Search

Abstract The ongoing coronavirus disease 2019 (COVID-19) pandemic presents serious public health threats. Omicron, the current most prevalent strain of COVID-19, has a low fatality rate and very high transmissibility, so the number of patients with mild symptoms of COVID-19 is rapidly increasing. This change of pandemic challenges medical systems worldwide in many aspects, including sharp increases in demands for hospital infrastructure, critical shortages in medical equipment, and medical staff. Predicting deterioration in mild patients could alleviate these problems. A novel scoring system was proposed for predicting the deterioration of patients whose condition may worsen rapidly and those who all still mild or asymptomatic. Retrospective cohorts of 954 and 2,035 patients that quarantined in the Residential Treatment Center were assembled for derivation and external validation of mild COVID-19, respectively. Deterioration was defined as transfer to a local hospital due to worsening condition of the patients during the 2-week isolation period. A total of 15 variables: sex, age, seven pre-existing conditions (diabetes, hypertension, cardiovascular disease, respiratory disease, liver disease, kidney disease, and organ transplant), and five vital signs (systolic blood pressure (SBP), diastolic blood pressure (DBP), heart rate (HR), body temperature, and oxygen saturation (SpO2)) were collected. A scoring system was developed using seven variables (age, pulse rate, SpO2, SBP, DBP, temperature, and hypertension) with significant differences between the transfer and not transfer groups in logistic regression. The proposed system was compared with existing scoring systems that assess the severity of patient conditions. The performance of the proposed scoring system to predict deterioration in patients with mild COVID-19 showed an area under the receiver operating characteristic (AUC) of 0.868. This is a statistically significant improvement compared to the performance of the previous patient condition assessment scoring systems. During external validation, the proposed system showed the best and most robust predictive performance (AUC = 0.768; accuracy = 0.899). In conclusion, we proposed a novel scoring system for predicting patients with mild COVID-19 who will experience deterioration which could predict the deterioration of the patient’s condition early with high predictive performance. Furthermore, because the scoring system does not require special calculations, it can be easily measured to predict the deterioration of a patients’ condition. This system can be used as effective tool for early detection of deterioration in mild COVID-19 patients.

BackgroundDuring the height of the COVID-19 pandemic, the Korean government temporarily allowed full scale telehealth care for safety and usability. However, limited studies have evaluated the impact of telehealth by analyzing the physical and/or mental health data of patients with COVID-19 diagnosis collected through telehealth targeting Korean population.ObjectiveThis study aimed to identify subgroup of depressive symptom trajectories in patients with clinically mild COVID-19 using collected longitudinal data from a telehealth-based contactless clinical trial.MethodsA total of 199 patients with COVID-19 were accrued for contactless clinical trial using telehealth from March 23 to July 20, 2022. Depressive symptoms were measured using the patient health questionnaire-9 on the start day of quarantine, on the final day of quarantine, and 1 month after release from quarantine. Additionally, acute COVID-19 symptoms were assessed every day during quarantine. This study used a latent class mixed model to differentiate subgroups of depressive symptom trajectories and a logistic regression model with Firth’s correction to identify associations between acute COVID-19 symptoms and the subgroups.ResultsTwo latent classes were identified: class 1 with declining linearity at a slow rate and class 2 with increasing linearity. Among COVID-19 symptoms, fever, chest pain, and brain fog 1 month after release from quarantine showed strong associations with class 2 (fever: OR, 19.43, 95% CI, 2.30–165.42; chest pain: OR, 6.55, 95% CI, 1.15–34.61; brain fog: OR, 7.03, 95% CI 2.57–20.95). Sleeping difficulty and gastrointestinal symptoms were also associated with class 2 (gastrointestinal symptoms: OR, 4.76, 95% CI, 1.71–14.21; sleeping difficulty: OR, 3.12, 95% CI, 1.71–14.21).ConclusionThese findings emphasize the need for the early detection of depressive symptoms in patients in the acute phase of COVID-19 using telemedicine. Active intervention, including digital therapeutics, may help patients with aggravated depressive symptoms.

As a key modifiable risk factor, alcohol consumption is clinically crucial information that allows medical professionals to further understand their patients’ medical conditions and suggest appropriate lifestyle modifying interventions. However, identifying alcohol-related information from unstructured free-text clinical notes is often challenging. Not only are the formats of the notes inconsistent, but they also include a massive amount of non-alcohol-related information. Furthermore, for medical institutions outside of English-speaking countries, these clinical notes contain both a mixture of English and local languages, inducing additional difficulty in the extraction. Thanks to the increasing availability of electronic medical record (EMR), several previous works explored the idea of using natural language processing (NLP) to train machine learning models that automatically identify alcohol-related information from unstructured clinical notes. However, all these previous works are limited to English clinical notes, thereby able to leverage various large-scale external ontologies during the text preprocessing. Furthermore, they rely on simple NLP techniques such as the bag-of-words models that suffer from high dimensionality and out-of-vocabulary issues. Addressing these issues, we adopt fine-tuning multilingual transformers. By leveraging their linguistically rich contextual information learned during their pre-training, we are able to extract alcohol-related information from unstructured clinical notes without preprocessing the clinical notes on any external ontologies. Furthermore, our work is the first to explore the use of transformers in bilingual clinical notes to extract alcohol-related information. Even with minimal text preprocessing, we achieve extraction accuracy of 84.70% in terms of macro F-1 score.

Coronavirus has caused many casualties and is still spreading. Some people experience rapid deterioration that is mild at first. The aim of this study is to develop a deterioration prediction model for mild COVID-19 patients during the isolation period. We collected vital signs from wearable devices and clinical questionnaires. The derivation cohort consisted of people diagnosed with COVID-19 between September and December 2021, and the external validation cohort collected between March and June 2022. To develop the model, a total of 50 participants wore the device for an average of 77 h. To evaluate the model, a total of 181 infected participants wore the device for an average of 65 h. We designed machine learning-based models that predict deterioration in patients with mild COVID-19. The prediction model, 10 min in advance, showed an area under the receiver characteristic curve (AUC) of 0.99, and the prediction model, 8 h in advance, showed an AUC of 0.84. We found that certain variables that are important to model vary depending on the point in time to predict. Efficient deterioration monitoring in many patients is possible by utilizing data collected from wearable sensors and symptom self-reports.

Health information exchange can improve health outcomes and reduce unnecessary medical expenses. An important task in health information exchange is to prove data integrity and strengthen the right to self-determination of individuals. This can be addressed using blockchain technology and dynamic consents. We aimed to develop a blockchain-based mobile platform called HealthPocket to exchange reliable health information with proven integrity through a dynamic consent system based on the HL7 FHIR standards. Through HealthPocket, subjects can selectively provide their consent to share specific medical and PGHD through proper authorization, and each response is converted into the JSON format with FHIR compatibility. We conducted a usability test to demonstrate health information exchange between primary and tertiary medical institutions. A total of 116 subjects used the HealthPocket mobile application to selectively share health information for at least one month, and conducted a questionnaire about their experience. In addition, medical staff of the institution could access the medical information shared by the participants and use it for treatment. The user surveys for patients examined the perceived usefulness, perceived ease of use, and overall service satisfaction. The mean overall satisfaction with the health information exchange service was the highest (4.67 out of 5 points). As security, personal information protection, and interoperability are main concerns related to health information exchange, blockchain technology can provide suitable solutions. For instance, when health information is exchanged using blockchain technology, it is impossible to alter data blocks. Therefore, by using a blockchain-based platform and dynamic consent system, we ensured the integrity of medical data.

BackgroundTo date, the association between acute signs and symptoms of COVID-19 and the exacerbation of depression and anxiety in patients with clinically mild COVID-19 has not been evaluated. ObjectiveThis study was designed to assess the correlation between acute signs and symptoms of COVID-19 and the exacerbation of depression and anxiety in patients with clinically mild COVID-19 at a residential treatment center in South Korea. MethodsThis retrospective study assessed 2671 patients with COVID-19 admitted to 4 residential treatment centers operated by Seoul National University Hospital, South Korea, from March 2020 to April 2022. Depression and anxiety were assessed using the 2-item Patient Health Questionnaire (PHQ-2) and 2-item Generalized Anxiety Disorder (GAD-2) scale, respectively. The exacerbation of depression and anxiety symptoms was identified from the differences in PHQ-2 and GAD-2 scores between admission and discharge, respectively. The patients’ clinical characteristics, including acute signs and symptoms of COVID-19, GAD-2 and PHQ-2 scores, were obtained from electronic health records. Demographic characteristics, a summary of vital signs, and COVID-19 symptoms were analyzed and compared between the patient groups with and those without exacerbated PHQ-2 and GAD-2 scores using the chi-square test. We applied logistic regression to identify the association between acute signs and symptoms of COVID-19 and the exacerbation of depression and anxiety. ResultsSleep disorders were associated with exacerbated depression (odds ratio [OR] 1.09, 95% CI 1.05-1.13) and anxiety (OR 1.1, 95% CI 1.06-1.14), and the sore throat symptom was associated with exacerbated anxiety symptoms (OR 1.03, 95% CI 1.00-1.07). Patients with abnormal oxygen saturation during quarantine were more likely to have exacerbated depression (OR 1.27, 95% CI 1.00-1.62), and those with an abnormal body temperature during quarantine were more likely to experience anxiety (OR 1.08, 95% CI 1.01-1.16). As anticipated, patients who experienced psychological symptoms at admission were more likely to experience depression (OR 1.91, 95% CI 1.52-2.41) and anxiety (OR 1.98, 95% CI 1.54-2.53). Meanwhile, the PHQ-2 and GAD-2 scores measured at admission revealed that lower the score, higher the possibility of exacerbation of both depression (OR 0.15, 95% CI 0.11-0.22) and anxiety (OR 0.13, 95% CI 0.10-0.19). ConclusionsResults from this study suggest the importance of further interventions for patients with abnormal oxygen saturation, abnormal body temperatures, sore throat, and sleep disorder symptoms or initial psychological symptoms to mitigate the exacerbation of depression and anxiety. In addition, this study highlights the usability of short and efficient scales such as the PHQ-2 and GAD-2 in the assessment of the mental health of patients with clinically mild COVID-19 symptoms who were quarantined at home during the pandemic era.

IntroductionTo effectively manage patients with coronavirus disease 2019 (COVID-19) while minimizing contact between medical staff, clinical trial protocol that facilitates contactless patient management was designed to predict deterioration of disease condition and monitor mental health status.MethodsThrough consultation with infectious disease specialists and psychiatrists, this study identified main clinical indicators related to respiratory and non-respiratory outcomes, and mental health. Telehealth devices that could collect relevant data indicators were explored. The following three modes were identified: wearable devices, video calls, and online questionnaires. Clinical trial protocol was implemented to patients confirmed with COVID-19 infection and admitted to Seongnam residential treatment centers between September 1, 2021 and December 30, 2021. Data were collected from wearable devices, video calls, online questionnaires, and from electronic health records. Participant satisfaction was assessed through an online survey at the time of discharge.ResultsIn total, 120 asymptomatic and mildly symptomatic COVID-19 patients participated in this trial. Seven types of physiological and life log data were collected from 87 patients using wearable devices, video and audio recordings, and online mental health-related questionnaire. Most participants were satisfied with the overall trial process, but perceived difficulties in using telehealth devices.ConclusionThis trial collected simultaneously generated multimodal patient data using various telehealth devices in a contactless setting for COVID-19 patients. Data collected in this study will be used to build a remote patient management system based on the prediction algorithms.

BackgroundDepression and suicide are critical social problems worldwide, but tools to objectively diagnose them are lacking. Therefore, this study aimed to diagnose depression through machine learning and determine whether it is possible to identify groups at high risk of suicide through words spoken by the participants in a semi-structured interview.MethodsA total of 83 healthy and 83 depressed patients were recruited. All participants were recorded during the Mini-International Neuropsychiatric Interview. Through the suicide risk assessment from the interview items, participants with depression were classified into high-suicide-risk (31 participants) and low-suicide-risk (52 participants) groups. The recording was transcribed into text after only the words uttered by the participant were extracted. In addition, all participants were evaluated for depression, anxiety, suicidal ideation, and impulsivity. The chi-square test and student’s T-test were used to compare clinical variables, and the Naive Bayes classifier was used for the machine learning text model.ResultsA total of 21,376 words were extracted from all participants and the model for diagnosing patients with depression based on this text confirmed an area under the curve (AUC) of 0.905, a sensitivity of 0.699, and a specificity of 0.964. In the model that distinguished the two groups using statistically significant demographic variables, the AUC was only 0.761. The DeLong test result (p-value 0.001) confirmed that the text-based classification was superior to the demographic model. When predicting the high-suicide-risk group, the demographics-based AUC was 0.499, while the text-based one was 0.632. However, the AUC of the ensemble model incorporating demographic variables was 0.800.ConclusionThe possibility of diagnosing depression using interview text was confirmed; regarding suicide risk, the diagnosis accuracy increased when demographic variables were incorporated. Therefore, participants’ words during an interview show significant potential as an objective and diagnostic marker through machine learning.

BackgroundDigital health care is an important strategy in the war against COVID-19. South Korea introduced living and treatment support centers (LTSCs) to control regional outbreaks and care for patients with asymptomatic or mild COVID-19. Seoul National University Hospital (SNUH) introduced information and communications technology (ICT)–based solutions to manage clinically healthy patients with COVID-19. ObjectiveThis study aims to investigate satisfaction and usability by patients and health professionals in the optimal use of a mobile app and wearable device that SNUH introduced to the LTSC for clinically healthy patients with COVID-19. MethodsOnline surveys and focus group interviews were conducted to collect quantitative and qualitative data. ResultsRegarding usability testing of the wearable device, perceived usefulness had the highest mean score of 4.45 (SD 0.57) points out of 5. Regarding usability of the mobile app, perceived usefulness had the highest mean score of 4.62 (SD 0.48) points out of 5. Regarding satisfaction items for the mobile app among medical professionals, the “self-reporting” item had the highest mean score of 4.42 (SD 0.58) points out of 5. In focus group interviews of health care professionals, hospital information system interfacing was the most important functional requirement for ICT-based COVID-19 telemedicine. ConclusionsImprovement of patient safety and reduction of the burden on medical staff were the expected positive outcomes. Stability and reliability of the device, patient education, accountability, and reimbursement issues should be considered as part of the development of remote patient monitoring. In responding to a novel contagious disease, telemedicine and a wearable device were shown to be useful during a global crisis.

Background. The prevalence of nonalcoholic fatty liver disease (NAFLD) has been increasing in the general population. This study evaluated the association between NAFLD and significant coronary stenosis in asymptomatic adults and evaluated sex-based differences. Methods. We performed a retrospective cross-sectional study in participants without previous cardiovascular diseases who visited the Seoul National University Hospital Health Promotion Center for a health checkup between January 1, 2010, and December 31, 2015. NAFLD was diagnosed on sonography, while coronary artery stenosis (CAS) was assessed on coronary computed tomography angiography (CCTA). Results. We obtained 3,693 participants who met the inclusion criteria, and 3,449 of them had no significant stenosis. Among the participants with significant stenosis, the prevalence of NAFLD was 59.4% (145 patients). The prevalence of NAFLD was 47.26% in male participants, which was higher than that in female participants. The association between NAFLD and significant CAS persisted after adjusting for age, body mass index, glycated hemoglobin, and Framingham risk factors. The correlation between NAFLD and significant coronary stenosis appeared to be stronger in women than in men, but the absolute risk was higher in men than in women. Conclusion. NAFLD was strongly associated with CAS. We should be alert about an increased cardiovascular risk in patients with NAFLD and more intensively provide primary prevention by performing tests to detect subclinical atherosclerosis.

Smoking is an important variable for clinical research, but there are few studies regarding automatic obtainment of smoking classification from unstructured bilingual electronic health records (EHR). We aim to develop an algorithm to classify smoking status based on unstructured EHRs using natural language processing (NLP). With acronym replacement and Python package Soynlp, we normalize 4711 bilingual clinical notes. Each EHR notes was classified into 4 categories: current smokers, past smokers, never smokers, and unknown. Subsequently, SPPMI (Shifted Positive Point Mutual Information) is used to vectorize words in the notes. By calculating cosine similarity between these word vectors, keywords denoting the same smoking status are identified. Compared to other keyword extraction methods (word co-occurrence-, PMI-, and NPMI-based methods), our proposed approach improves keyword extraction precision by as much as 20.0%. These extracted keywords are used in classifying 4 smoking statuses from our bilingual EHRs. Given an identical SVM classifier, the F1 score is improved by as much as 1.8% compared to those of the unigram and bigram Bag of Words. Our study shows the potential of SPPMI in classifying smoking status from bilingual, unstructured EHRs. Our current findings show how smoking information can be easily acquired for clinical practice and research.

Background The combined effect of transitions of metabolic health and weight on cardiovascular disease (CVD) remains unclear. We aimed to examine the association of concurrent changes of metabolic health and weight on CVD over time. Methods and Results The study population consisted of 205 394 from the Korean National Health Insurance Service. Metabolic health was determined by fasting serum glucose, total cholesterol, and blood pressure levels, while obesity was determined by body mass index. All participants were divided into either metabolically healthy nonobese (MHNO), metabolically healthy obese, metabolically unhealthy nonobese, or metabolically unhealthy obese for each of the first (2002–2003) and second (2004–2005) health screening periods, after which participants were followed‐up for CVD from 2006 to 2015. Cox proportional hazards regression was used to determine adjusted hazard ratios (aHRs) and 95% CIs. Among initial MHNO participants, those who became metabolically healthy obese (aHR, 1.25; 95% CI, 1.10–1.41), metabolically unhealthy nonobese (aHR, 1.23; 95% CI, 1.15–1.31), and metabolically unhealthy obese (aHR, 1.34; 95% CI, 1.12–1.61) had elevated risk for CVD compared with those who remained MHNO. Conversely, improving metabolic health and obesity were associated with reduced CVD risk among initially metabolically unhealthy nonobese to secondary MHNO (aHR, 0.79; 95% CI, 0.73–0.84), metabolically unhealthy obese to MHNO (aHR, 0.68; 95% CI, 0.58–0.81), and metabolically unhealthy obese to metabolically healthy obese (aHR, 0.73; 95% CI, 0.66–0.80) participants. Conclusions Changes toward metabolically unhealthy or obese states resulted in increased CVD risk. Improving metabolic health along with reducing weight may lead to decreased risk of CVD.

With growing interest in machine learning, text standardization is becoming an increasingly important aspect of data pre-processing within biomedical communities. As performances of machine learning algorithms are affected by both the amount and the quality of their training data, effective data standardization is needed to guarantee consistent data integrity. Furthermore, biomedical organizations, depending on their geographical locations or affiliations, rely on different sets of text standardization in practice. To facilitate easier machine learning-related collaborations between these organizations, an effective yet practical text data standardization method is needed. In this paper, we introduce MARIE (a context-aware term mapping method with string matching and embedding vectors), an unsupervised learning-based tool, to find standardized clinical terminologies for queries, such as a hospital’s own codes. By incorporating both string matching methods and term embedding vectors generated by BioBERT (bidirectional encoder representations from transformers for biomedical text mining), it utilizes both structural and contextual information to calculate similarity measures between source and target terms. Compared to previous term mapping methods, MARIE shows improved mapping accuracy. Furthermore, it can be easily expanded to incorporate any string matching or term embedding methods. Without requiring any additional model training, it is not only effective, but also a practical term mapping method for text data standardization and pre-processing.

Osteoporosis is a medical condition of global concern, with increasing incidence in both sexes. Bone mineral density (BMD), a highly heritable trait, has been proven a useful diagnostic factor in predicting fracture. Because medical information is lacking about male osteoporotic genetics, we conducted a genome-wide association study of BMD in Korean men. With 1,176 participants, we analyzed 4,414,664 single nucleotide polymorphisms (SNPs) after genomic imputation, and identified five SNPs and three loci correlated with bone density and strength. Multivariate linear regression models were applied to adjust for age and body mass index interference. Rs17124500 (p = 6.42 × 10-7), rs34594869 (p = 6.53 × 10-7) and rs17124504 (p = 6.53 × 10-7) in 14q31.3 and rs140155614 (p = 8.64 × 10-7) in 15q25.1 were significantly associated with lumbar spine BMD (LS-BMD), while rs111822233 (p = 6.35 × 10-7) was linked with the femur total BMD (FT-BMD). Additionally, we analyzed the relationship between BMD and five genes previously identified in Korean men. Rs61382873 (p = 0.0009) in LRP5, rs9567003 (p = 0.0033) in TNFSF11 and rs9935828 (p = 0.0248) in FOXL1 were observed for LS-BMD. Furthermore, rs33997547 (p = 0.0057) in ZBTB and rs1664496 (p = 0.0012) in MEF2C were found to influence FT-BMD and rs61769193 (p = 0.0114) in ZBTB to influence femur neck BMD. We identified five SNPs and three genomic regions, associated with BMD. The significance of our results lies in the discovery of new loci, while also affirming a previously significant locus, as potential osteoporotic factors in the Korean male population.

How we perceive property of surfaces with distinct geometry and reflectance under various illumination conditions is not fully understood. One widely studied approach to understanding perceptual surface property is to derive statistics from images of surfaces with the goal of constructing models that can estimate surface property attributes. This paper presents machine-learning-based methods to estimate the lightness and glossiness of surfaces. Instead of deriving image statistics and building estimation models on top of them, we use deep networks to estimate the perceptual surface property directly from surface images. We adopt the attention models in our networks to allow the networks to estimate the surface property based on features in certain parts of images. This approach can rule out image variations due to geometry, reflectance, and illumination when making the estimations. The networks are trained with perceptual lightness and glossiness data obtained from psychophysical experiments. The trained deep networks provide accurate estimations of the surface property that correlate well with human perception. The network performances are compared with various image statistics derived for the estimation of perceptual surface property.

The concept of MyData emerged as a paradigm shift in personal data management and the process of seeking to transform the current organization-centered system. MyData enables the utilization of one’s own personal information that is scattered among various institutions as a system for data subjects to exercise rights of self-determination. We aimed to develop and demonstrate a MyData platform (MyHealthData) that allows data subjects to download and manage health-related personal data stored in various medical institutions. The platform consists of a mobile app for users, API (application–program interface) for data conversion and exchange installed in the hospital information system (HIS), and a relay server connected to the blockchain to ensure data integrity. User surveys were conducted to explore perceived usefulness, perceived ease of use, and satisfaction. We provided four services to users through the platform developed in this study: inquiring about medical and health checkup records, health coaching, checking conditions of participation in clinical research, and claims, all by using an app. A total of 1228 participants signed up for the service and the overall user satisfaction was high, especially with ‘inquire about medical and health checkup records’. MyData brings a user-centered paradigm in which data subjects can directly participate in the use of their own data. MyData will improve healthcare data interoperability, allowing it to be used not only in research areas but also in other areas by sharing and integrating various healthcare data.