Evaluating Biomedical Word Embeddings for Vocabulary Alignment at Scale in the UMLS Metathesaurus Using Siamese Networks.

Recent work uses a Siamese Network, initialized with BioWordVec embeddings (distributed word embeddings), for predicting synonymy among biomedical terms to automate a part of the UMLS (Unified Medical Language System) Metathesaurus construction process. We evaluate the use of contextualized word embeddings extracted from nine different biomedical BERT-based models for synonymy prediction in the UMLS by replacing BioWordVec embeddings with embeddings extracted from each biomedical BERT model using different feature extraction methods. Surprisingly, we find that Siamese Networks initialized with BioWordVec embeddings still outperform the Siamese Networks initialized with embedding extracted from biomedical BERT model.

Context-Enriched Learning Models for Aligning Biomedical Vocabularies at Scale in the UMLS Metathesaurus.

The Unified Medical Language System (UMLS) Metathesaurus construction process mainly relies on lexical algorithms and manual expert curation for integrating over 200 biomedical vocabularies. A lexical-based learning model (LexLM) was developed to predict synonymy among Metathesaurus terms and largely outperforms a rule-based approach (RBA) that approximates the current construction process. However, the LexLM has the potential for being improved further because it only uses lexical information from the source vocabularies, while the RBA also takes advantage of contextual information. We investigate the role of multiple types of contextual information available to the UMLS editors, namely source synonymy (SS), source semantic group (SG), and source hierarchical relations (HR), for the UMLS vocabulary alignment (UVA) problem. In this paper, we develop multiple variants of context-enriched learning models (ConLMs) by adding to the LexLM the types of contextual information listed above. We represent these context types in context-enriched knowledge graphs (ConKGs) with four variants ConSS, ConSG, ConHR, and ConAll. We train these ConKG embeddings using seven KG embedding techniques. We create the ConLMs by concatenating the ConKG embedding vectors with the word embedding vectors from the LexLM. We evaluate the performance of the ConLMs using the UVA generalization test datasets with hundreds of millions of pairs. Our extensive experiments show a significant performance improvement from the ConLMs over the LexLM, namely +5.0% in precision (93.75%), +0.69% in recall (93.23%), +2.88% in F1 (93.49%) for the best ConLM. Our experiments also show that the ConAll variant including the three context types takes more time, but does not always perform better than other variants with a single context type. Finally, our experiments show that the pairs of terms with high lexical similarity benefit most from adding contextual information, namely +6.56% in precision (94.97%), +2.13% in recall (93.23%), +4.35% in F1 (94.09%) for the best ConLM. The pairs with lower degrees of lexical similarity also show performance improvement with +0.85% in F1 (96%) for low similarity and +1.31% in F1 (96.34%) for no similarity. These results demonstrate the importance of using contextual information in the UVA problem.

Biomedical Vocabulary Alignment at Scale in the UMLS Metathesaurus.

With 214 source vocabularies, the construction and maintenance process of the UMLS (Unified Medical Language System) Metathesaurus terminology integration system is costly, time-consuming, and error-prone as it primarily relies on (1) lexical and semantic processing for suggesting groupings of synonymous terms, and (2) the expertise of UMLS editors for curating these synonymy predictions. This paper aims to improve the UMLS Metathesaurus construction process by developing a novel supervised learning approach for improving the task of suggesting synonymous pairs that can scale to the size and diversity of the UMLS source vocabularies. We evaluate this deep learning (DL) approach against a rule-based approach (RBA) that approximates the current UMLS Metathesaurus construction process. The key to the generalizability of our approach is the use of various degrees of lexical similarity in negative pairs during the training process. Our initial experiments demonstrate the strong performance across multiple datasets of our DL approach in terms of recall (91-92%), precision (88-99%), and F1 score (89-95%). Our DL approach largely outperforms the RBA method in recall (+23%), precision (+2.4%), and F1 score (+14.1%). This novel approach has great potential for improving the UMLS Metathesaurus construction process by providing better synonymy suggestions to the UMLS editors.

Siamese KG-LSTM: A deep learning model for enriching UMLS Metathesaurus synonymy.

The Unified Medical Language System, or UMLS, is a repository of medical terminology developed by the U.S. National Library of Medicine for improving the computer system's ability of understanding the biomedical and health languages. The UMLS Metathesaurus is one of the three UMLS knowledge sources, containing medical terms and their relationships. Due to the rapid increase in the number of medical terms recently, the current construction of UMLS Metathesaurus, which heavily depends on lexical tools and human editors, is error-prone and time-consuming. This paper takes advantages of the emerging deep learning models for learning to predict the synonyms and non-synonyms between the pairs of biomedical terms in the Metathesaurus. Our learning approach focuses a subset of specific terms instead of the whole Metathesaurus corpus. Particularly, we train the models with biomedical terms from the Disorders semantic group. To strengthen the models, we enrich the inputs with different strategies, including synonyms and hierarchical relationships from source vocabularies. Our deep learning model adopts the Siamese KG-LSTM (Siamese Knowledge Graph - Long Short-Term Memory) in the architecture. The experimental results show that this approach yields excellent performance when handling the task of synonym detection for Disorders semantic group in the Metathesaurus. This shows the potential of applying machine learning techniques in the UMLS Metathesaurus construction process. Although the work in this paper focuses only on specific semantic group of Disorders, we believe that the proposed method can be applied to other semantic groups in the UMLS Metathesaurus.

Determination of Personalized Asthma Triggers From Multimodal Sensing and a Mobile App: Observational Study.

BACKGROUND: Asthma is a chronic pulmonary disease with multiple triggers. It can be managed by strict adherence to an asthma care plan and by avoiding these triggers. Clinicians cannot continuously monitor their patients' environment and their adherence to an asthma care plan, which poses a significant challenge for asthma management. OBJECTIVE: In this study, pediatric patients were continuously monitored using low-cost sensors to collect asthma-relevant information. The objective of this study was to assess whether kHealth kit, which contains low-cost sensors, can identify personalized triggers and provide actionable insights to clinicians for the development of a tailored asthma care plan. METHODS: The kHealth asthma kit was developed to continuously track the symptoms of asthma in pediatric patients and monitor the patients' environment and adherence to their care plan for either 1 or 3 months. The kit consists of an Android app-based questionnaire to collect information on asthma symptoms and medication intake, Fitbit to track sleep and activity, the Peak Flow meter to monitor lung functions, and Foobot to monitor indoor air quality. The data on the patient's outdoor environment were collected using third-party Web services based on the patient's zip code. To date, 107 patients consented to participate in the study and were recruited from the Dayton Children's Hospital, of which 83 patients completed the study as instructed. RESULTS: Patient-generated health data from the 83 patients who completed the study were included in the cohort-level analysis. Of the 19% (16/83) of patients deployed in spring, the symptoms of 63% (10/16) and 19% (3/16) of patients suggested pollen and particulate matter (PM2.5), respectively, to be their major asthma triggers. Of the 17% (14/83) of patients deployed in fall, symptoms of 29% (4/17) and 21% (3/17) of patients suggested pollen and PM2.5, respectively, to be their major triggers. Among the 28% (23/83) of patients deployed in winter, PM2.5 was identified as the major trigger for 83% (19/23) of patients. Similar correlations were not observed between asthma symptoms and factors such as ozone level, temperature, and humidity. Furthermore, 1 patient from each season was chosen to explain, in detail, his or her personalized triggers by observing temporal associations between triggers and asthma symptoms gathered using the kHealth asthma kit. CONCLUSIONS: The continuous monitoring of pediatric asthma patients using the kHealth asthma kit generates insights on the relationship between their asthma symptoms and triggers across different seasons. This can ultimately inform personalized asthma management and intervention plans.

kBot: Knowledge-enabled Personalized Chatbot for Asthma Self-Management.

There is a well-recognized need for a shift to proactive asthma care given the impact asthma has on overall healthcare costs. The demand for continuous monitoring of patient's adherence to the medication care plan, assessment of environmental triggers, and management of asthma can be challenging in traditional clinical settings and taxing on clinical professionals. Recent years have seen a robust growth of general purpose conversational systems. However, they lack the capabilities to support applications such an individual's health, which requires the ability to contextualize, learn interactively, and provide the proper hyper-personalization needed to hold meaningful conversations. In this paper, we present kBot, a knowledge-enabled personalized chatbot system designed for health applications and adapted to help pediatric asthmatic patients (age 8 to 15) to better control their asthma. Its core functionalities include continuous monitoring of the patient's medication adherence and tracking of relevant health signals and environment data. kBot takes the form of an Android application with a frontend chat interface capable of conversing in both text and voice, and a backend cloud-based server application that handles data collection, processing, and dialogue management. It achieves contextualization by piecing together domain knowledge from online sources and inputs from our clinical partners. The personalization aspect is derived from patient answering questionnaires and day-to-day conversations. kBOT's preliminary evaluation focused on chatbot quality, technology acceptance, and system usability involved eight asthma clinicians and eight researchers. For both groups, kBot achieved an overall technology acceptance value of greater than 8 on the 11-point Likert scale and a mean System Usability Score (SUS) greater than 80.

How Will the Internet of Things Enable Augmented Personalized Health?

The Internet of Things refers to network-enabled technologies, including mobile and wearable devices, which are capable of sensing and actuation as well as interaction and communication with other similar devices over the Internet. The IoT is profoundly redefining the way we create, consume, and share information. Ordinary citizens increasingly use these technologies to track their sleep, food intake, activity, vital signs, and other physiological statuses. This activity is complemented by IoT systems that continuously collect and process environment-related data that has a bearing on human health. This synergy has created an opportunity for a new generation of healthcare solutions.