Validation of a specialized evaluation system for COVID-19 in Japan: A retrospective, multicenter cohort study.

INTRODUCTION: Evaluation of a severity grade (SG) is important to classify patients for efficient use of limited medical resources. This study validates two existing evaluation systems for the prevention of the coronavirus disease 2019 (COVID-19) in Japan: a criterion of SG and a list of 14 specialized underlying diseases (SUDs). METHODS: A retrospective cohort was created using electronic medical records from 18 research institutes. The cohort includes 6,050 COVID-19 patients with two types of diagnosis information as follows: SG at hospitalization among mild, moderate I, moderate II, and severe and aggravation after hospitalization. RESULTS: A crude mortality rate and an aggravation rate increased by the worsening of SG in the COVID-19 cohort. The transition of the aggravation rate was notable for COVID-19 patients with SUD. A conditional probability of the mortality given the aggravation in the COVID-19 cohort was 87.4% compared to mild or moderate patients (approximately 21%-45%) who have the possibility of the aggravation. An odds ratio of the mortality and aggravation information about the SUD list was higher than other variables. CONCLUSIONS: We demonstrated the possibility of improving the criteria of SG by including the SUD list for more effective operation of the criteria of SG. Furthermore, we demonstrated the importance of the prevention of the aggravation based on the conditional probability, and the possibility of predicting the aggravation using the risk factors.

Developing Artificial Intelligence Models for Extracting Oncologic Outcomes from Japanese Electronic Health Records.

INTRODUCTION: A framework that extracts oncological outcomes from large-scale databases using artificial intelligence (AI) is not well established. Thus, we aimed to develop AI models to extract outcomes in patients with lung cancer using unstructured text data from electronic health records of multiple hospitals. METHODS: We constructed AI models (Bidirectional Encoder Representations from Transformers [BERT], Naïve Bayes, and Longformer) for tumor evaluation using the University of Miyazaki Hospital (UMH) database. This data included both structured and unstructured data from progress notes, radiology reports, and discharge summaries. The BERT model was applied to the Life Data Initiative (LDI) data set of six hospitals. Study outcomes included the performance of AI models and time to progression of disease (TTP) for each line of treatment based on the treatment response extracted by AI models. RESULTS: For the UMH data set, the BERT model exhibited higher precision accuracy compared to the Naïve Bayes or the Longformer models, respectively (precision [0.42 vs. 0.47 or 0.22], recall [0.63 vs. 0.46 or 0.33] and F1 scores [0.50 vs. 0.46 or 0.27]). When this BERT model was applied to LDI data, prediction accuracy remained quite similar. The Kaplan-Meier plots of TTP (months) showed similar trends for the first (median 14.9 [95% confidence interval 11.5, 21.1] and 16.8 [12.6, 21.8]), the second (7.8 [6.7, 10.7] and 7.8 [6.7, 10.7]), and the later lines of treatment for the predicted data by the BERT model and the manually curated data. CONCLUSION: We developed AI models to extract treatment responses in patients with lung cancer using a large EHR database; however, the model requires further improvement.

The use of artificial intelligence (AI) to derive health outcomes from large electronic health records is not well established. Thus, we built three different AI models: Bidirectional Encoder Representations from Transformers (BERT), Naïve Bayes, and Longformer to serve this purpose. Initially, we developed these models based on data from the University of Miyazaki Hospital (UMH) and later improved them using the Life Data Initiative (LDI) data set of six hospitals. The performance of the BERT model was better than the other two, and it showed similar results when it was applied to the LDI data set. The Kaplan­Meier plots of time to progression of disease for the predicted data by the BERT model showed similar trends to those for the manually curated data. In summary, we developed an AI model to extract health outcomes using a large electronic health database in this study; however, the performance of the AI model could be improved using more training data.

Development of a quantitative methylation-specific droplet digital PCR assay for detecting Dickkopf-related protein 3.

OBJECTIVE: The detection and monitoring of DNA methylation status in circulating tumor cell DNA (ctDNA) provides critical insights into cancer diagnosis and progression. The methylation status of the Dickkopf-related protein 3 (DKK3) promoter region is correlated with the metastasis and recurrence of multiple cancers. Thus, detecting the methylation status via non-invasive methods is essential for the diagnosis and prognosis of cancers. Using a droplet digital polymerase chain reaction approach, we have developed a highly sensitive and quantitative measurement of methylated and unmethylated DKK3 derived from circulating cell-free DNA (ccfDNA). RESULTS: We confirmed the specificity of droplet digital methylation specific polymerase chain reaction (ddMSP). We selected the optimal bisulfite conversion method using commercially available kits. We validated the ddMSP analysis system by analyzing the methylation status of genomic DNA extracted from cultured mesothelioma cells and mesothelial cells. Our system quantified approximately 30 copies of cell-free DNA per 4 mL, which is sufficient for detecting ctDNA. Finally, we quantified methylated and unmethylated DKK3 copies in ccfDNA from 21 patients with malignant mesothelioma.

Behavioral signs of an unintended error in nursing information sharing with electronic clinical pathways: a mixed research approach.

Electronic clinical pathways (ECPs) strongly encourage the standardization of medical treatment and the sharing of information among medical staff. The goal of this study was to determine the influence of ECPs on information sharing among nurses in a university hospital. Four experienced nurses, selected based on ECP composing and operation experience, were recruited from the department with the most frequent users in the first-round interview, 132 nurses' questionnaire answers were analyzed, and eight nurses participated in the second-round interview. This study conducted a mixed-method (interview-questionnaire-interview) investigation to extract the behavioral signs of unintended errors in information sharing after the ethical approval was obtained. On the basis of ANOVA and t-test for the questionnaire and constant comparison for interview, this study found that the greater extent of user dependency on convenient ECPs in the frequent-use group led to mistakes under hectic conditions. This study also found evidence of poor management of ECPs when problems occurred. The immature design of ECPs provoked inappropriate behaviors among nurses even though they brought about some benefits such as mitigation of the burden of daily recording tasks. The findings empirically showed the ECP user's behavioral changes regarding the technology-induced error.

Financial Estimation of the Uncertainty in Medicine Using Present Value of Medical Fees and a Mortality Risk Prediction Model: a Retrospective Cohort Study.

This study aimed to develop a method to enable the financial estimation of each patient's uncertainty without focusing on healthcare technology. We define financial uncertainty (FU) as the difference between an actual amount of claim (AC) and the discounted present value of the AC (DAC). DAC can be calculated based on a discounted present value calculated using a cash flow, a period of investment, and a discount rate. The present study considered these three items as AC, the length of hospital stay, and the predicted mortality rate. The mortality prediction model was built using typical data items in standard level electronic medical records such as sex, age, and disease information. The performance of the prediction model was moderate because an area under curve was approximately 85%. The empirical analysis primarily compares the FU of the top 20 diseases with the actual AC using a retrospective cohort in the University of Miyazaki Hospital. The observational period is 5 years, from April 1, 2013, to March 31, 2018. The analysis demonstrates that the proportion of FU to actual AC is higher than 20% in low-weight children, patients with leukemia, brain tumor, myeloid leukemia, or non-Hodgkin's lymphoma. For these diseases, patients cannot avoid long hospitalization; therefore, the medical fee payment system should be designed based on uncertainty. Our method is both practical and generalizable because it uses a small number of data items that are required in standard electronic medical records. This method contributes to the decision-making processes of health policymakers.

REIC/Dkk-3 Gene Therapy Induces Immunogenic Cell Death in a Mouse Model of Malignant Mesothelioma.

BACKGROUND/AIM: The adenovirus vector- carrying reduced expression in immortalized cell (REIC) gene (Ad-REIC) increases endoplasmic reticulum stress chaperone GRP78/BiP expression and induces the JNK-mediated apoptotic pathway. We aimed to determine whether Ad-REIC-induced apoptotic cell death can trigger immunogenic cell death (ICD). MATERIALS AND METHODS: We examined the emission of damage-associated molecular patterns in vitro and the vaccination effect in vivo. We determined the immunological changes in the tumour microenvironment by putative ICD inducers and the combined effects of immune checkpoint blockade therapies. RESULTS: Ad-REIC induced the release of high-mobility group box 1 and adenosine triphosphate and the translocation of calreticulin in murine mesothelioma AB12 cells. The vaccination effect was elicited by Ad-REIC treatment in vivo. The effect of Ad-REIC was potentiated by anti-cytotoxic T-lymphocyte-associated protein 4 antibody treatment in a murine mesothelioma AB1-HA cell model. CONCLUSION: Ad-REIC induces ICD in malignant mesothelioma.

Development of a General Statistical Analytical System Using Nationally Standardized Medical Information.

In Japan, since the Next Generation Medical Infrastructure Act regarding anonymized medical data contributing to R&D came into force in 2018, it is expected to exploit medical data for R&D. The Millennial Medical Record Project has been collected a large amount of standardized medical data of a number of hospitals stored in a database under the act. In order for users to widely exploit the medical data when carrying out trial-and-error, there is a difficulty of data access because of a highly secured management of non-anonymous medical data. To solve the data access problem, we develop a general statistical analytical system for executing a variety of statistical significance tests with statistical power analysis in an environment of trial-and-error for users' analyses without programming. In the analytical system, the front-end is a registration form as the input and the analysis results as the output on Microsoft Excel, and the back-end is based on Python, R and SQL. Although the fixed registration form covers limited application for the analysis, since the analysis results using the stored Millennial Medical Record data is provided in a short time without collecting the necessary data for the analysis, the exploitation of medical data could widely and rapidly promote by medical experts/researchers in the manner of trial-and-error. The developed system could apply to make protocols for clinical research and clinical trial, and the potential to discover real-world evidence could be increased.

Event Surrogate from Clinical Pathway Completion to Daily Meal for Availability Extension Using Standard Electronic Medical Records: a Retrospective Cohort Study.

This study aimed to improve generalizability of our previous study that analyzed clinical pathway (CP) completion. Although our previous study demonstrated that CP completion can reduce the length of hospital stay, it is possible for few medical organizations to extract the implementation of treatment registered on CP from typical electronic medical records. Therefore, we have defined a prospective event for event substitution, called meal completion (MC), in which patients can take their meal daily. Data were collected from April 2013 to March 2018 from the electronic medical records of the University of Miyazaki Hospital. We used propensity score matching to extract records from 8033 patients. Patients were further divided into the MC and non-MC groups; 2577 patients in each group were available for data analysis. The numbers of patients with CP completion were 646 (28.1%) in the MC group and 411 (18.2%) in the non-MC group. The P value of the chi-square test was <0.001. According to this result, there was the causation from MC to increase in CP completion. Additionally, it was possible to consider the inclusion relationship in all treatments (universal set), treatments registered on CP (subset of all treatments), and meals (subset of treatments registered on CP). In conclusion, MC can substitute for CP completion because the demonstration is appropriate for the Prentice criterion, which is often used for the evaluation of a surrogate endpoint.

Bursa atrophy at 28 days old caused by variant infectious bursal disease virus has a negative economic impact on broiler farms in Japan.

Infectious bursal disease (IBD), caused by IBD virus (IBDV), is highly contagious, immunosuppressive and causes a negative economic impact on poultry industry. IBDV-vaccinated broiler farms at south Kyushu, Japan had a bursa-to-bodyweight ratio (BB ratio) reduction at 28 days (d) old, followed by high mortality 30 d later. We analysed the influence of the IBDV on atrophy of the bursa of fabricius (BF) and the subsequent mortality after 30 d. Ten broilers were sampled at each timepoint from the farm with high mortality at 21, 25, 28 and 35 d. A second flock from the same farm was sampled at 14, 21, 25, 28, 35 and 42 d. IBDV was detected in BF samples at 25, 28 and 35 d and at 21, 25, 28 and 35 d in the first and second flocks, respectively, using immunohistochemical staining and RT-PCR. IBDV isolates from both flocks were closely related to the China KM523643 strain. Histopathology and TUNEL assay indicated apoptosis, severe lymphoid depletion, vacuoles within follicles, lymphoid follicle atrophy and fibrosis in the BF. We observed 75% of the polyserositis and 10% of the airsacculitis at 30 D in dead broilers. The antigenic variant IBDV infection was appeared to be the main influencing factor on BF atrophy and BB ratio reduction in the broilers. High mortality in the broilers after 30 d could be due to secondary infection. The disease caused by IBDV had a negative economic impact in the farm. RESEARCH HIGHLIGHTS New variant IBDV caused bursa atrophy and reduced BB ratio in 28-day-old broilers. After vIBDV had infected broilers, at 21 days old they became immunosuppressed. High mortality at 30 days old in broilers was due to secondary infection. New vIBDV has a negative economic impact on broiler farms in Japan.

A random forest algorithm-based approach to capture latent decision variables and their cutoff values.

Although reference intervals (RIs) and clinical decision limits (CDLs) are vital laboratory information for supporting the interpretation of numerical clinical pathology results, there is evidence that RIs and CDLs vary in certain contexts as well as other evidence that RIs and CDLs are flawed. We propose a random forest algorithm-based exploration methodology by using phenotype transformation of independent variables in relation to dependent variables to capture latent decision variables and their cutoff values. We denote certain CDLs within the RIs estimated by an indirect method that affect some diagnostics or outcomes in the context of specific patients' conditions as latent CDLs. We then apply the proposed methodology to clinical laboratory data regarding bodily fluids, such as blood, urine at the admission of patients for the exploration of latent CDLs of hospital length of stay (HLOS) for each patients' condition identified by diseases of patients who undergoing surgeries. From the exploration results, we found that free Thyroxine (T4) above five unique cutoff values: 1.16 ng/dL, 1.19 ng/dL, 1.2 ng/dL, 1.23 ng/dL and 1.25 ng/dL for tachyarrhythmia predicted longer HLOS, though these cutoff values fall within the estimated RIs as well as the hospital-determined RIs. In addition to the evidence that higher free Thyroxine (T4) levels within the RIs have an association with the corresponding disease, on the whole, the cutoff values except 1.16 ng/dL tended to affect long HLOS with the significant differences. The cutoff values could be taken up for discussion among clinical experts whether it is meaningful to alert the risk of patients' conditions and the long HLOS at the admission of patients. If clinical experts appreciate its meaningfulness in clinical practice, the alerts could be embedded in electronic medical records for handling those risks at the admission of patients.

Causation between Pathway Completion and Reduced Hospital Stay in Patients with Lung Cancer: a Retrospective Cohort Study Using Propensity Score Matching.

We have previously demonstrated that clinical pathway completion helps reduce hospital stays. However, our previous results showed only a correlation, not causation. Therefore, the current study's aim was to analyze the causation between clinical pathway completion and reduced hospital stays for patients with lung cancer. Data were collected from April 2013 to March 2018 from the electronic medical records of the University of Miyazaki Hospital. We used propensity score matching to extract records from 227 patients. Patients were further divided into a pathway completed group and a pathway not completed group; 74 patients in each group were available for data analysis. Our main analysis involved estimating the discharge curve, which was comprised of the in-hospital rate and hospital stay. Additional analyzes were performed to compare the frequency of medical treatments registered in the clinical pathway but not implemented (termed deviated medical treatments). The occurrence of these treatments meant that the clinical pathway was not completed. The main results indicated a decrease in the in-hospital rate of the completion group, compared with the not completed group. The p value of the log-rank test was <0.001 for total patients and patients who underwent resection, and 0.017 for patients who did not undergo resection. Additional results indicated that a number of intravenous drips were not implemented, despite their registration on clinical pathways. Our results indicate that clinical pathway completion contributes to improved efficiency and safety. This simplified procedure is expected to be applicable to other diseases and clinical indicators.

The Development of Medical Markup Language Version 4 as a Clinical Document Exchange Format for Nationwide EHR Systems.

Medical Markup Language (MML) is a standard format for exchange of healthcare data among healthcare providers. Following the last major update (version 3), we developed new modules and discussed the requirements for the next major updates. Subsequently, in 2016 we released MML version 4 and used it to obtain clinical data from healthcare providers for a nationwide electronic health records (EHR) system. In this article we provide an overview of this major update of MML version 4 and discuss its interoperability for clinical data.

Original Laboratory Test Code Mapping System Using Test Result Data on Electronic Health Record.

Laboratory tests results have potential secondary usage. Each healthcare facility has a laboratory test code. Hence, test code mapping is required to support laboratory technicians. An automatic code mapping can reduce the burden of manual mapping during data preparation. The authors developed a semi-automatic mapping support system that uses the newest test results generated in the electronic health record.

Development of an "Alert Framework" Based on the Practices in the Medical Front.

At the University of Miyazaki Hospital (UMH), we have accumulated and semantically structured a vast amount of medical information since the activation of the electronic health record system approximately 10 years ago. With this medical information, we have decided to develop an alert system for aiding in medical treatment. The purpose of this investigation is to not only to integrate an alert framework into the electronic heath record system, but also to formulate a modeling method of this knowledge. A trial alert framework was developed for the staff in various occupational categories at the UMH. Based on findings of subsequent interviews, a more detailed and upgraded alert framework was constructed, resulting in the final model. Based on our current findings, an alert framework was developed with four major items. Based on the analysis of the medical practices from the trial model, it has been concluded that there are four major risk patterns that trigger the alert. Furthermore, the current alert framework contains detailed definitions which are easily substituted into the database, leading to easy implementation of the electronic health records.

Effect on Completion of Clinical Pathway for Improving Clinical Indicator: Cases of Hospital Stay, Mortality Rate, and Comprehensive-Volume Ratio.

Many studies have analyzed the effects of clinical pathways, but most have considered only single diseases. The purpose of the present study was to exploratively analyze electronic medical records related to the use of clinical pathways, seeking trends that could usefully benefit clinical activity. From the data warehouse of University of Miyazaki Hospital, collected from April 2014 to March 2016, we retrospectively identified 6523 patients for whom a clinical pathway was applied. Other inclusion criteria were single hospitalization, the type of medical fee was comprehensive, and data were available so that all clinical indicators could be calculated. Two types of deviation from the clinical pathway were defined: cancellation (described in the clinical pathway but not implemented) and addition (not described in the clinical pathway but nevertheless implemented). If the code of International Classification of Diseases describing the clinical pathway differed from that describing where the medical resource was mostly spent, we considered this as indicating a complication. We compared principal clinical indicators (length of hospital stay, mortality rate, and comprehensive-volume ratio) by completion rate for the clinical pathway. Regardless of whether patients had complications, completing the clinical pathway was associated with a significant reduction in length of hospital stay. This finding indicated that length of hospital stay could be shortened if all medical treatments described in the clinical pathway were implemented. Our results demonstrated that it is possible to shorten the length of hospital stay by improving clinical pathways to include medical treatment for preventing complications.

Privacy Policy Implementation on the Nation-Wide EHR in Japan for Hospitals and Patients.

Shared clinical information is an important contribution to regional medicine. Clinical information sharing with patients is also recommended to motivate patients and promote health. On the other hand, the threat of information leaks, caused by internet connected records, is critical to hospitals. The traditional approach is complete isolation of hospital networks, instead of information sharing. The authors propose methods here to maximize information sharing by following hospital preferences for electronic health records.

Expansion of medical vocabularies using distributional semantics on Japanese patient blogs.

BACKGROUND: Research on medical vocabulary expansion from large corpora has primarily been conducted using text written in English or similar languages, due to a limited availability of large biomedical corpora in most languages. Medical vocabularies are, however, essential also for text mining from corpora written in other languages than English and belonging to a variety of medical genres. The aim of this study was therefore to evaluate medical vocabulary expansion using a corpus very different from those previously used, in terms of grammar and orthographics, as well as in terms of text genre. This was carried out by applying a method based on distributional semantics to the task of extracting medical vocabulary terms from a large corpus of Japanese patient blogs. METHODS: Distributional properties of terms were modelled with random indexing, followed by agglomerative hierarchical clustering of 3 ×100 seed terms from existing vocabularies, belonging to three semantic categories: Medical Finding, Pharmaceutical Drug and Body Part. By automatically extracting unknown terms close to the centroids of the created clusters, candidates for new terms to include in the vocabulary were suggested. The method was evaluated for its ability to retrieve the remaining n terms in existing medical vocabularies. RESULTS: Removing case particles and using a context window size of 1+1 was a successful strategy for Medical Finding and Pharmaceutical Drug, while retaining case particles and using a window size of 8+8 was better for Body Part. For a 10n long candidate list, the use of different cluster sizes affected the result for Pharmaceutical Drug, while the effect was only marginal for the other two categories. For a list of top n candidates for Body Part, however, clusters with a size of up to two terms were slightly more useful than larger clusters. For Pharmaceutical Drug, the best settings resulted in a recall of 25 % for a candidate list of top n terms and a recall of 68 % for top 10n. For a candidate list of top 10n candidates, the second best results were obtained for Medical Finding: a recall of 58 %, compared to 46 % for Body Part. Only taking the top n candidates into account, however, resulted in a recall of 23 % for Body Part, compared to 16 % for Medical Finding. CONCLUSIONS: Different settings for corpus pre-processing, window sizes and cluster sizes were suitable for different semantic categories and for different lengths of candidate lists, showing the need to adapt parameters, not only to the language and text genre used, but also to the semantic category for which the vocabulary is to be expanded. The results show, however, that the investigated choices for pre-processing and parameter settings were successful, and that a Japanese blog corpus, which in many ways differs from those used in previous studies, can be a useful resource for medical vocabulary expansion.

Integrating HL7 RIM and ontology for unified knowledge and data representation in clinical decision support systems.

BACKGROUND AND OBJECTIVES: The broad adoption of clinical decision support systems within clinical practice has been hampered mainly by the difficulty in expressing domain knowledge and patient data in a unified formalism. This paper presents a semantic-based approach to the unified representation of healthcare domain knowledge and patient data for practical clinical decision making applications. METHODS: A four-phase knowledge engineering cycle is implemented to develop a semantic healthcare knowledge base based on an HL7 reference information model, including an ontology to model domain knowledge and patient data and an expression repository to encode clinical decision making rules and queries. A semantic clinical decision support system is designed to provide patient-specific healthcare recommendations based on the knowledge base and patient data. RESULTS: The proposed solution is evaluated in the case study of type 2 diabetes mellitus inpatient management. The knowledge base is successfully instantiated with relevant domain knowledge and testing patient data. Ontology-level evaluation confirms model validity. Application-level evaluation of diagnostic accuracy reaches a sensitivity of 97.5%, a specificity of 100%, and a precision of 98%; an acceptance rate of 97.3% is given by domain experts for the recommended care plan orders. CONCLUSIONS: The proposed solution has been successfully validated in the case study as providing clinical decision support at a high accuracy and acceptance rate. The evaluation results demonstrate the technical feasibility and application prospect of our approach.

Development of Unified Lab Test Result Master for Multiple Facilities.

A clinical study requires massive amounts of of lab test data, especially for rare diseases. Before creating a protocol, the hypothesis if the protocol will work with enough amount of patients' dataset has to be proved. However, a single facility, such as a university hospital, often faces a lack of number of patients for specific target diseases. Even if collecting datasets from several facilities, there is no active master table that can merge lab test results between the facility datasets. Therefore, the authors develop a unified lab test result master. Because test master standards such as JLAC10 and LOINC are provided from a viewpoint of academic classification of laboratory medicine, the classification does not fit clinical classification, which doctors understand with a mind-set of establishing a clinical study protocol. The authors establish a method to unify masters using an active lab test result master from two university hospitals.

Creating personalised clinical pathways by semantic interoperability with electronic health records.

OBJECTIVE: There is a growing realisation that clinical pathways (CPs) are vital for improving the treatment quality of healthcare organisations. However, treatment personalisation is one of the main challenges when implementing CPs, and the inadequate dynamic adaptability restricts the practicality of CPs. The purpose of this study is to improve the practicality of CPs using semantic interoperability between knowledge-based CPs and semantic electronic health records (EHRs). METHODS: Simple protocol and resource description framework query language is used to gather patient information from semantic EHRs. The gathered patient information is entered into the CP ontology represented by web ontology language. Then, after reasoning over rules described by semantic web rule language in the Jena semantic framework, we adjust the standardised CPs to meet different patients' practical needs. RESULTS: A CP for acute appendicitis is used as an example to illustrate how to achieve CP customisation based on the semantic interoperability between knowledge-based CPs and semantic EHRs. A personalised care plan is generated by comprehensively analysing the patient's personal allergy history and past medical history, which are stored in semantic EHRs. Additionally, by monitoring the patient's clinical information, an exception is recorded and handled during CP execution. According to execution results of the actual example, the solutions we present are shown to be technically feasible. CONCLUSION: This study contributes towards improving the clinical personalised practicality of standardised CPs. In addition, this study establishes the foundation for future work on the research and development of an independent CP system.