top results of PriPath in the literature and found that most predictions can be supported by previous experimental research. CONCLUSIONS: PriPath can thus aid in determining dysregulated pathways, which applies to medical diagnostics. In the future, we aim to advance this approach so that it can perform patient stratification based on gene expression and identify druggable targets. Thereby, we cover two aspects of precision medicine.

MedKPL: A heterogeneous knowledge enhanced prompt learning framework for transferable diagnosis.

Artificial Intelligence (AI) based diagnosis systems have emerged as powerful tools to reform traditional medical care. Each clinician now wants to have his own intelligent diagnostic partner to expand the range of services he can provide. However, the implementation of intelligent decision support systems based on clinical note has been hindered by the lack of extensibility of end-to-end AI diagnosis algorithms. When reading a clinical note, expert clinicians make inferences with relevant medical knowledge, which serve as prompts for making accurate diagnoses. Therefore, external medical knowledge is commonly employed as an augmentation for medical text classification tasks. Existing methods, however, cannot integrate knowledge from various knowledge sources as prompts nor can fully utilize explicit and implicit knowledge. To address these issues, we propose a Medical Knowledge-enhanced Prompt Learning (MedKPL) diagnostic framework for transferable clinical note classification. Firstly, to overcome the heterogeneity of knowledge sources, such as knowledge graphs or medical QA databases, MedKPL uniform the knowledge relevant to the disease into text sequences of fixed format. Then, MedKPL integrates medical knowledge into the prompt designed for context representation. Therefore, MedKPL can integrate knowledge into the models to enhance diagnostic performance and effectively transfer to new diseases by using relevant disease knowledge. The results of our experiments on two medical datasets demonstrate that our method yields superior medical text classification results and performs better in cross-departmental transfer tasks under few-shot or even zero-shot settings. These findings demonstrate that our MedKPL framework has the potential to improve the interpretability and transferability of current diagnostic systems.

[Inter- and Transdisciplinarity as a Normative Dynamic: Challenges and Opportunities for Political Science]. / Inter- und Transdisziplinarität als normative Dynamik: Herausforderungen und Chancen für die Politikwissenschaft.

Inter- and transdisciplinarity (ITD) has been part of political science for quite some time now, but although political science regularly deals with its self-understanding, the consequences for research and researchers of ITD have not yet been systematically considered. To stimulate this debate, we conceptualize ITD as a spectrum of knowledge integration, application, and participation. We use International Relations norm research as a theoretical framework to describe, analyze, and reflect on ITD as a normative dynamic. Autoethnographically and through participatory observation, we examine ITD as a normative dynamic, with insights from three research projects in the field of sustainability. Specifically, we ask what implications ITD has for researchers and research in political science. As a result, we find that ITD offers both opportunities and challenges. In the context of knowledge integration, we discuss the importance of the participation of political science in major societal issues in contrast to ITD's preferences for a particular understanding of knowledge and research. We reflect on ITD's application bias in terms of problem-solving opportunities and output orientation. In addition, we consider the participation postulate of ITD and weigh potential democratizing effects against the conditions under which these might be realized. Finally, we address where further research seems useful to continue reflection on ITD.

RTX-KG2: a system for building a semantically standardized knowledge graph for translational biomedicine.

BACKGROUND: Biomedical translational science is increasingly using computational reasoning on repositories of structured knowledge (such as UMLS, SemMedDB, ChEMBL, Reactome, DrugBank, and SMPDB in order to facilitate discovery of new therapeutic targets and modalities. The NCATS Biomedical Data Translator project is working to federate autonomous reasoning agents and knowledge providers within a distributed system for answering translational questions. Within that project and the broader field, there is a need for a framework that can efficiently and reproducibly build an integrated, standards-compliant, and comprehensive biomedical knowledge graph that can be downloaded in standard serialized form or queried via a public application programming interface (API). RESULTS: To create a knowledge provider system within the Translator project, we have developed RTX-KG2, an open-source software system for building-and hosting a web API for querying-a biomedical knowledge graph that uses an Extract-Transform-Load approach to integrate 70 knowledge sources (including the aforementioned core six sources) into a knowledge graph with provenance information including (where available) citations. The semantic layer and schema for RTX-KG2 follow the standard Biolink model to maximize interoperability. RTX-KG2 is currently being used by multiple Translator reasoning agents, both in its downloadable form and via its SmartAPI-registered interface. Serializations of RTX-KG2 are available for download in both the pre-canonicalized form and in canonicalized form (in which synonyms are merged). The current canonicalized version (KG2.7.3) of RTX-KG2 contains 6.4M nodes and 39.3M edges with a hierarchy of 77 relationship types from Biolink. CONCLUSION: RTX-KG2 is the first knowledge graph that integrates UMLS, SemMedDB, ChEMBL, DrugBank, Reactome, SMPDB, and 64 additional knowledge sources within a knowledge graph that conforms to the Biolink standard for its semantic layer and schema. RTX-KG2 is publicly available for querying via its API at arax.rtx.ai/api/rtxkg2/v1.2/openapi.json . The code to build RTX-KG2 is publicly available at github:RTXteam/RTX-KG2 .

Advantages and challenges of fostering cognitive integration through virtual collaborative learning: a qualitative study.

BACKGROUND: The drastic shift from face-to-face classes to online learning due to the COVID-19 pandemic has enabled educators to ensure the continuity of learning for health professions students in higher education. Collaborative learning, a pedagogy used to facilitate knowledge integration by helping students translate theory from basic sciences to clinical application and practice, has thus been transformed from a face-to-face to a virtual strategy to achieve the learning objectives of a multi-disciplinary and integrated module. OBJECTIVES: This study aimed to describe and evaluate, through focus group discussions, a virtual collaborative learning activity implemented to assist first year undergraduate nursing students to develop cognitive integration in a module consisting of pathophysiology, pharmacology, and nursing practice. METHODS: Fourteen first year undergraduate students and four faculty involved in facilitating the virtual collaboration participated in the study. Focus group discussions were conducted to elicit the perceptions of students and staff on the virtual collaborative learning session conducted at the end of the semester. RESULTS: Three themes were generated from the thematic analysis of the students' focus group scripts. These were: (1) achieving engagement and interaction, (2) supporting the collaborative process, and (3) considering practical nuances. The three themes were further subdivided into subthemes to highlight noteworthy elements captured during focus group discussions. Three themes also emerged from the focus group discussion scripts of faculty participants: (1) learning to effectively manage, (2) facing engagement constraints, and (3) achieving integration. These themes were further sectioned into salient subthemes. CONCLUSION: The virtual collaborative learning pedagogy is valuable in fostering cognitive integration. However, meticulous planning considering various variables prior to implementation is needed. With better planning directed at addressing the learners' needs and the faculty's capabilities and readiness for online learning pedagogies, and with a strong institutional support to help mitigate the identified constraints of virtual collaborative learning, students and faculty will benefit.

Beyond the divide between indigenous and academic knowledge: Causal and mechanistic explanations in a Brazilian fishing community.

Transdisciplinary research challenges the divide between Indigenous and academic knowledge by bringing together epistemic resources of heterogeneous stakeholders. The aim of this article is to explore causal explanations in a traditional fishing community in Brazil that provide resources for transdisciplinary collaboration, without neglecting differences between Indigenous and academic experts. Semi-structured interviews were carried out in a fishing village in the North shore of Bahia and our findings show that community members often rely on causal explanations for local ecological phenomena with different degrees of complexity. While these results demonstrate the ecological expertise of local community members, we also argue that recognition of local expertise needs to reflect on differences between epistemic communities by developing a culturally sensitive model of transdisciplinary knowledge negotiation.

Self-directed Science Learning During COVID-19 and Beyond.

Prompted by the sudden shift to remote instruction in March 2020 brought on by the COVID-19 pandemic, teachers explored online resources to support their students learning from home. We report on how twelve teachers identified and creatively leveraged open educational resources (OERs) and practices to facilitate self-directed science learning. Based on interviews and logged data, we illustrate how teachers' use of OER starkly differed from the typical uses of technology for transmitting information or increasing productivity. These experiences provide insights into ways teachers and professional developers can take advantage of OER to promote self-directed learning when in-person instruction resumes.

Integrating trans health knowledge through instructional design: preparing learners for a continent - not an island - of primary care with trans people.

In recent years the need to teach primary care providers to better care for transgender and non-binary (trans) patients has garnered significant scholarly and public attention. The alarming why motivating this surge in trans health primary care education has already been firmly established and needs no further comment. Instead, we offer new perspectives on how to do trans health primary care education. From treasured 'trans 101' educational interventions to trans health 'clinical pearls', the prevailing model used to teach primary care learners represents time-limited cultural competency-based education, which we argue creates an isolated education 'island'. In rethinking this approach, we present an introduction to the concepts of knowledge integration and the transfer of learning and apply them to show how trans health knowledge and skills should be structured within existing curricula to support effective learning and application. These instructional design considerations have yet to be extensively explored when teaching primary care learners trans health content and may be critical to building pedagogy that ultimately improves healthcare delivery. We conclude that trans health - and trans patients themselves - must not be treated as an isolated education island of knowledge and practice. Rather, it is the responsibility of educators to design instruction that encourages learners to integrate this knowledge with foundational principles of primary care; building bridges across a continent of primary care practice landscapes in turn.

DynOVis: a web tool to study dynamic perturbations for capturing dose-over-time effects in biological networks.

BACKGROUND: The development of high throughput sequencing techniques provides us with the possibilities to obtain large data sets, which capture the effect of dynamic perturbations on cellular processes. However, because of the dynamic nature of these processes, the analysis of the results is challenging. Therefore, there is a great need for bioinformatics tools that address this problem. RESULTS: Here we present DynOVis, a network visualization tool that can capture dynamic dose-over-time effects in biological networks. DynOVis is an integrated work frame of R packages and JavaScript libraries and offers a force-directed graph network style, involving multiple network analysis methods such as degree threshold, but more importantly, it allows for node expression animations as well as a frame-by-frame view of the dynamic exposure. Valuable biological information can be highlighted on the nodes in the network, by the integration of various databases within DynOVis. This information includes pathway-to-gene associations from ConsensusPathDB, disease-to-gene associations from the Comparative Toxicogenomics databases, as well as Entrez gene ID, gene symbol, gene synonyms and gene type from the NCBI database. CONCLUSIONS: DynOVis could be a useful tool to analyse biological networks which have a dynamic nature. It can visualize the dynamic perturbations in biological networks and allows the user to investigate the changes over time. The integrated data from various online databases makes it easy to identify the biological relevance of nodes in the network. With DynOVis we offer a service that is easy to use and does not require any bioinformatics skills to visualize a network.

Advancing the science of unfinished nursing care: Exploring the benefits of cross-disciplinary knowledge exchange, knowledge integration and transdisciplinarity.

AIMS: The aims of this paper are to explore the role of cross-disciplinary knowledge exchange and integration in advancing the science of unfinished nursing care and to offer preliminary guidance for theory development activities for this growing international community of scholars. BACKGROUND: Unfinished nursing care, also known as missed care or rationed care is a highly prevalent problem with negative consequences for patients, nurses and healthcare organizations around the world. It presents as a 'wicked' sustainability problem resulting from structural obstacles to effective resource allocation that have been resistant to conventional solutions. Research activity related to this problem is on the rise internationally but is hindered by inconsistencies in conceptualizations of the problem and lack of robust theory development around the phenomenon. A unified conceptual framework is needed to focus scholarly activities and facilitate advancement of a robust science of unfinished nursing care. DESIGN: Discussion paper. DATA SOURCES: This discussion paper is based on our own experiences in international and interdisciplinary research partnerships related to unfinished nursing care. These experiences are placed in the context of both classic and current literature related to the evolution of scientific knowledge. IMPLICATIONS FOR NURSING: The problem of unfinished nursing care crosses multiple scientific disciplines. It is imperative that the community of scholars interested in solving this wicked problem engage in meaningful cross-disciplinary knowledge integration and move towards transdisciplinarity. CONCLUSION: Metatheorizing guided by structuration theory should be considered as a strategy to promote transdiciplinarity around the problem of unfinished nursing care.

Phelan-McDermid syndrome data network: Integrating patient reported outcomes with clinical notes and curated genetic reports.

The heterogeneity of patient phenotype data are an impediment to the research into the origins and progression of neuropsychiatric disorders. This difficulty is compounded in the case of rare disorders such as Phelan-McDermid Syndrome (PMS) by the paucity of patient clinical data. PMS is a rare syndromic genetic cause of autism and intellectual deficiency. In this paper, we describe the Phelan-McDermid Syndrome Data Network (PMS_DN), a platform that facilitates research into phenotype-genotype correlation and progression of PMS by: a) integrating knowledge of patient phenotypes extracted from Patient Reported Outcomes (PRO) data and clinical notes-two heterogeneous, underutilized sources of knowledge about patient phenotypes-with curated genetic information from the same patient cohort and b) making this integrated knowledge, along with a suite of statistical tools, available free of charge to authorized investigators on a Web portal https://pmsdn.hms.harvard.edu. PMS_DN is a Patient Centric Outcomes Research Initiative (PCORI) where patients and their families are involved in all aspects of the management of patient data in driving research into PMS. To foster collaborative research, PMS_DN also makes patient aggregates from this knowledge available to authorized investigators using distributed research networks such as the PCORnet PopMedNet. PMS_DN is hosted on a scalable cloud based environment and complies with all patient data privacy regulations. As of October 31, 2016, PMS_DN integrates high-quality knowledge extracted from the clinical notes of 112 patients and curated genetic reports of 176 patients with preprocessed PRO data from 415 patients.

Model Informed Pediatric Development Applied to Bilastine: Ontogenic PK Model Development, Dose Selection for First Time in Children and PK Study Design.

PURPOSE: Bilastine is an H1 antagonist whose pharmacokinetics (PK) and pharmacodynamics (PD) have been resolved in adults with a therapeutic oral dose of 20 mg/day. Bilastine has favorable characteristics for use in pediatrics but the PK/PD and the optimal dose in children had yet to be clinically explored. The purpose is to: (1) Develop an ontogenic predictive model of bilastine PK linked to the PD in adults by integrating current knowledge; (2) Use the model to design a PK study in children; (3) Confirm the selected dose and the study design through the evaluation of model predictability in the first recruited children; (4) Consider for inclusion the group of younger children (< 6 years). METHODS: A semi-mechanistic approach was applied to predict bilastine PK in children assuming the same PD as described in adults. The model was used to simulate the time evolution of plasma levels and wheal and flare effects after several doses and design an adaptive PK trial in children that was then confirmed using data from the first recruits by comparing observations with model predictions. RESULTS: PK/PD simulations supported the selection of 10 mg/day in 2 to <12 year olds. Results from the first interim analysis confirmed the model predictions and design hence trial continuation. CONCLUSION: The model successfully predicted bilastine PK in pediatrics and optimally assisted the selection of the dose and sampling scheme for the trial in children. The selected dose was considered suitable for younger children and the forthcoming safety study in children aged 2 to <12 years.

Integrating research evidence and physical activity policy making-REPOPA project.

Evidence shows that regular physical activity is enhanced by supporting environment. Studies are needed to integrate research evidence into health enhancing, cross-sector physical activity (HEPA) policy making. This article presents the rationale, study design, measurement procedures and the initial results of the first phase of six European countries in a five-year research project (2011-2016), REsearch into POlicy to enhance Physical Activity (REPOPA). REPOPA is programmatic research; it consists of linked studies; the first phase studied the use of evidence in 21 policies in implementation to learn more in depth from the policy making process and carried out 86 qualitative stakeholder interviews. The second, ongoing phase builds on the central findings of the first phase in each country; it consists of two sets of interventions: game simulations to study cross-sector collaboration and organizational change processes in the use of evidence and locally tailored interventions to increase knowledge integration. The results of the first two study phases will be tested and validated among policy makers and other stakeholders in the third phase using a Delphi process. Initial results from the first project phase showed the lack of explicit evidence use in HEPA policy making. Facilitators and barriers of the evidence use were the availability of institutional resources and support but also networking between researchers and policy makers. REPOPA will increase understanding use of research evidence in different contexts; develop guidance and tools and establish sustainable structures such as networks and platforms between academics and policy makers across relevant sectors.

Evolution of the "drivers" of translational cancer epidemiology: analysis of funded grants and the literature.

Concurrently with a workshop sponsored by the National Cancer Institute, we identified key "drivers" for accelerating cancer epidemiology across the translational research continuum in the 21st century: emerging technologies, a multilevel approach, knowledge integration, and team science. To map the evolution of these "drivers" and translational phases (T0-T4) in the past decade, we analyzed cancer epidemiology grants funded by the National Cancer Institute and published literature for 2000, 2005, and 2010. For each year, we evaluated the aims of all new/competing grants and abstracts of randomly selected PubMed articles. Compared with grants based on a single institution, consortium-based grants were more likely to incorporate contemporary technologies (P = 0.012), engage in multilevel analyses (P = 0.010), and incorporate elements of knowledge integration (P = 0.036). Approximately 74% of analyzed grants and publications involved discovery (T0) or characterization (T1) research, suggesting a need for more translational (T2-T4) research. Our evaluation indicated limited research in 1) a multilevel approach that incorporates molecular, individual, social, and environmental determinants and 2) knowledge integration that evaluates the robustness of scientific evidence. Cancer epidemiology is at the cusp of a paradigm shift, and the field will need to accelerate the pace of translating scientific discoveries in order to impart population health benefits. While multi-institutional and technology-driven collaboration is happening, concerted efforts to incorporate other key elements are warranted for the discipline to meet future challenges.

Participatory Causal Loop Diagrams Building for Supporting Decision-Makers Integrating Flood Risk Management in an Urban Regeneration Process.

Several modeling tools commonly used for supporting flood risk assessment and management are highly effective in representing physical phenomena, but provide a rather limited understanding of the multiple implications that flood risk and flood risk reduction measures have on highly complex systems such as urban areas. In fact, most of the available modeling tools do not fully account for this complexity-and related uncertainty-which heavily affects the interconnections between urban systems evolution and flood risk, ultimately resulting in an ineffective flood risk management. The present research proposes an innovative methodological framework to support decision-makers involved in an urban regeneration process at a planning/strategic level, accounting for the multi-dimensional implications of flood risk and of different flood risk management strategies. The adopted approach is based on the use of System Thinking principles and participatory System Dynamics modeling techniques, and pursues an integration between scientific and stakeholder knowledge. Reference is made to one of the case studies of the CUSSH and CAMELLIA projects, namely Thamesmead (London), a formerly inhospitable marshland currently undergoing a process of urban regeneration, and perceived as being increasingly vulnerable to flooding. It represents an interesting opportunity for building a replicable modeling approach to integrate urban development dynamics with flood risk, ultimately supporting policy and decision-makers in identifying mitigation/prevention measures and understanding how they could help achieve multi-dimensional benefits (e.g., environmental, social and economic).

AI-powered microscopy image analysis for parasitology: integrating human expertise.

Microscopy image analysis plays a pivotal role in parasitology research. Deep learning (DL), a subset of artificial intelligence (AI), has garnered significant attention. However, traditional DL-based methods for general purposes are data-driven, often lacking explainability due to their black-box nature and sparse instructional resources. To address these challenges, this article presents a comprehensive review of recent advancements in knowledge-integrated DL models tailored for microscopy image analysis in parasitology. The massive amounts of human expert knowledge from parasitologists can enhance the accuracy and explainability of AI-driven decisions. It is expected that the adoption of knowledge-integrated DL models will open up a wide range of applications in the field of parasitology.

Enhancing Clinical Relevance of Pretrained Language Models Through Integration of External Knowledge: Case Study on Cardiovascular Diagnosis From Electronic Health Records.

BACKGROUND: Despite their growing use in health care, pretrained language models (PLMs) often lack clinical relevance due to insufficient domain expertise and poor interpretability. A key strategy to overcome these challenges is integrating external knowledge into PLMs, enhancing their adaptability and clinical usefulness. Current biomedical knowledge graphs like UMLS (Unified Medical Language System), SNOMED CT (Systematized Medical Nomenclature for Medicine-Clinical Terminology), and HPO (Human Phenotype Ontology), while comprehensive, fail to effectively connect general biomedical knowledge with physician insights. There is an equally important need for a model that integrates diverse knowledge in a way that is both unified and compartmentalized. This approach not only addresses the heterogeneous nature of domain knowledge but also recognizes the unique data and knowledge repositories of individual health care institutions, necessitating careful and respectful management of proprietary information. OBJECTIVE: This study aimed to enhance the clinical relevance and interpretability of PLMs by integrating external knowledge in a manner that respects the diversity and proprietary nature of health care data. We hypothesize that domain knowledge, when captured and distributed as stand-alone modules, can be effectively reintegrated into PLMs to significantly improve their adaptability and utility in clinical settings. METHODS: We demonstrate that through adapters, small and lightweight neural networks that enable the integration of extra information without full model fine-tuning, we can inject diverse sources of external domain knowledge into language models and improve the overall performance with an increased level of interpretability. As a practical application of this methodology, we introduce a novel task, structured as a case study, that endeavors to capture physician knowledge in assigning cardiovascular diagnoses from clinical narratives, where we extract diagnosis-comment pairs from electronic health records (EHRs) and cast the problem as text classification. RESULTS: The study demonstrates that integrating domain knowledge into PLMs significantly improves their performance. While improvements with ClinicalBERT are more modest, likely due to its pretraining on clinical texts, BERT (bidirectional encoder representations from transformer) equipped with knowledge adapters surprisingly matches or exceeds ClinicalBERT in several metrics. This underscores the effectiveness of knowledge adapters and highlights their potential in settings with strict data privacy constraints. This approach also increases the level of interpretability of these models in a clinical context, which enhances our ability to precisely identify and apply the most relevant domain knowledge for specific tasks, thereby optimizing the model's performance and tailoring it to meet specific clinical needs. CONCLUSIONS: This research provides a basis for creating health knowledge graphs infused with physician knowledge, marking a significant step forward for PLMs in health care. Notably, the model balances integrating knowledge both comprehensively and selectively, addressing the heterogeneous nature of medical knowledge and the privacy needs of health care institutions.