Medical domain knowledge in domain-agnostic generative AI

The text-guided diffusion model GLIDE (Guided Language to Image Diffusion for Generation and Editing) is the state of the art in text-to-image generative artificial intelligence (AI). GLIDE has rich representations, but medical applications of this model have not been systematically explored. If GLIDE had useful medical knowledge, it could be used for medical image analysis tasks, a domain in which AI systems are still highly engineered towards a single use-case. Here we show that the publicly available GLIDE model has reasonably strong representations of key topics in cancer research and oncology, in particular the general style of histopathology images and multiple facets of diseases, pathological processes and laboratory assays. However, GLIDE seems to lack useful representations of the style and content of radiology data. Our findings demonstrate that domain-agnostic generative AI models can learn relevant medical concepts without explicit training. Thus, GLIDE and similar models might be useful for medical image processing tasks in the future - particularly with additional domain-specific fine-tuning.

Anatomical Terms in Jurgis Žilinskas’ Short Textbook of Osteology and Syndesmology

Medical terminology has an extensive and rich history in Latin and Greek languages. When Romans conquered Greece, the knowledge and language of both cultures merged, resulting in new medical concepts regarding disease treatment and containment. Medical records were chronicled by hand, creating medical terms and books. Although medical terms have been drawn from many languages, a large majority originate from Greek and Latin. Terms of Greek origin occur mainly in clinical terminology, while Latin terms make up the majority of anatomical terminology. Another reason for a large number of Greek medical terms is that the Greek language is quite suitable for building compound words. The article discusses the current state of anatomical terminology in Lithuania. The history of the Lithuanian nomenclature of anatomy dates back several centuries, during which the most frequently used Lithuanian anatomical terms were gradually developed. Every time, writing and publishing textbooks, methodological aids, and other anatomy books in the Lithuanian language, the authors interpreted various Latin anatomy terms in their own way and introduced new equivalents in the Lithuanian language. However, they often did not agree on the translation of various Latin terms into Lithuanian and their application to define various structures. The development and perfection of medical terminology is a long process. The most significant contribution in regulating Lithuanian anatomical terminology was made by Jurgis Žilinskas. The terms that we currently use can be found in his textbooks “Osteologija ir syndesmologija” (“Osteology and syndesmology”) (1932) and “Splanchnologija” (“Splanchnology”) (1934) (Litevkiene, Korosteliova 2012, 208). He initiated term regulation in his first textbook, “Lectures of Neurology” (1923), containing only Latin terms, well-formed according to Baseler’s nomenclature. The nomenclature of anatomy compiled by him was applied in other anatomy textbooks and the Dictionary of Medical Terms.

Interpretable segmentation of medical free-text records based on word embeddings

Medical free-text records store a lot of useful information that can be exploited in developing computer-supported medicine. However, extracting the knowledge from the unstructured text is difficult and depends on the language. In the paper, we apply Natural Language Processing methods to process raw medical texts in Polish and propose a new methodology for clustering of patients’ visits. We (1) extract medical terminology from a corpus of free-text clinical records, (2) annotate data with medical concepts, (3) compute vector representations of medical concepts and validate them on the proposed term analogy tasks, (4) compute visit representations as vectors, (5) introduce a new method for clustering of patients’ visits and (6) apply the method to a corpus of 100,000 visits. We use several approaches to visual exploration that facilitate interpretation of segments. With our method, we obtain stable and separated segments of visits which are positively validated against final medical diagnoses. In this paper we show how algorithm for segmentation of medical free-text records may be used to aid medical doctors. In addition to this, we share implementation of described methods with examples as open-source package .

Access to health in an intercultural setting: the role of corpora and images in grasping term variation

Medical concepts can often be lexicalized in several ways depending on aspects such as the facet of the concept being underlined or the particular communicative setting in which the concept is being used. This feature of terminology is known as terminological variation. In this paper we consider terminological variation as a tool to improve interlinguistic and intercultural communication, a key issue in the provision of universal access to health care. To facilitate the identification and analysis of terminological variation, the paper also proposes some search strategies to highlight this phenomenon in corpora, the main source of terminological information. Finally, images are proposed as a key issue in the localization process needed to bridge communication gaps between health care providers and lay audiences. The data used in the paper are taken from an international cooperation project aimed at providing health providers in Yucatan, Mexico, with materials and training in intercultural communication for healthcare mainly in Spanish and Mayan, and from a research project on lexical variation .

Ayurvedic Remedies of Covid-19

The Sanskrit term Ayurveda has translated knowledge of life. It is one of the world's oldest healing systems that originated in eastern culture and it includes numerous medical concepts and it’s a hypothesis for treatment and prevention of disease. In ancient times near to 3000 years ago in India when there are no synthetic medicines was developed then people used Ayurvedic plants to get cures for different diseases. Ayurveda is based on a belief that health and wellness depend on a delicate balance of mind, body, and spirit. Ayurvedic herbs are key components of Ayurveda. COVID-19 is an infectious disease found in December 2019 and it has now become a pandemic. The COVID-19 infection is produced by virulent severe acute respiratory syndrome-Coronavirus-2 (SARS-CoV-2) virus although various antiviral drugs are available for controlling the infection but sometimes, they lack in supply for treating the worldwide population. So, it has become imperative to develop an effective medical strategy for the management of COVID-19 which has become a major threat to humanity. Herbs exhibit various biological activities so; they can effectively help with managing the pandemic. This review discussed some herbs which have the potential for the treatment of COVID-19.

Automatically disambiguating medical acronyms with ontology-aware deep learning

Modern machine learning (ML) technologies have great promise for automating diverse clinical and research workflows; however, training them requires extensive hand-labelled datasets. Disambiguating abbreviations is important for automated clinical note processing; however, broad deployment of ML for this task is restricted by the scarcity and imbalance of labeled training data. In this work we present a method that improves a model’s ability to generalize through novel data augmentation techniques that utilizes information from biomedical ontologies in the form of related medical concepts, as well as global context information within the medical note. We train our model on a public dataset (MIMIC III) and test its performance on automatically generated and hand-labelled datasets from different sources (MIMIC III, CASI, i2b2). Together, these techniques boost the accuracy of abbreviation disambiguation by up to 17% on hand-labeled data, without sacrificing performance on a held-out test set from MIMIC III.

Leveraging of Weighted Ensemble Technique for Identifying Medical Concepts from Clinical Texts at Word and Phrase Level

Concept identification from medical texts becomes important due to digitization. However, it is not always feasible to identify all such medical concepts manually. Thus, in the present attempt, we have applied five machine learning classifiers (Support Vector Machine, K-Nearest Neighbours, Logistic Regression, Random Forest and Naïve Bayes) and one deep learning classifier (Long Short Term Memory) to identify medical concepts by training a total of 27.383K sentences. In addition, we have also developed a rule based phrase identification module to help the existing classifiers for identifying multi- word medical concepts. We have employed word2vec technique for feature extraction and PCA and T- SNE for conducting ablation study over various features to select important ones. Finally, we have adopted two different ensemble approaches, stacking and weighted sum to improve the performance of the individual classifier and significant improvements were observed with respect to each of the classifiers. It has been observed that phrase identification module plays an important role when dealing with individual classifier in identifying higher order ngram medical concepts. Finally, the ensemble approach enhances the results over SVM that was showing initial improvement even after the application of phrase based module.

History of Epidemics in China

The global pandemic of COVID-19 as a zoonotic disease invites new reflections on the human-animal relationship in the history of epidemics. Historians have explored medical concepts, social impacts, and other aspects of epidemics in China at different geographical and temporal scales. Relevant research significantly enriches historical understanding, yet animals seldom occupy the center of attention despite the fact that a variety of human infectious diseases such as plague are zoonotic in origin. This article suggests the need for a reappraisal of epidemics in Chinese history, with particular consideration of historical information on the multifold involvement of animals in human infections and anticontagious measures. Rethinking historically the interactions between humans and animals within the epidemic context helps to raise our awareness that Chinese medical thinkers were sensitive to the possibility of zoonotic infection, and prompt new analyses of how they understood the human-animal boundary and beyond.

Test–Retest Reliability of a Semi-Structured Interview to Aid in Pediatric Traumatic Brain Injury Diagnosis

Objective: Retrospective self-report is typically used for diagnosing previous pediatric traumatic brain injury (TBI). A new semi-structured interview instrument (New Mexico Assessment of Pediatric TBI; NewMAP TBI) investigated test–retest reliability for TBI characteristics in both the TBI that qualified for study inclusion and for lifetime history of TBI. Method: One-hundred and eight-four mTBI (aged 8–18), 156 matched healthy controls (HC), and their parents completed the NewMAP TBI within 11 days (subacute; SA) and 4 months (early chronic; EC) of injury, with a subset returning at 1 year (late chronic; LC). Results: The test–retest reliability of common TBI characteristics [loss of consciousness (LOC), post-traumatic amnesia (PTA), retrograde amnesia, confusion/disorientation] and post-concussion symptoms (PCS) were examined across study visits. Aside from PTA, binary reporting (present/absent) for all TBI characteristics exhibited acceptable (≥0.60) test–retest reliability for both Qualifying and Remote TBIs across all three visits. In contrast, reliability for continuous data (exact duration) was generally unacceptable, with LOC and PCS meeting acceptable criteria at only half of the assessments. Transforming continuous self-report ratings into discrete categories based on injury severity resulted in acceptable reliability. Reliability was not strongly affected by the parent completing the NewMAP TBI. Conclusions: Categorical reporting of TBI characteristics in children and adolescents can aid clinicians in retrospectively obtaining reliable estimates of TBI severity up to a year post-injury. However, test–retest reliability is strongly impacted by the initial data distribution, selected statistical methods, and potentially by patient difficulty in distinguishing among conceptually similar medical concepts (i.e., PTA vs. confusion).

Unifying Heterogenous Electronic Health Records Systems via Text-Based Code Embedding: Study of Predictive Modeling (Preprint)

BACKGROUND Substantial increase in the use of Electronic Health Records (EHRs) has opened new frontiers for predictive healthcare. However, while EHR systems are nearly ubiquitous, they lack a unified code system for representing medical concepts. Heterogeneous formats of EHR present a substantial barrier for the training and deployment of state-of-the-art deep learning models at scale. OBJECTIVE The aim of this study is to suggest a novel text embedding approach to overcome heterogeneity of EHR structure among different EHR systems. METHODS We introduce Description-based Embedding, DescEmb, a code-agnostic description-based representation learning framework for predictive modeling on EHR. DescEmb takes advantage of the flexibility of neural language understanding models while maintaining a neutral approach that can be combined with prior frameworks for task-specific representation learning or predictive modeling. RESULTS Based on five prediction tasks with two heterogeneous EHR datasets, DescEmb achieves comparable or superior performance to the traditional code-based embedding approach, especially under the zero-shot and few-shot transfer learning scenarios. We also demonstrate that DescEmb enables us to train a single model on a pooled dataset from heterogeneous EHR systems and achieve the same, if not better performance compared to training separate models for each EHR system. CONCLUSIONS Based on the promising results, we believe the description-based embedding approach on EHR will open a new direction for large-scale predictive modeling in healthcare.