Sodium-glucose cotransporter 2 inhibitor therapy reduces the administration frequency of steroid injection in patients with diabetic macular edema: A cohort study using the Japanese health insurance claims database.

AIMS/INTRODUCTION: Severe diabetic macular edema (DME) is often resistant to anti-vascular endothelial growth factor therapy. Steroids are particularly effective at reducing edema by suppressing inflammation; they are also used as an alternative to expensive anti-vascular endothelial growth factor therapy in some patients. Therefore, the use of steroids in DME reflects an unmet need for anti-vascular endothelial growth factor therapy. Notably, triamcinolone acetonide (TA) injections are widely used in Japan. Here, we evaluated the frequency of TA as an indicator of the efficacy of sodium-glucose cotransporter 2 inhibitors (SGLT2is) in DME treatment using a health insurance claims database. MATERIALS AND METHODS: In this cohort study, we retrospectively analyzed the health insurance claims data of 11 million Japanese individuals from 2005 to 2019. The frequency and duration of TA injection after the initiation of SGLT2is or other antidiabetic drugs were analyzed. RESULTS: Among the 2,412 matched patients with DME, the incidence rate of TA injection was 63.8 times per 1,000 person-years in SGLT2i users and 94.9 times per 1,000 person-years in non-users. SGLT2is reduced the risk for the first (P = 0.0024, hazard ratio 0.66, 95% confidence interval 0.50-0.87), second (P = 0.0019, hazard ratio 0.53, 95% confidence interval 0.35-0.80) and third TA (P = 0.0053, hazard ratio 0.44, 95% confidence interval 0.25-0.80) injections. A subanalysis of each baseline characteristic of the patients showed that SGLT2is were effective regardless of the background factors. CONCLUSIONS: The use of SGLT2is reduced the frequency of TA injection in patients with DME. Therefore, SGLT2i therapy might be a novel, noninvasive and low-cost adjunctive therapy for DME.

ARID5B is a negative modulator of IL-6 production in rheumatoid arthritis synovial fibroblasts.

Recent single-cell RNA-sequencing analysis of rheumatoid arthritis (RA) synovial tissues revealed the heterogeneity of RA synovial fibroblasts (SFs) with distinct functions such as high IL-6 production. The molecular mechanisms responsible for high IL-6 production will become a promising drug target of RASFs to treat RA. In this study, we performed siRNA screening of 65 transcription factors (TFs) differentially expressed among RASF subsets to identify TFs involved in IL-6 production. The siRNA screening identified 7 TFs including ARID5B, a RA risk gene, that affected IL-6 production. Both long and short isoforms of ARID5B were expressed and negatively regulated by TNF-α in RASFs. The siRNA knockdown and lentiviral overexpression of long and short isoforms of ARID5B revealed that the long isoform suppressed IL-6 production stimulated with TNF-α. eQTL analysis using 58 SFs demonstrated that RA risk allele, rs10821944, in intron 4 of the ARID5B gene had a trend of eQTL effects to the expression of long isoform of ARID5B in SFs treated with TNF-α. ARID5B was found to be a negative modulator of IL-6 production in RASFs. The RA risk allele of ARID5B intron may cause high IL-6 production, suggesting that ARID5B will become a promising drug target to treat RA.

Memory CD8 T cells are vulnerable to chronic IFN-γ signals but not to CD4 T cell deficiency in MHCII-deficient mice.

The mechanisms by which the number of memory CD8 T cells is stably maintained remains incompletely understood. It has been postulated that maintaining them requires help from CD4 T cells, because adoptively transferred memory CD8 T cells persist poorly in MHC class II (MHCII)-deficient mice. Here we show that chronic interferon-γ signals, not CD4 T cell-deficiency, are responsible for their attrition in MHCII-deficient environments. Excess IFN-γ is produced primarily by endogenous colonic CD8 T cells in MHCII-deficient mice. IFN-γ neutralization restores the number of memory CD8 T cells in MHCII-deficient mice, whereas repeated IFN-γ administration or transduction of a gain-of-function STAT1 mutant reduces their number in wild-type mice. CD127high memory cells proliferate actively in response to IFN-γ signals, but are more susceptible to attrition than CD127low terminally differentiated effector memory cells. Furthermore, single-cell RNA-sequencing of memory CD8 T cells reveals proliferating cells that resemble short-lived, terminal effector cells and documents global downregulation of gene signatures of long-lived memory cells in MHCII-deficient environments. We propose that chronic IFN-γ signals deplete memory CD8 T cells by compromising their long-term survival and by diverting self-renewing CD127high cells toward terminal differentiation.

Dimension reduction of microbiome data linked Bifidobacterium and Prevotella to allergic rhinitis.

Dimension reduction has been used to visualise the distribution of multidimensional microbiome data, but the composite variables calculated by the dimension reduction methods have not been widely used to investigate the relationship of the human gut microbiome with lifestyle and disease. In the present study, we applied several dimension reduction methods, including principal component analysis, principal coordinate analysis (PCoA), non-metric multidimensional scaling (NMDS), and non-negative matrix factorization, to a microbiome dataset from 186 subjects with symptoms of  allergic rhinitis (AR) and 106 controls. All the dimension reduction methods supported that the distribution of microbial data points appeared to be continuous rather than discrete. Comparison of the composite variables calculated from the different dimension reduction methods showed that the characteristics of the composite variables differed depending on the distance matrices and the dimension reduction methods. The first composite variables calculated from PCoA and NMDS with the UniFrac distance were strongly associated with AR (FDR adjusted P = 2.4 × 10-4 for PCoA and P = 2.8 × 10-4 for NMDS), and also with the relative abundance of Bifidobacterium and Prevotella. The abundance of Bifidobacterium was also linked to intake of several nutrients, including carbohydrate, saturated fat, and alcohol via composite variables. Notably, the association between the composite variables and AR was much stronger than the association between the relative abundance of individual genera and AR. Our results highlight the usefulness of the dimension reduction methods for investigating the association of microbial composition with lifestyle and disease in clinical research.

Preoperative prediction for periprosthetic bone loss and individual evaluation of bisphosphonate effect after total hip arthroplasty using artificial intelligence.

Aims: This study was designed to develop a model for predicting bone mineral density (BMD) loss of the femur after total hip arthroplasty (THA) using artificial intelligence (AI), and to identify factors that influence the prediction. Additionally, we virtually examined the efficacy of administration of bisphosphonate for cases with severe BMD loss based on the predictive model. Methods: The study included 538 joints that underwent primary THA. The patients were divided into groups using unsupervised time series clustering for five-year BMD loss of Gruen zone 7 postoperatively, and a machine-learning model to predict the BMD loss was developed. Additionally, the predictor for BMD loss was extracted using SHapley Additive exPlanations (SHAP). The patient-specific efficacy of bisphosphonate, which is the most important categorical predictor for BMD loss, was examined by calculating the change in predictive probability when hypothetically switching between the inclusion and exclusion of bisphosphonate. Results: Time series clustering allowed us to divide the patients into two groups, and the predictive factors were identified including patient- and operation-related factors. The area under the receiver operating characteristic (ROC) curve (AUC) for the BMD loss prediction averaged 0.734. Virtual administration of bisphosphonate showed on average 14% efficacy in preventing BMD loss of zone 7. Additionally, stem types and preoperative triglyceride (TG), creatinine (Cr), estimated glomerular filtration rate (eGFR), and creatine kinase (CK) showed significant association with the estimated patient-specific efficacy of bisphosphonate. Conclusion: Periprosthetic BMD loss after THA is predictable based on patient- and operation-related factors, and optimal prescription of bisphosphonate based on the prediction may prevent BMD loss.

Generative artificial intelligence and non-pharmacological bias: an experimental study on cancer patient sexual health communications.

Objectives The objective of this study was to explore the feature of generative artificial intelligence (AI) in asking sexual health among cancer survivors, which are often challenging for patients to discuss.Methods We employed the Generative Pre-trained Transformer-3.5 (GPT) as the generative AI platform and used DocsBot for citation retrieval (June 2023). A structured prompt was devised to generate 100 questions from the AI, based on epidemiological survey data regarding sexual difficulties among cancer survivors. These questions were submitted to Bot1 (standard GPT) and Bot2 (sourced from two clinical guidelines).Results No censorship of sexual expressions or medical terms occurred. Despite the lack of reflection on guideline recommendations, 'consultation' was significantly more prevalent in both bots' responses compared with pharmacological interventions, with ORs of 47.3 (p<0.001) in Bot1 and 97.2 (p<0.001) in Bot2.Discussion Generative AI can serve to provide health information on sensitive topics such as sexual health, despite the potential for policy-restricted content. Responses were biased towards non-pharmacological interventions, which is probably due to a GPT model designed with the 's prohibition policy on replying to medical topics. This shift warrants attention as it could potentially trigger patients' expectations for non-pharmacological interventions.

Explainable Machine Learning Classification to Identify Vulnerable Groups Among Parenting Mothers: Web-Based Cross-Sectional Questionnaire Study.

BACKGROUND: One life event that requires extensive resilience and adaptation is parenting. However, resilience and perceived support in child-rearing vary, making the real-world situation unclear, even with postpartum checkups. OBJECTIVE: This study aimed to explore the psychosocial status of mothers during the child-rearing period from newborn to toddler, with a classifier based on data on the resilience and adaptation characteristics of mothers with newborns. METHODS: A web-based cross-sectional survey was conducted. Mothers with newborns aged approximately 1 month (newborn cohort) were analyzed to construct an explainable machine learning classifier to stratify parenting-related resilience and adaptation characteristics and identify vulnerable populations. Explainable k-means clustering was used because of its high explanatory power and applicability. The classifier was applied to mothers with infants aged 2 months to 1 year (infant cohort) and mothers with toddlers aged >1 year to 2 years (toddler cohort). Psychosocial status, including depressed mood assessed by the Edinburgh Postnatal Depression Scale (EPDS), bonding assessed by the Postpartum Bonding Questionnaire (PBQ), and sleep quality assessed by the Pittsburgh Sleep Quality Index (PSQI) between the classified groups, was compared. RESULTS: A total of 1559 participants completed the survey. They were split into 3 cohorts, comprising populations of various characteristics, including parenting difficulties and psychosocial measures. The classifier, which stratified participants into 5 groups, was generated from the self-reported scores of resilience and adaptation in the newborn cohort (n=310). The classifier identified that the group with the greatest difficulties in resilience and adaptation to a child's temperament and perceived support had higher incidences of problems with depressed mood (relative prevalence [RP] 5.87, 95% CI 2.77-12.45), bonding (RP 5.38, 95% CI 2.53-11.45), and sleep quality (RP 1.70, 95% CI 1.20-2.40) compared to the group with no difficulties in perceived support. In the infant cohort (n=619) and toddler cohort (n=461), the stratified group with the greatest difficulties had higher incidences of problems with depressed mood (RP 9.05, 95% CI 4.36-18.80 and RP 4.63, 95% CI 2.38-9.02, respectively), bonding (RP 1.63, 95% CI 1.29-2.06 and RP 3.19, 95% CI 2.03-5.01, respectively), and sleep quality (RP 8.09, 95% CI 4.62-16.37 and RP 1.72, 95% CI 1.23-2.42, respectively) compared to the group with no difficulties. CONCLUSIONS: The classifier, based on a combination of resilience and adaptation to the child's temperament and perceived support, was able identify psychosocial vulnerable groups in the newborn cohort, the start-up stage of childcare. Psychosocially vulnerable groups were also identified in qualitatively different infant and toddler cohorts, depending on their classifier. The vulnerable group identified in the infant cohort showed particularly high RP for depressed mood and poor sleep quality.

Sodium-glucose co-transporter 2 inhibitor therapy reduces the administration frequency of anti-vascular endothelial growth factor agents in patients with diabetic macular oedema with a history of anti-vascular endothelial growth factor agent use: A cohort study using the Japanese health insurance claims database.

AIM: We assessed the effectiveness of sodium-glucose co-transporter 2 inhibitors (SGLT2is) in reducing the administration frequency of anti-vascular endothelial growth factor (VEGF) agents in patients with diabetic macular oedema (DMO) using a health insurance claims database. MATERIALS AND METHODS: This retrospective cohort study analysed health insurance claims data covering 11 million Japanese patients between 2005 and 2019. We analysed the frequency and duration of intravitreal injection of anti-VEGF agents after initiating SGLT2is or other antidiabetic drugs. RESULTS: Among 2412 matched patients with DMO, the incidence rates of anti-VEGF agent injections were 230.1 per 1000 person-year in SGLT2i users and 228.4 times per 1000 person-year in non-users, respectively, and the risk ratio for events was unchanged in both groups. Sub-analysis of each baseline characteristic of the patients showed that SGLT2is were particularly effective in patients with a history of anti-VEGF agent use [p = .027, hazard ratio (HR): 0.44, 95% confidence interval (CI): 0.22-0.91]. SGLT2is reduced the risk for the first (p = .023, HR: 0.45, 95% CI: 0.22-0.91) and second (p = .021, HR: 0.39, 95% CI: 0.17-0.89) anti-VEGF agent injections. CONCLUSIONS: There was no difference in the risk ratio for the addition of anti-VEGF therapy between the two treatment groups. However, the use of SGLT2is reduced the frequency of anti-VEGF agent administration in patients with DMO requiring anti-VEGF therapy. Therefore, SGLT2i therapy may be a novel, non-invasive, low-cost adjunctive therapy for DMO requiring anti-VEGF therapy.

Best practices for multimodal clinical data management and integration: An atopic dermatitis research case.

BACKGROUND: In clinical research on multifactorial diseases such as atopic dermatitis, data-driven medical research has become more widely used as means to clarify diverse pathological conditions and to realize precision medicine. However, modern clinical data, characterized as large-scale, multimodal, and multi-center, causes difficulties in data integration and management, which limits productivity in clinical data science. METHODS: We designed a generic data management flow to collect, cleanse, and integrate data to handle different types of data generated at multiple institutions by 10 types of clinical studies. We developed MeDIA (Medical Data Integration Assistant), a software to browse the data in an integrated manner and extract subsets for analysis. RESULTS: MeDIA integrates and visualizes data and information on research participants obtained from multiple studies. It then provides a sophisticated interface that supports data management and helps data scientists retrieve the data sets they need. Furthermore, the system promotes the use of unified terms such as identifiers or sampling dates to reduce the cost of pre-processing by data analysts. We also propose best practices in clinical data management flow, which we learned from the development and implementation of MeDIA. CONCLUSIONS: The MeDIA system solves the problem of multimodal clinical data integration, from complex text data such as medical records to big data such as omics data from a large number of patients. The system and the proposed best practices can be applied not only to allergic diseases but also to other diseases to promote data-driven medical research.

Developing Medical Education Curriculum Reform Strategies to Address the Impact of Generative AI: Qualitative Study.

BACKGROUND: Generative artificial intelligence (GAI), represented by large language models, have the potential to transform health care and medical education. In particular, GAI's impact on higher education has the potential to change students' learning experience as well as faculty's teaching. However, concerns have been raised about ethical consideration and decreased reliability of the existing examinations. Furthermore, in medical education, curriculum reform is required to adapt to the revolutionary changes brought about by the integration of GAI into medical practice and research. OBJECTIVE: This study analyzes the impact of GAI on medical education curricula and explores strategies for adaptation. METHODS: The study was conducted in the context of faculty development at a medical school in Japan. A workshop involving faculty and students was organized, and participants were divided into groups to address two research questions: (1) How does GAI affect undergraduate medical education curricula? and (2) How should medical school curricula be reformed to address the impact of GAI? The strength, weakness, opportunity, and threat (SWOT) framework was used, and cross-SWOT matrix analysis was used to devise strategies. Further, 4 researchers conducted content analysis on the data generated during the workshop discussions. RESULTS: The data were collected from 8 groups comprising 55 participants. Further, 5 themes about the impact of GAI on medical education curricula emerged: improvement of teaching and learning, improved access to information, inhibition of existing learning processes, problems in GAI, and changes in physicians' professionality. Positive impacts included enhanced teaching and learning efficiency and improved access to information, whereas negative impacts included concerns about reduced independent thinking and the adaptability of existing assessment methods. Further, GAI was perceived to change the nature of physicians' expertise. Three themes emerged from the cross-SWOT analysis for curriculum reform: (1) learning about GAI, (2) learning with GAI, and (3) learning aside from GAI. Participants recommended incorporating GAI literacy, ethical considerations, and compliance into the curriculum. Learning with GAI involved improving learning efficiency, supporting information gathering and dissemination, and facilitating patient involvement. Learning aside from GAI emphasized maintaining GAI-free learning processes, fostering higher cognitive domains of learning, and introducing more communication exercises. CONCLUSIONS: This study highlights the profound impact of GAI on medical education curricula and provides insights into curriculum reform strategies. Participants recognized the need for GAI literacy, ethical education, and adaptive learning. Further, GAI was recognized as a tool that can enhance efficiency and involve patients in education. The study also suggests that medical education should focus on competencies that GAI hardly replaces, such as clinical experience and communication. Notably, involving both faculty and students in curriculum reform discussions fosters a sense of ownership and ensures broader perspectives are encompassed.

Multifaceted analysis of cross-tissue transcriptomes reveals phenotype-endotype associations in atopic dermatitis.

Atopic dermatitis (AD) is a skin disease that is heterogeneous both in terms of clinical manifestations and molecular profiles. It is increasingly recognized that AD is a systemic rather than a local disease and should be assessed in the context of whole-body pathophysiology. Here we show, via integrated RNA-sequencing of skin tissue and peripheral blood mononuclear cell (PBMC) samples along with clinical data from 115 AD patients and 14 matched healthy controls, that specific clinical presentations associate with matching differential molecular signatures. We establish a regression model based on transcriptome modules identified in weighted gene co-expression network analysis to extract molecular features associated with detailed clinical phenotypes of AD. The two main, qualitatively differential skin manifestations of AD, erythema and papulation are distinguished by differential immunological signatures. We further apply the regression model to a longitudinal dataset of 30 AD patients for personalized monitoring, highlighting patient heterogeneity in disease trajectories. The longitudinal features of blood tests and PBMC transcriptome modules identify three patient clusters which are aligned with clinical severity and reflect treatment history. Our approach thus serves as a framework for effective clinical investigation to gain a holistic view on the pathophysiology of complex human diseases.

Non-negative tensor factorization workflow for time series biomedical data.

Non-negative tensor factorization (NTF) enables the extraction of a small number of latent components from high-dimensional biomedical data. However, NTF requires many steps, which is a hurdle to implementation. Here, we provide a protocol for TensorLyCV, an easy to run and reproducible NTF analysis pipeline using Snakemake workflow management system and Docker container. Using vaccine adverse reaction data as an example, we describe steps for data processing, tensor decomposition, optimal rank parameter estimation, and visualization of factor matrices. For complete details on the use and execution of this protocol, please refer to Kei Ikeda et al.1.

Synovial Tissue Heterogeneity in Japanese Patients With Rheumatoid Arthritis Elucidated Using a Cell-Type Deconvolution Approach.

OBJECTIVE: Recent advances in single-cell RNA sequencing technology have improved our understanding of the immunological landscape of rheumatoid arthritis (RA). We aimed to stratify the synovium from East Asian patients with RA by immune cell compositions and gain insight into the inflammatory drivers of each synovial phenotype. METHODS: Synovial tissues were obtained from East Asian patients in Japan with RA (n = 41) undergoing articular surgery. The cellular composition was quantified by a deconvolution approach using a public single-cell-based reference. Inflammatory pathway activity was calculated by gene set variation analysis, and chromatin accessibility was evaluated using assay of transposase accessible chromatin-sequencing. RESULTS: We stratified RA synovium into three distinct subtypes based on the hierarchical clustering of cellular composition data. One subtype was characterized by abundant HLA-DRAhigh synovial fibroblasts, autoimmune-associated B cells, GZMK+ GZMB+ CD8+ T cells, interleukin (IL)1-ß+ monocytes, and plasmablasts. In addition, tumor necrosis factor (TNF)-α, interferons (IFNs), and IL-6 signaling were highly activated in this subtype, and the expression of various chemokines was significantly enhanced. Moreover, we found an open chromatin region overlapping with RA risk locus rs9405192 near the IRF4 gene, suggesting the genetic background influences the development of this inflammatory synovial state. The other two subtypes were characterized by increased IFNs and IL-6 signaling, and expression of molecules associated with degeneration, respectively. CONCLUSION: This study adds insights into the synovial heterogeneity in East Asian patients and shows a promising link with predominant inflammatory signals. Evaluating the site of inflammation has the potential to lead to appropriate drug selection that matches the individual pathology.

Predicting dry weight change in Hemodialysis patients using machine learning.

BACKGROUND: Machine Learning has been increasingly used in the medical field, including managing patients undergoing hemodialysis. The random forest classifier is a Machine Learning method that can generate high accuracy and interpretability in the data analysis of various diseases. We attempted to apply Machine Learning to adjust dry weight, the appropriate volume status of patients undergoing hemodialysis, which requires a complex decision-making process considering multiple indicators and the patient's physical conditions. METHODS: All medical data and 69,375 dialysis records of 314 Asian patients undergoing hemodialysis at a single dialysis center in Japan between July 2018 and April 2020 were collected from the electronic medical record system. Using the random forest classifier, we developed models to predict the probabilities of adjusting the dry weight at each dialysis session. RESULTS: The areas under the receiver-operating-characteristic curves of the models for adjusting the dry weight upward and downward were 0.70 and 0.74, respectively. The average probability of upward adjustment of the dry weight had sharp a peak around the actual change over time, while the average probability of downward adjustment of the dry weight formed a gradual peak. Feature importance analysis revealed that median blood pressure decline was a strong predictor for adjusting the dry weight upward. In contrast, elevated serum levels of C-reactive protein and hypoalbuminemia were important indicators for adjusting the dry weight downward. CONCLUSIONS: The random forest classifier should provide a helpful guide to predict the optimal changes to the dry weight with relative accuracy and may be useful in clinical practice.

Protocol to acquire time series data on adverse reactions following vaccination using a smartphone or web-based platform.

Data collection on adverse reactions in recipients after vaccination is vital to evaluate potential health issues, but health observation diaries are onerous for participants. Here, we present a protocol to collect time series information using a smartphone or web-based platform, thus eliminating the need for paperwork and data submission. We describe steps for setting up the platform using the Model-View-Controller web framework, uploading lists of recipients, sending notifications, and managing respondent data. For complete details on the use and execution of this protocol, please refer to Ikeda et al. (2022).1.

Machine-learning predicts time-series prognosis factors in metastatic prostate cancer patients treated with androgen deprivation therapy.

Machine learning technology is expected to support diagnosis and prognosis prediction in medicine. We used machine learning to construct a new prognostic prediction model for prostate cancer patients based on longitudinal data obtained from age at diagnosis, peripheral blood and urine tests of 340 prostate cancer patients. Random survival forest (RSF) and survival tree were used for machine learning. In the time-series prognostic prediction model for metastatic prostate cancer patients, the RSF model showed better prediction accuracy than the conventional Cox proportional hazards model for almost all time periods of progression-free survival (PFS), overall survival (OS) and cancer-specific survival (CSS). Based on the RSF model, we created a clinically applicable prognostic prediction model using survival trees for OS and CSS by combining the values of lactate dehydrogenase (LDH) before starting treatment and alkaline phosphatase (ALP) at 120 days after treatment. Machine learning provides useful information for predicting the prognosis of metastatic prostate cancer prior to treatment intervention by considering the nonlinear and combined impacts of multiple features. The addition of data after the start of treatment would allow for more precise prognostic risk assessment of patients and would be beneficial for subsequent treatment selection.

Monitoring of blood biochemical markers for periprosthetic joint infection using ensemble machine learning and UMAP embedding.

INTRODUCTION: Periprosthetic joint infection (PJI) is a serious complication after total joint arthroplasty. It is important to accurately identify PJI and monitor postoperative blood biochemical marker changes for the appropriate treatment strategy. In this study, we aimed to monitor the postoperative blood biochemical characteristics of PJI by contrasting with non-PJI joint replacement cases to understand how the characteristics change postoperatively. MATERIALS AND METHODS: A total of 144 cases (52 of PJI and 92 of non-PJI) were reviewed retrospectively and split into development and validation cohorts. After exclusion of 11 cases, a total of 133 (PJI: 50, non-PJI: 83) cases were enrolled finally. An RF classifier was developed to discriminate between PJI and non-PJI cases based on 18 preoperative blood biochemical tests. We evaluated the similarity/dissimilarity between cases based on the RF model and embedded the cases in a two-dimensional space by Uniform Manifold Approximation and Projection (UMAP). The RF model developed based on preoperative data was also applied to the same 18 blood biochemical tests at 3, 6, and 12 months after surgery to analyze postoperative pathological changes in PJI and non-PJI. A Markov chain model was applied to calculate the transition probabilities between the two clusters after surgery. RESULTS: PJI and non-PJI were discriminated with the RF classifier with the area under the receiver operating characteristic curve of 0.778. C-reactive protein, total protein, and blood urea nitrogen were identified as the important factors that discriminates between PJI and non-PJI patients. Two clusters corresponding to the high- and low-risk populations of PJI were identified in the UMAP embedding. The high-risk cluster, which included a high proportion of PJI patients, was characterized by higher CRP and lower hemoglobin. The frequency of postoperative recurrence to the high-risk cluster was higher in PJI than in non-PJI. CONCLUSIONS: Although there was overlap between PJI and non-PJI, we were able to identify subgroups of PJI in the UMAP embedding. The machine-learning-based analytical approach is promising in consecutive monitoring of diseases such as PJI with a low incidence and long-term course.

Machine learning approach to stratify complex heterogeneity of chronic heart failure: A report from the CHART-2 study.

AIMS: Current approaches to classify chronic heart failure (HF) subpopulations may be limited due to the diversity of pathophysiology and co-morbidities in chronic HF. We aimed to elucidate the clusters of chronic patients with HF by data-driven approaches with machine learning in a hospital-based registry. METHODS AND RESULTS: A total of 4649 patients with a broad spectrum of left ventricular ejection fraction (LVEF) in the CHART-2 (Chronic Heart Failure Analysis and Registry in the Tohoku District-2) study were enrolled to this study. Chronic HF patients were classified using random forest clustering with 56 multiscale clinical parameters. We assessed the influence of the clusters on cardiovascular death, non-cardiovascular death, all-cause death, and free from hospitalization by HF. Latent class analysis using random forest clustering identified 10 clusters with four primary components: cardiac function (LVEF, left atrial and ventricular diameters, diastolic blood pressure, and brain natriuretic peptide), renal function (glomerular filtration rate and blood urea nitrogen), anaemia (red blood cell, haematocrit, haemoglobin, and platelet count), and nutrition (albumin and body mass index). All 11 significant clinical parameters in the four primary components and two disease aetiologies (ischaemic heart disease and valvular heart disease) showed statistically significant differences among the 10 clusters (P < 0.01). Cluster 1 (26.7% of patients), which is characterized by preserved LVEF (<59%, 37% of the total) with lowest brain natriuretic peptide (>111.3 pg/mL, 0.9%) and lowest left atrial diameter (>42 mm, 37.4%), showed the best 5 year survival rate of 98.1% for cardiovascular death, 95.9% for non-cardiovascular death, 92.9% for all-cause death, and 91.7% for free from hospitalization by HF. Cluster 10 (6.0% of the total), which is co-morbid disorders of all four primary components, showed the worst survival rate of 39.1% for cardiovascular death, 68.9% for non-cardiovascular death, 23.9% for all-cause death, and 28.1% for free from hospitalization by HF. CONCLUSIONS: These results suggest the potential applicability of the machine leaning approach, providing useful clinical prognostic information to stratify complex heterogeneity in patients with HF.