Improved Evaluation Metrics for Sentence Suggestions in Nursing and Elderly Care Record Applications

This paper presents a new approach called EmbedHDP, which aims to enhance the evaluation models utilized for assessing sentence suggestions in nursing care record applications. The primary objective is to determine the alignment of the proposed evaluation metric with human evaluators who are caregivers. It is crucial due to the direct relevance of the provided provided to the health or condition of the elderly. The motivation for this proposal arises from challenges observed in previous models. Our analysis examines the mechanisms of current evaluation metrics such as BERTScore, cosine similarity, ROUGE, and BLEU to achieve reliable metrics evaluation. Several limitations were identified. In some cases, BERTScore encountered difficulties in effectively evaluating the nursing care record domain and consistently providing quality assessments of generated sentence suggestions above 60%. Cosine similarity is a widely used method, but it has limitations regarding word order. This can lead to potential misjudgments of semantic differences within similar word sets. Another technique, ROUGE, relies on lexical overlap but tends to ignore semantic accuracy. Additionally, while BLEU is helpful, it may not fully capture semantic coherence in its evaluations. After calculating the correlation coefficient, it was found that EmbedHDP is effective in evaluating nurse care records due to its ability to handle a variety of sentence structures and medical terminology, providing differentiated and contextually relevant assessments. Additionally, this research used a dataset comprising 320 pairs of sentences with correspondingly equivalent lengths. The results revealed that EmbedHDP outperformed other evaluation models, achieving a coefficient score of 61%, followed by cosine similarity, with a score of 59%, and BERTScore, with 58%. This shows the effectiveness of our proposed approach in improving the evaluation of sentence suggestions in nursing care record applications.

Relabeling for Indoor Localization Using Stationary Beacons in Nursing Care Facilities

In this study, we propose an augmentation method for machine learning based on relabeling data in caregiving and nursing staff indoor localization with Bluetooth Low Energy (BLE) technology. Indoor localization is used to monitor staff-to-patient assistance in caregiving and to gain insights into workload management. However, improving accuracy is challenging when there is a limited amount of data available for training. In this paper, we propose a data augmentation method to reuse the Received Signal Strength (RSS) from different beacons by relabeling to the locations with less samples, resolving data imbalance. Standard deviation and Kullback-Leibler divergence between minority and majority classes are used to measure signal pattern to find matching beacons to relabel. By matching beacons between classes, two variations of relabeling are implemented, specifically full and partial matching. The performance is evaluated using the real-world dataset we collected for five days in a nursing care facility installed with 25 BLE beacons. A Random Forest model is utilized for location recognition, and performance is compared using the weighted F1-score to account for class imbalance. By increasing the beacon data with our proposed relabeling method for data augmentation, we achieve a higher minority class F1-score compared to augmentation with Random Sampling, Synthetic Minority Oversampling Technique (SMOTE) and Adaptive Synthetic Sampling (ADASYN). Our proposed method utilizes collected beacon data by leveraging majority class samples. Full matching demonstrated a 6 to 8% improvement from the original baseline overall weighted F1-score.

Optimizing Forecasted Activity Notifications with Reinforcement Learning

In this paper, we propose the notification optimization method by providing multiple alternative times as a reminder for a forecasted activity with and without probabilistic considerations for the activity that needs to be completed and needs notification. It is important to consider various factors when sending notifications to people after obtaining the results of the forecasted activity. We should not send notifications only when we have forecasted results because future daily activities are unpredictable. Therefore, it is important to strike a balance between providing useful reminders and avoiding excessive interruptions, especially for low probabilities of forecasted activity. Our study investigates the impact of the low probability of forecasted activity and optimizes the notification time with reinforcement learning. We also show the gaps between forecasted activities that are useful for self-improvement by people for the balance of important tasks, such as tasks completed as planned and additional tasks to be completed. For evaluation, we utilize two datasets: the existing dataset and data we collected in the field with the technology we have developed. In the data collection, we have 23 activities from six participants. To evaluate the effectiveness of these approaches, we assess the percentage of positive responses, user response rate, and response duration as performance criteria. Our proposed method provides a more effective way to optimize notifications. By incorporating the probability level of activity that needs to be done and needs notification into the state, we achieve a better response rate than the baseline, with the advantage of reaching 27.15%, as well as than the other criteria, which are also improved by using probability.

Impact of low-dose or under-dose direct oral anticoagulant on coagulation and fibrinolytic markers in patients with atrial fibrillation

Background

Atrial fibrillation (AF) is known to increase the risks of cerebral and systemic embolism. Apart from vitamin K antagonists, edoxaban, a direct oral anticoagulant (DOAC), has been approved for oral anticoagulation in patients with non-valvular AF. On the other hand, DOACs are sometimes prescribed at off-label under-doses for patients who have undergone ablation for AF. Prothrombin fragment F1+2 is an activation peptide released from prothrombin during thrombin formation. The purpose of this study is to compare the effects of DOAC doses on coagulation and fibrinolytic markers.

Methods and results

A total of 88 patients with AF (age: 68±11 years, male:45%, paroxysmal AF n=49, persistent AF n=39) were recruited. All patients were received edoxaban (60mg or 30mg) once a day. For the purpose of the study, patients were divided into three groups according to whether they had been treated before the ablation procedure under an appropriate standard dose group (n=30 [34.1%]), appropriate low-dose group (n=35 [39.8%]), or off-label under-dose group (n=23 [26.1%]). We examined the coagulation and fibrinolytic markers, and echocardiographic parameters before ablation. All patients were followed up for 12 months after AF ablation. Creatinine clearance was significantly higher in appropriate standard-dose group than in appropriate low-dose or off-label under-dose group (101.1±38.4, 57.1±15.9 and 73.2±14.6 mL/min, respectively; P<0.001). There were no significant baseline differences in AF type, history of stroke/transient ischemic attack (TIA), plasma B-type natriuretic peptide, protein C, fibrinogen, D-dimer level, left ventricular ejection fraction or left atrium dimension among the three groups. Prothrombin fragment F1+2 level was significantly lower in the appropriate standard-dose group than the appropriate low-dose and off-label under-dose groups (105.9±29.4, 142.6±41.3 and 142.8±84.9 pmol/L, respectively; P=0.011, Figure). One patient in the appropriate low-dose group had a TIA and 1 patient in the off-label under-dose group had a bleeding event during the follow up period after ablation.

Conclusion

Our results suggest that an appropriate standard dose of edoxaban is needed to suppress hypercoagulability in patients with AF.

Funding Acknowledgement

Type of funding sources: None. Prothrombin fragment F1+2 level

Integrating a spoken dialogue system, nursing records, and activity data collection based on smartphones

BACKGROUND AND OBJECTIVE

This study describes the integration of a spoken dialogue system and nursing records on an Android smartphone application intending to help nurses reduce documentation time and improve the overall experience of a healthcare setting. The application also incorporates with collecting personal sensor data and activity labels for activity recognition.

METHODS

We developed a joint model based on a bidirectional long-short term memory and conditional random fields (Bi-LSTM-CRF) to identify user intention and extract record details from user utterances. Then, we transformed unstructured data into record inputs on the smartphone application.

RESULTS

The joint model achieved the highest F1-score at 96.79%. Moreover, we conducted an experiment to demonstrate the proposed model's capability and feasibility in recording in realistic settings. Our preliminary evaluation results indicate that when using the dialogue-based, we could increase the percentage of documentation speed to 58.13% compared to the traditional keyboard-based.

CONCLUSIONS

Based on our findings, we highlight critical and promising future research directions regarding the design of the efficient spoken dialogue system and nursing records.

Risk factors for thrombocytopenia and analysis of time to platelet transfusion after azacitidine treatment

The use of azacitidine (AZA) has been known to lead to a high incidence of hematotoxic adverse events. The aims of this study were to identify the risk factors for thrombocytopenia after the administration of AZA and to analyze time to the initial platelet transfusion. Sixty-two patients with myelodysplastic syndrome (MDS), who were treated with AZA in Gifu Municipal Hospital between March 2012 and June 2020, were included in this study. The risk factors for thrombocytopenia were identified using univariate analysis of patient characteristics, disease type, and laboratory values immediately before the start of treatment. Variables with p<0.2 identified in the univariate analysis were used as independent variables in the multivariate analysis. This analysis identified "creatinine clearance (CCr) <60 mL/min" as a significant factor (odds ratio, 4.790; 95% confidence interval [CI], 1.380-16.70; p=0.014). Subsequently, time in days to the initial platelet transfusion after the initial administration of AZA was analyzed using the log-rank test. The overall median time in days to platelet transfusion was 370 days. The log-rank test was used to determine the influence of patient characteristics, disease type, and laboratory values immediately before the start of treatment. The subsequent Cox proportional hazard regression analysis using variables with p<0.2 as independent variables identified "hemoglobin (Hb) <8.0 g/dL" as a significant factor (hazard ratio, 2.143; 95% CI, 1.001-4.573; p=0.048). The results of this study led to the following clinical implications: first, patients with CCr of <60 mL/min at the start of treatment should be treated with caution due to the risk of thrombocytopenia. Second, patients with Hb of <8.0 g/dL at the start of treatment may require platelet transfusion in the early stage of treatment.

On-Device Deep Personalization for Robust Activity Data Collection

One of the biggest challenges of activity data collection is the need to rely on users and keep them engaged to continually provide labels. Recent breakthroughs in mobile platforms have proven effective in bringing deep neural networks powered intelligence into mobile devices. This study proposes a novel on-device personalization for data labeling for an activity recognition system using mobile sensing. The key idea behind this system is that estimated activities personalized for a specific individual user can be used as feedback to motivate user contribution and improve data labeling quality. First, we exploited fine-tuning using a Deep Recurrent Neural Network to address the lack of sufficient training data and minimize the need for training deep learning on mobile devices from scratch. Second, we utilized a model pruning technique to reduce the computation cost of on-device personalization without affecting the accuracy. Finally, we built a robust activity data labeling system by integrating the two techniques outlined above, allowing the mobile application to create a personalized experience for the user. To demonstrate the proposed model’s capability and feasibility, we developed and deployed the proposed system to realistic settings. For our experimental setup, we gathered more than 16,800 activity windows from 12 activity classes using smartphone sensors. We empirically evaluated the proposed quality by comparing it with a baseline using machine learning. Our results indicate that the proposed system effectively improved activity accuracy recognition for individual users and reduced cost and latency for inference for mobile devices. Based on our findings, we highlight critical and promising future research directions regarding the design of efficient activity data collection with on-device personalization.

Impact of stress myocardial blood flow as an important predictor for major adverse cardiac and cerebrovascular event in hemodialysis patients, even in patients without myocardial perfusion abnormality

Background

In the clinical setting, ischemic heart disease (IHD) is a major problem not only in general patients but also in regular hemodialysis (HD) patients. Positron emission tomography (PET) is becoming a reliable modality for detecting coronary artery disease. Of course, PET illustrates myocardial perfusion (MP), PET also measures myocardial blood flow (MBF) directly. We have reported stress MBF is an independent predictor in HD population. Although some prior studies show CFR is an independent predictor for their prognosis in patients without MP abnormality, there is limited data about the predictability of stress MBF in HD patients without MP abnormality.

Methods

A total 438 of HD patients who undergone 13NH3PET for suspected IHD were enrolled. All patients were undergone13NH3PET at Nagoya Radiological Diagnosis Foundation. After we excluded patients whose summed stress score (SSS) &lt;4, we identified 182 eligible patients. Patients were divided into two group according to the median value of CFR levels; low stress MBF group (≤2.56) and high stress MBF group (&gt;2.56). We followed up them up to 4.2 years (median 2.4 years) and collected their data. We evaluated their major adverse cardiac cerebrovascular event. We performed Kaplan-Meyer analysis and multivariable cox regression models. Furthermore, we evaluated the incremental value with C-index, net reclassification improvement (NRI) and integrated discrimination improvement (IDI) when CFR added into a model with established risk factors.

Result

There were intergroup difference in baseline characteristics: age, gender, prior CVD and diabetes. Kaplan-Meyer analysis shows statistically intergroup difference [log rank p=0.013, hazard ratio (HR) 0.413, 95% confidential interval (CI) 0.220–0.775]. Multivariable cox regression model for MACCE shows CFR is an independent risk factor (p=0.004, HR 0.311, 95% CI 0.137–0.684). As regarding model discrimination, all of C-index (0.832 vs 0.796, p=0.15), NRI (0.513, p=0.008) and IDI (0.032, p=0.033) were greatest in a predicting model with established risk factors plus stress MBF.

Conclusion

The low stress MBF group has poor prognosis in MACCE comparing to the high stress MBF group. Stress MBF is an independent risk factor for MACCE. Adding stress MBF on conventional risk factors could more accurately predict MACCE in HD patients, even in patients without MP abnormality.

Funding Acknowledgement

Type of funding source: None

Diabetes and hemodialysis are important factor for decrease coronary flow reserve even in the patients with normal myocardial perfusion

Background

In clinical setting, patients with traditional coronary risk factors are at high risk for coronary artery disease (CAD). Such patients who complain chest discomfort are usually performed nuclear myocardial perfusion (MP) test. We sometimes find patients whose PET result shows normal MP and abnormal coronary flow reserve (CFR). However, there are limited data about the predictors for decreased CFR. In the view of describe above, we have investigated the parameters for decreased CFR in the patients without MP abnormality.

Methods and results

From 20th April 2013 to 21st December 2018, we performed 2,930 13N- ammonia PET for suspected CAD. After excluding the follows; 966 patients with repeated test, 54 patients with incomplete data, one patient missed, we investigated 1,909 eligible patients' data. We performed least square to identify the factors decreased CFR. Hemodialysis (HD), age, prior revascularization, diabetes (DM) and body mass index (BMI) were independent risk factor for decreased CFR in all population. On the other hand, HD, age, DM, hypertension and BMI were independent risk factor for decreasing CFR in patients without MP abnormality. According to the result of least square methods, we classified all patients into four groups; without DM/ without HD group, with DM/ without HD group, without DM/ with HD group and with DM/ with HD group. The value of CFR in each group were as follows: without DM/ without HD group (median, 1st quartile-3rd quartile; 2.88, 2.21–3.52), with DM/ without HD group (2.65, 2.00–3.38), without DM/ with HD group (2.29, 1.67–2.95) and with DM/ with HD group (1.97, 1.43–2.68). There were statistically significant intergroup differences. The value of CFR in the patients without MP abnormality were as follows: without DM/ without HD group (3.04, 2.47–3.65), with DM/ without HD group (2.98, 2.40–3.61), without DM/ with HD group (2.52, 2.10–3.08) and with DM/ with HD group (2.38, 1.86–2.97). Even in the patients without MP abnormality, there were also statistically significant intergroup differences.

Conclusion

According to our 13N-ammonia PET data analysis, DM and HD were important and independent factors for decreased CFR. Even in the patients without MP abnormality, DM and HD were important factor for decreased CFR.

Funding Acknowledgement

Type of funding source: None

Automatic Labeled Dialogue Generation for Nursing Record Systems

The integration of digital voice assistants in nursing residences is becoming increasingly important to facilitate nursing productivity with documentation. A key idea behind this system is training natural language understanding (NLU) modules that enable the machine to classify the purpose of the user utterance (intent) and extract pieces of valuable information present in the utterance (entity). One of the main obstacles when creating robust NLU is the lack of sufficient labeled data, which generally relies on human labeling. This process is cost-intensive and time-consuming, particularly in the high-level nursing care domain, which requires abstract knowledge. In this paper, we propose an automatic dialogue labeling framework of NLU tasks, specifically for nursing record systems. First, we apply data augmentation techniques to create a collection of variant sample utterances. The individual evaluation result strongly shows a stratification rate, with regard to both fluency and accuracy in utterances. We also investigate the possibility of applying deep generative models for our augmented dataset. The preliminary character-based model based on long short-term memory (LSTM) obtains an accuracy of 90% and generates various reasonable texts with BLEU scores of 0.76. Secondly, we introduce an idea for intent and entity labeling by using feature embeddings and semantic similarity-based clustering. We also empirically evaluate different embedding methods for learning good representations that are most suitable to use with our data and clustering tasks. Experimental results show that fastText embeddings produce strong performances both for intent labeling and on entity labeling, which achieves an accuracy level of 0.79 and 0.78 f1-scores and 0.67 and 0.61 silhouette scores, respectively.

Bounds on Lorentz Invariance Violation from MAGIC Observation of GRB 190114C

On January 14, 2019, the Major Atmospheric Gamma Imaging Cherenkov telescopes detected GRB 190114C above 0.2 TeV, recording the most energetic photons ever observed from a gamma-ray burst. We use this unique observation to probe an energy dependence of the speed of light in vacuo for photons as predicted by several quantum gravity models. Based on a set of assumptions on the possible intrinsic spectral and temporal evolution, we obtain competitive lower limits on the quadratic leading order of speed of light modification.

Anti-allergic effect of the Src family kinase inhibitor saracatinib

The aim of this study was to evaluate the anti-anaphylactic and anti-allergic potentials of saracatinib, a Src family kinase inhibitor that was already shown to be safe in clinical trials when it was used as an anti-cancer drug. Using in vitro mast cell models, we found that saracatinib inhibited the degranulation response and cytokine production in RBL2H3 cells that were stimulated with IgE and antigen without affecting cell viability. Phosphorylation of Lyn, Akt, a PI3K substrate, and MAPKs including ERK, JNK, and p38, as well as the intracellular Ca²⁺ increase induced by this stimulation were also suppressed by saracatinib. This drug also inhibited symptoms in our established anaphylaxis mouse model, anaphylaxis-dependent spotted distribution of immune complex in skin (ASDIS). The intravenous injection of the mixture of IgE and antigen induced acute spotted distribution of immune complex in skin in hairless HR-1 mice, and its inhibition by intradermal injection of saracatinib was observed. Moreover, toluidine blue-stained skin sections indicated that the degranulation ratio of dermal mast cells was reduced in saracatinib-treated skin compared with vehicle-treated skin. Because only a few signaling inhibitors are used as anti-anaphylaxis and anti-allergic drugs, these results indicated the valuable suggestion that saracatinib and the Src family kinase inhibitors are good candidates for anti-anaphylaxis and anti-allergic drugs.

Wearable Sensor-Based Gait Analysis for Age and Gender Estimation

Wearable sensor-based systems and devices have been expanded in different application domains, especially in the healthcare arena. Automatic age and gender estimation has several important applications. Gait has been demonstrated as a profound motion cue for various applications. A gait-based age and gender estimation challenge was launched in the 12th IAPR International Conference on Biometrics (ICB), 2019. In this competition, 18 teams initially registered from 14 countries. The goal of this challenge was to find some smart approaches to deal with age and gender estimation from sensor-based gait data. For this purpose, we employed a large wearable sensor-based gait dataset, which has 745 subjects (357 females and 388 males), from 2 to 78 years old in the training dataset; and 58 subjects (19 females and 39 males) in the test dataset. It has several walking patterns. The gait data sequences were collected from three IMUZ sensors, which were placed on waist-belt or at the top of a backpack. There were 67 solutions from ten teams-for age and gender estimation. This paper extensively analyzes the methods and achieved-results from various approaches. Based on analysis, we found that deep learning-based solutions lead the competitions compared with conventional handcrafted methods. We found that the best result achieved 24.23% prediction error for gender estimation, and 5.39 mean absolute error for age estimation by employing angle embedded gait dynamic image and temporal convolution network.

Characterizing Word Embeddings for Zero-Shot Sensor-Based Human Activity Recognition

In this paper, we address Zero-shot learning for sensor activity recognition using word embeddings. The goal of Zero-shot learning is to estimate an unknown activity class (i.e., an activity that does not exist in a given training dataset) by learning to recognize components of activities expressed in semantic vectors. The existing zero-shot methods use mainly 2 kinds of representation as semantic vectors, attribute vector and embedding word vector. However, few zero-shot activity recognition methods based on embedding vector have been studied; especially for sensor-based activity recognition, no such studies exist, to the best of our knowledge. In this paper, we compare and thoroughly evaluate the Zero-shot method with different semantic vectors: (1) attribute vector, (2) embedding vector, and (3) expanded embedding vector and analyze their correlation to performance. Our results indicate that the performance of the three spaces is similar but the use of word embedding leads to a more efficient method, since this type of semantic vector can be generated automatically. Moreover, our suggested method achieved higher accuracy than attribute-vector methods, in cases when there exist similar information in both the given sensor data and in the semantic vector; the results of this study help select suitable classes and sensor data to build a training dataset.

P5410Combination assessment of renal and hepatic dysfunction improves the predictability of prognosis in patients with acute decompensated heart failure

Background

Renal dysfunction is associated with poor mortality in patients with heart failure (HF). Hepatic dysfunction, assessed by Fibrosis-4 (FIB4) index, has also prediction ability in acute decompensated HF (ADHF) patients. We investigated whether the assessment of the combination of FIB4 index and renal dysfunction improves predictability in patients with ADHF.

Methods

We retrospectively enrolled consecutive 758 patients who admitted due to ADHF from January 2011 to February 2018 and followed up for one year. FIB4 index on admission was calculated by the formula: age (yrs) × AST[U/L] / (platelets [103/μL] × (ALT[U/L])1/2). Study subjects were divided into high FIB4 index (>3.25) and low FIB4 index (≤3.25), furthermore each group were classified by the presence/absence of CKD (estimated glomerular filtration rate <60 ml/min/1.73m). We have generated four groups; low FIB4/without CKD (n=154), low FIB4/with CKD (n=294), high FIB4/without CKD (n=56), and high FIB4/with CKD (n=254). The primary outcome was defined as all-cause mortality in one year. We performed Kaplan-Meyer analysis and multivariable Cox regression models. Furthermore, we evaluated the incremental value with C-index, net reclassification improvement (NRI) and integrated discrimination improvement (IDI) when FIB4 index and renal dysfunction added to a baseline model.

Results

In total, 106 patients died in one year. High FIB4 index and CKD showed significantly higher 1-year mortality (high FIB4 index: 19.7% vs 10.3%, p<0.001, CKD: 17.0% vs 6.7%, p<0.001, respectively). Kaplan-Meyer analysis shows that high FIB4 index with CKD showed statistically higher mortality than the others (vs low FIB4/without CKD, p<0.001, vs high FIB4/without CKD, p=0.031, vs low FIB4/with CKD, p<0.001, respectively).

Multivariate Cox regression model revealed that both high FIB4 index and CKD were an independent risk predictor of 1-year mortality (FIB4 index: p<0.001, HR 1.06, 95% CI 1.035–1.087, CKD: p=0.004, HR 1.834, 95% CI 1.213–2.773, respectively) in patients with ADHF.

A baseline model for prediction of 1-year mortality was determined by multivariable logistic regression including age, body mass index, systolic blood pressure, and serum albumin (C-index: 0.688). Adding high FIB4 index and CKD to the baseline model, all of C-index (0.738, p=0.04), NRI (0.122, p=0.067), and IDI (0.024, p=0.004) were improved.

Receiver operating characteristic curves

Conclusions

Combination assessment of renal and hepatic dysfunction could improve the predictability of prognosis in patients with ADHF.

Acknowledgement/Funding

None

P4551Prognostic utility of Palliative Prognostic Index for prediction of 30-day and 1-year outcome in patients with acute decompensated heart failure

Background

The prognosis of heart failure remains poor similar to the terminal cancer patients, although recent progress in medical treatment. Palliative Prognostic Index (PPI) is a widely used prognostic index for terminal cancer patients (PPI includes: Palliative Performance Scale, oral intake, oedema, dyspnea at rest and delirium), suggesting the short-term prognostic marker of terminal cancer patients.

Purpose

The purpose of this study was to evaluate the impact of PPI on 30-day mortality, 1-year mortality and 1-year events (including all-cause mortality, readmission due to heart failure and new onset of cerebral infarction after hospital discharge) among acute decompensated heart failure (ADHF) patients.

Method

Study subjects comprised of consecutive 764 patients who admitted due to ADHF and followed up for 1-year. PPI were calculated at the time of hospital admission. Study subjects were divided into two groups based on the PPI: L-PPI (PPI<6) and H-PPI (6≤PPI). We calculated the C-index, net reclassification improvement (NRI) and the integrated discrimination improvement (IDI) to evaluate the improvement of prediction ability on 30-day mortality.

Result

H-PPI showed significantly higher 30-day mortality than L-PPI [7.9% vs 2.0%, log rank p<0.001, Hazard retio (HR): 1.26, 95% confidential interval(CI): 1.14–1.37, p<0.001], 1-year mortality [20.0% vs 12.7%, log rank p=0.022, HR 1.15, 95% CI 1.09–1.21, p<0.001]and 1-year events [45.5% vs 31.1%, log rank p<0.001, HR 1.13, 95% CI 1.09–1.17, p<0.001]. Multivariate cox proportional hazard models adjusted with several covariates revealed that PPI was an independent predictor of 30-day mortality (HR: 1.23, 95% CI: 1.10–1.36, p<0.001), 1-year mortality (HR: 1.10, 95% CI: 1.04–1.16, p<0.001) and 1-year events (HR: 1.11, 95% CI: 1.07–1.15, p<0.001), respectively.

A reference model for prediction of 30-day mortality was determined including left ventricular ejection fraction and serum albumin concentration by multivariable logistic regression analysis. (P<0.05) (C-index: 0.720) Adding PPI to the reference model (C-index: 0.773) significantly improved both NRI (0.458, p=0.038) and IDI (0.046, p=0.007), respectively.

Conclusion

We suggest that assessment of PPI showed good prognostic ability for 30-day and 1-year outcome, while PPI provided additional prognostic information in patients with ADHF.

P6240The impact of stress myocardial blood flow as a very strong predictor for all-cause mortality, cardiovascular mortality and adverse cardiac and cerebrovascular event in hemodialysis population

Background

In the clinical setting, ischemic heart disease (IHD) is a major problem not only in general patients but also in regular hemodialysis (HD) patients. 13ammonia positron emission tomography (13NH3PET) is an established and excellent diagnostic device for IHD. Although coronary flow reserve is the most important index in IHD diagnosis, there are limited data about stress myocardial blood flow (MBF). We investigated the prognosis predictability of stress MBF in all-cause mortality, cardiovascular (CV) mortality and adverse cardiac and cerebrovascular event (MACCE).

Methods and results

A total 438 of HD patients who undergone 13NH3PET for suspected IHD were enrolled. 29 cases were excluded due to revascularization therapy in 60 days. In total we collected 409 eligible cases. All patients were undergone13NH3PET at Nagoya Radiological Diagnosis Foundation. Patients were divided into two group according to the median value of stress MBF levels; low stress MBF group (<2.12) and high stress MBF group (≥2.12). We followed up them up to 4.2 years (median 2.4 years) and collected their data. We evaluated their all-cause mortality, CV mortality and MACCE. Kaplan-Meyer analysis shows that intergroup difference in all-cause mortality (log rank p=0.001, hazard ratio [HR] 0.411, 95% confident interval [CI] 0.261–0.632), CV mortality (log rank p=0.002, HR 0.324, 95% CI 0.157–0.625) and MACCE (log rank p<0.001, HR 0.465, 95% CI 0.324–0.657). Multiple cox analysis that include established risk factors shows CFR is an independent risk factor for all-cause mortality (HR 0.261, 95% CI 0.154–0.442), CV mortality (HR 0.172, 95% CI 0.079–0.374) and MACCE (HR 0.329, 95% CI 0.213–0.503). As a result of the incremental value with C-index, net reclassification improvement (NRI) and integrated discrimination improvement (IDI) when CFR added into a model with established risk factors, each indicator shows adding stress MBF on established risk factors improve the predictability in all-cause mortality, CV mortality and MACCE (all-cause mortality; NRI 0.642, p<0.001, IDI 0.091, p<0.001, CV mortality NRI 0.809, p<0.001, IDI 0.116, p<0.001, MACCE; NRI 0.646, p<0.001, IDI 0.072, p<0.001).

Conclusion

Considering prognosis of HD population, stress MBF is an important and independent predictor for all-cause mortality, CV mortality and MACCE. As a result of our investigation, stress MBF is one of most strong predictors in HD population.

P3764Morphological mechanisms of atrial functional mitral regurgitation in patients with atrial fibrillation

Background

Atrial functional mitral regurgitation (AFMR) occurs in patients with atrial fibrillation. However, morphological mechanisms of AFMR are poorly understood.

Purpose

The purpose of this study was to examine the morphological characteristics in patients with AFMR.

Methods

Among consecutive 795 patients undergoing initial radiofrequency catheter ablation (RFCA) at our hospital, twenty-five patients with persistent AF accompanied by AFMR (≥ moderate) before RFCA (AFMR group) were studied. Age-matched 25 patients with persistent AF without MR were defined as a control group.

Results

Left ventricular ejection fraction (LVEF) was lower and left atrium volume index was larger in the AFMR group (Table). Mitral valve annulus diameter and length of anterior mitral leaflet (AML) were similar between groups, whereas length of posterior mitral leaflet (PML) was significantly shorter in the AFMR group. Smaller tethering angle of AML (γ in the figure) and shorter tethering height were significantly associated with the occurrence of AFMR, which were different from morphology of functional mitral regurgitation in patients with dilated LV. Multiple regression analysis revealed that less tenting height (p<0.05) and LA dilatation toward the posterior (p<0.01) were significantly related to AFMR.

Echocardiographic parameters AFMR (n=25) Control (n=25) P value Age, y 69±8 66±10 NS Male, n (%) 9 (36) 20 (80) P=0.001 LVEF,% 60±9 67±6 P=0.004 LAD, mm 44±5 41±7 NS LAVI, ml/m2 56±17 41±13 P<0.001 MV diameter, mm 3.9±0.4 3.8±0.5 NS α angle, ° 34±9 35±7 NS β angle, ° 48±9 50±8 NS γ angle, ° 32±5 37±5 P=0.0005 AML length, mm 3.0±0.5 3.0±0.5 NS PML length, mm 2.1±0.1 2.4±0.1 P=0.03 Tenting height, mm 1.5±0.1 1.8±0.1 P=0.02 D, mm 0.8±0.3 0.5±0.3 P=0.001 LVEF: left ventricular ejection fraction; LAD: left atrial diameter; LAVI: left atrial volume index; AML: anterior mitral leaflet; PML: posterior mitral leaftlet.

Conclusions

AFMR occurs in patients with unique morphological features, such as less tethering height and LA dilatation toward the posterior.

REDUCTION IN AMBIENT GAMMA DOSE RATE FROM RADIOCESIUM DUE TO SNOW COVER

In the present study, variations in ambient gamma dose rate associated with snow cover were examined in a radioactive-contaminated site in Fukushima Prefecture, Japan. The ambient gamma dose rates decreased with increasing snow depth. The reduction trends were different between fresh snow (0.1-0.2 g/cm3) and granular snow (0.3-0.4 g/cm3) depending on snow density. Snow cover water content (snow water equivalent) calculated from snow depth and density was a key parameter governing the reduction in the ambient gamma dose rate. The ambient gamma dose rates reduced to 0.6 and 0.5 at 4 g/cm2 and 8 g/cm2 of snow water equivalent, respectively. Based on gamma-ray flux density distributions, the ambient gamma dose rates from the primary gamma rays decreased more compared to those from scattered gamma rays due to snow cover.

P1025Clinical predictors for the improvement of left ventricular ejection fraction and prognosis after catheter ablation of atrial fibrillation in patients with systolic dysfunction

Background

Catheter ablation (CA) of atrial fibrillation (AF) improves left ventricular ejection fraction (LVEF) and clinical outcomes in patients with left ventricular systolic dysfunction (LVSD). However, predictors of the improvement of LV function and clinical outcomes by CA were poorly understood.

Purpose

We examined the efficacy of CA in AF patients with LVSD and predictive factors associated with clinical outcomes.

Method

Among consecutive 795 patients undergone initial RFCA at our hospital, we studied 51 patients with LVSD (LVEF ≤50%). Improved LVEF more then 5% at 1-year after CA was classified as “responder” to CA. We analyzed clinical variables and echocardiographic parameters before and after the CAs.

Results

In the responder group, LVEF was significantly improved 1-year after catheter ablation compared with the non-responder group. (ΔLVEF 22±12% vs. −1±4%, p<0.001). The responder group was significantly younger, had more non-paroxysmal AF, smaller LV systolic diameter and lower plasma BNP level before CA (Table). Late gadolinium enhancement (LGE)-positive rate in cardiovascular magnetic resonance imaging (CMR) before CA was higher in the non-responder group than in the responder group (100% [6/6] vs. 38% [5/13], p<0.005). After CAs of AF, event-free survival from hospitalization for heart failure was significantly higher in the responder group (Figure) with less AF recurrence (27% vs. 47%, p=0.04) than in the non-responder group.

Baseline characteristics Responder (N=35) Non-Responder (N=16) P value Age, y 62±11 69±8 p<0.01 Male, n (%) 26 (74) 13 (76) NS Non-pAF 26 (74) 4 (24) p<0.01 LAD, mm 48±7 48±8 NS LAVI, ml/m2 54±17 58±20 NS LVDd, mm 54±7 58±10 NS LVDs, mm 43±7 48±10 p=0.05 EF, % 37±8 38±8 NS BNP (pg/ml) 278±225 684±848 p<0.05

Conclusion

Younger age, absence of LV dilatation, lower plasma BNP, or absence of LGE may well predict favorable clinical outcomes after CA in patients with LVSD.

On-Device Deep Learning Inference for Efficient Activity Data Collection

Labeling activity data is a central part of the design and evaluation of human activity recognition systems. The performance of the systems greatly depends on the quantity and "quality" of annotations; therefore, it is inevitable to rely on users and to keep them motivated to provide activity labels. While mobile and embedded devices are increasingly using deep learning models to infer user context, we propose to exploit on-device deep learning inference using a long short-term memory (LSTM)-based method to alleviate the labeling effort and ground truth data collection in activity recognition systems using smartphone sensors. The novel idea behind this is that estimated activities are used as feedback for motivating users to collect accurate activity labels. To enable us to perform evaluations, we conduct the experiments with two conditional methods. We compare the proposed method showing estimated activities using on-device deep learning inference with the traditional method showing sentences without estimated activities through smartphone notifications. By evaluating with the dataset gathered, the results show our proposed method has improvements in both data quality (i.e., the performance of a classification model) and data quantity (i.e., the number of data collected) that reflect our method could improve activity data collection, which can enhance human activity recognition systems. We discuss the results, limitations, challenges, and implications for on-device deep learning inference that support activity data collection. Also, we publish the preliminary dataset collected to the research community for activity recognition.

Impact of a systematic education model for palliative care in cancer

In clinical practice, pharmacists are continually required to improve their knowledge and expertise; however, the postgraduate education system for professional development cannot be confidently stated to be well established. The establishment of a systematic and multifaceted educational curriculum should be useful to improve home care and pharmacists' contribution; therefore, we developed a curriculum in collaboration with the university faculty of pharmaceutical sciences, department of pharmacy in hospital, and the Fukuoka Pharmaceutical Association. Class topics were extracted from the "Kanwa-Iryou-Yakugaku", edited by the Japanese Society for Pharmaceutical Palliative Care and Sciences. The items are necessary to perform palliative care as a pharmacist. A class schedule of 6 days (24 classes in total) was formulated. Questionnaires on comprehension degree before and after each class were provided to the participants. Comprehension was assessed on a scale of 1 to 10, where "I do not understand at all" was 1 and "I understand enough" was 10. The average recovery rates of questionnaires from each class were 92.6 % and 88.9 % before and after class, respectively. The average number of participants who completely answered the questionnaire before and after class was 45.6; therefore, these data were analyzed. Comprehension degree on each topic had significantly increased after attendance of all classes (p < 0.01). The comprehension degree of participants of the medical science of palliative care did greatly improve. Consequently, it is clear that the standard education model constructed was meaningful for the professional development of pharmacists in palliative care medicine.

Cost utility analysis of pharmacist counseling care for breast cancer chemotherapy outpatients

Chemotherapy for cancer is increasingly implemented in the outpatient setting. Pharmacists contribute to cancer treatment by conducting counseling during outpatient chemotherapy visits. They provide guidance on drug treatment, side effects, and side effect countermeasures on every visit. However, there have been few economic evaluations of pharmacist involvement in outpatient chemotherapy. Therefore, we performed a cost utility analysis. We assigned usual care (control) and pharmacist counseling to two groups of 19 patients receiving outpatient chemotherapy for breast cancer at Gifu Municipal hospital. Quality of life was measured at three timepoints before and during chemotherapy treatment using the EuroQol 5 dimension instrument (EQ-5D). EQ-5D values across the timepoints were 0.831, 0.757, and 0.791 for the control group, and 0.882, 0.883, and 0.921 for the pharmacist counseling group. The additional cost in the pharmacist counseling group was 2,227 yen per counseling session. The change in quality-adjusted life years (QALY) was a maximum of -0.021±0.186 in the control group and 0.007±0.199 in the pharmacist counseling group. The maximum cost for one QALY was 1,360,558 yen (≈12,460 US dollars). Pharmacists' counseling in outpatient cancer chemotherapy for breast cancer patients had an acceptable incremental cost-effect ratio, contributing to improved patient quality of life without significant additional expenditure to healthcare.

Performance verification of a precise vibrating-wire magnet alignment technique for next-generation light sources

The high-accuracy alignment of magnets is a key issue in the development of next-generation light-source rings. To obtain adequate dynamic apertures, the magnets must be aligned to an accuracy of 10 µm or better. Recently, a new technique that utilizes a vibrating wire has attracted attention for this purpose as it can directly determine with high resolution the magnetic centers in a series of multipole magnets on a straight section between bending magnets. In conventional vibrating-wire alignment techniques, wire sag, which causes alignment errors, is determined from the theoretical catenary curve. By contrast, in the present study, we have measured the sag profiles of various wires in the longitudinal direction to micrometer-order accuracy. We concluded that we can reduce deviations of the actual wire sag from the theoretical curve by choosing a suitable wire. By setting up a test bench of a vibrating-wire alignment system for a series of multipole magnet on a straight section, we have achieved the total error of the magnetic-center measurements of micrometer-order in the standard deviation. Moreover, two systematic error factors, the drift of the magnetic centers due to thermal deformations of the magnets after they are excited and the change in the magnetic centers due to reassembly of the magnets after installing the vacuum chamber, are included in practical magnet alignments. We have experimentally investigated these error factors using the test bench.

Platelet count and mean platelet volume are associated with not only bone, soft tissue, and lymph node metastases but also with malignant pleural effusion in lung cancer patients

An increased platelet count is often observed in lung cancer patients. Whether and how the platelets affect cancer progression have yet to be established. The aim of the study was to investigate the involvement of the platelet count and mean platelet volume (MPV) in the prognosis and progression of lung cancer patients. This retrospective study included 146 patients with newly diagnosed primary lung cancer. The platelet count and MPV were measured before invasive diagnostic procedures and treatment. These platelet indices, overall survival of the patients, and tumor metastases for each organ were analyzed. On Kaplan-Meier survival analysis, the overall survivals of patients with platelet counts ≤ 244.0 × 109/L or MPV > 9.7 fL were longer than those of patients with platelet counts > 244.0 × 109/L or MPV ≤ 9.7 fL. Cox regression analysis showed that poor performance status, increased platelet count, and increased C-reactive protein were independent prognostic factors. The platelet indices were associated with metastases to bone, soft tissue, and lymph node, in addition to malignant pleural effusion. Increased platelet count and decreased MPV were unfavorable prognostic factors for patients with lung cancer, and they were involved in bone, soft tissue, and lymph node metastases and malignant pleural effusion.

Postoperative intensive care unit admission does not affect outcomes in elective surgical patients with severe comorbidity

OBJECTIVE

The impact of postoperative intensive care upon patient outcomes was evaluated by retrospectively investigating the rate of poor outcomes among miscellaneous elective surgical patients with severe comorbidities.

DESIGN

A retrospective cohort study was carried out.

SETTING

University hospital.

PATIENTS

Surgical patients with severe comorbidities.

INTERVENTION

The outcomes of 1218 surgical patients treated in intensive care units (ICUs) and postsurgical wards (ICU group vs. non-ICU group) were reviewed for poor outcomes (i.e., no discharge or death). A propensity score analysis was used to generate 248 matched pairs of ICU-admitted patients and controls.

VARIABLES OF INTEREST

Poor outcome rates on postoperative day 90 and mortality on postoperative days 30 and 90.

RESULTS

No significant between-group differences were observed in terms of poor outcomes on postoperative day 90 [ICU vs. non-ICU: 33/248 (13%) vs. 28/248 (11%), respectively; ICU odds ratio (OR): 1.19, 95% confidence interval (CI), 0.71-2.01, p=0.596] or in between-group differences in terms of mortality on postoperative days 30 and 90 [ICU vs. non-ICU: 4/248 (1.6%) vs. 2/248 (0.8%) on postoperative day 30 and 5/248 (2.0%) vs. 3/248 (1.2%) on day 90, respectively; ICU OR (95% CI), 2.00 (0.37-10.9) and 1.67 (0.40-6.97) for postoperative 30- and 90-day mortality, respectively (p=0.683 and 0.724)]. Low preoperative body weight was negatively correlated to patient outcomes [OR (95% CI): 0.82/10kg (0.70-0.97), p=0.019], whereas regional analgesia combined with general anesthesia was positively correlated to patient outcomes [OR (95% CI): 0.39 (0.69-0.96), p=0.006]. Extra ICU admission was correlated to poor patient outcomes [OR (95% CI): 4.18 (2.23-7.81), p < 0.0001].

CONCLUSIONS

Postoperative ICU admission failed to demonstrate any meaningful benefits in patients with severe comorbidities undergoing miscellaneous elective surgeries.

Amiodarone does not affect brain injury in a rat model of transient forebrain ischemia

OBJECTIVES

Although amiodarone may cause neurotoxicity that can affect patient outcomes when used during cardiopulmonary resuscitation (CPR), it has been commonly prescribed during CPR. This study investigated the possible neurotoxic effects of amiodarone in a rat model of transient forebrain ischemia.

DESIGN

A prospective laboratory animal study was carried out.

SETTING

Animal laboratory.

MATERIALS

Male Sprague-Dawley rats.

INTERVENTION

Eight minutes of forebrain ischemia was induced in rats by bilateral carotid occlusion and hypotension (mean arterial pressure=35mmHg) under isoflurane (1.5%) anesthesia. Amiodarone (0, 50, 100 and 150mg/kg) with saline was injected intraperitoneally 10min after ischemia. Rats given 0mg/kg of amiodarone were used as saline-treated controls. Sham operated rats received no treatment.

VARIABLES OF INTEREST

Animals were evaluated neurologically on postoperative days 4-7, and histologically after a one-week recovery period.

RESULTS

The greatest improvement in water maze test performance corresponded to the sham operated group (p=0.015 vs. saline-treated controls). No differences in performance were seen in amiodarone-treated rats compared with saline-treated controls. In the control group, 45% of the CA1 hippocampal neurons survived, compared with 78% in the sham operated group (p=0.009). Neuron survival after ischemia in the amiodarone treatment groups (50, 100 and 150mg/kg) (58%, 40% and 36%, respectively) and in the control rats did not differ significantly.

CONCLUSIONS

The administration of amiodarone immediately after transient forebrain ischemia did not worsen spatial cognitive function or neuronal survival in the hippocampal CA1 region in rats. The current results must be applied with caution in humans. However, they indicate that the potential neurotoxicity induced by amiodarone during resuscitation after cardiac arrest may be negligible.