Nearest-neighbor Approach Research Articles

Industrial designs and firm performance: evidence from publicly traded Canadian companies, 1990–2014

ABSTRACT Industrial designs (IDs) are a specialized type of intellectual property that protects a product's form and presentational features, giving it uniqueness that differentiates it from other products. This paper estimates the effect on firm revenue and profitability of holding IDs. Using a unique data set linking Canadian ID holdings with Canadian publicly traded firms over the years 1990-2014, the authors use two methods to identify a positive association between holding IDs and firm revenue per employee and net income per employee. We contribute to the literature on IDs by introducing controls for patenting, R&D spending, and time-varying sector effects. To determine the marginal effect of each additional ID held, we use a fixed effects regression and find that a 1% increase in the stock of IDs increases revenue per employee by 0.19%. With a nearest-neighbor matching approach, we find a 10% total premium in revenue per employee and a 20% premium in net income per employee among firms holding at least one ID, compared to those with none.

Large Margin Weighted k-Nearest Neighbors Label Distribution Learning for Classification.

Label distribution learning (LDL) helps solve label ambiguity and has found wide applications. However, it may suffer from the challenge of objective inconsistency when adopted to classification problems because the learning objective of LDL is inconsistent with that of classification. Some LDL algorithms have been proposed to solve this issue, but they presume that label distribution can be represented by the maximum entropy model, which may not hold in many real-world problems. In this article, we design two novel LDL methods based on the k -nearest neighbors ( k NNs) approach without assuming any form of label distribution. First, we propose the large margin weighted k NN LDL (LW- k NNLDL). It learns a weight vector for the k NN algorithm to learn label distribution and implement a large margin to address the objective inconsistency. Second, we put forward the large margin distance-weighted k NN LDL (LD k NN-LDL) that learns distance-dependent weight vectors to consider the difference in the neighborhoods of different instances. Theoretical results show that our methods can learn any general-form label distribution. Moreover, extensive experimental studies validate that our methods significantly outperform state-of-the-art LDL approaches.

HIDIM: A novel framework of network intrusion detection for hierarchical dependency and class imbalance

Deep learning-based network intrusion detection has been extensively explored as a data-driven approach. Therefore, paying attention to the data’s characteristics is essential. By analyzing the attribute dependence and sample distribution of intrusion data, there are the following problems: “hierarchical dependency omission” and “decision boundary discontinuity.” The former means the previous attribute embedding models failed to incorporate network protocol hierarchy. The latter indicates that the small disjuncts distribution leads to sub-concept fragmentation, exacerbating the difficulty in handling class imbalance. To address these problems, we propose a novel detection framework for Hierarchical Dependency and Class Imbalance (HIDIM). First, we treat semantic attributes as words and introduce the protocol hierarchy of attributes into a paragraph embedding model. Second, we design a synthetic oversampling method. It adopts a mutual nearest neighbor approach to determine the boundaries of each disjunct. Then, it synthesizes high-quality samples within those boundary areas by crossing or mutating features based on their importance. The experimental results on multiple real-world datasets demonstrate that the proposed framework is superior to other state-of-the-art models in terms of accuracy, F1-score, and false negative rate by 2.23%, 2.12%, and 1.43% on average, respectively.

Rainfall characteristics over the Congo Air Boundary Region in southern Africa: A comparison of station and gridded rainfall products

Strong meridional rainfall gradients exist between the tropics and subtropics in southwestern Africa, bounded to the north by the moist Congo basin and to the south by the Kalahari Desert. This domain received relatively little scientific attention compared to the rest of southern Africa. In this study, the limited available station data are assessed against six gridded rainfall products (CHIRPS, PERSIANN-CDR, ERA5, GPCC and CPC) for various rainfall characteristics. The nearest neighbour approach was used to match the closest rainfall dataset pixel to each station location, with the assumption that each rain gauge represents observation of various pixels of products, irrespective of product resolution. Results reveal that ERA5, CHIRPS and PERSIANN-CDR tend to represent the monthly rainfall totals and number of rainy days well for most stations although magnitudes and monthly peaks differ. CPC and GPCC tend to perform poorly for magnitudes of rainfall, rainy days and monthly cycles especially for Angolan stations. These products also fail to adequately capture spatial distributions of rainfall, with poor representation of the strong gradients found in the region.Correlations between various gridded rainfall products mostly show good agreement in rainfall totals and rainy days. For early summer (October–November-December) moderate rainy days, ERA5 and CHIRPS products tend to have more days than the stations while CPC and GPCC products perform poorly over Angola and in the south. ERA5 generally overestimates rainfall in mountainous regions, while other products tend to underestimate it. Based on the Simple Daily Intensity Index, it was found that for most of the gridded rainfall products tend to overestimate rainfall during rainy days in the northern and wetter part of the domain. Furthermore, for heavy rainfall, CPC and GPCC tend to compare fairly well with stations in the southern and eastern parts of the domain but poorly with those in the western parts. PERSIANN-CDR tends to underestimate heavy rainy days for most stations in early and late summer (January–February-March-April). However, CHIRPS compares well at several stations, while ERA5 performs well for stations located in the south. This study helps provide useful guidance in choosing suitable rainfall gridded datasets for assessing long term rainfall cycles, daily rainfall characteristics as well as extremes over southwestern Africa.

Hindcasting and updating Landsat-based forest structure mapping across years to support forest management and planning

Forest vegetation mapping that integrates forest inventory data with multispectral remote sensing data provides valuable geospatial products for public land management agencies, but resource managers may require rapid updating of maps as new imagery becomes available (updating) or retrospective mapping for times prior to forest inventory plot measurement (hindcasting). While forest attribute mapping using Landsat multispectral imagery is common, the accuracy of applying models outside of reference epoch to support long-term forest monitoring is not normally quantified. We examine whether a Landsat-based mapping approach can support robust, temporally consistent multivariate mapping of forest structure and composition data in support of forest management planning and landscape analysis. Specifically, we ask: how accurate forest attribute mapping was when hindcasting or updating outside of a period of time when forest inventory plot data were available (reference epoch)? In the western Cascade Mountains of Oregon and California, USA, we used the gradient nearest neighbor approach to annually impute USDA Forest Inventory and Analysis (FIA) plot data (2001–2016) to all 30-m forested pixels based on temporally smoothed Landsat multispectral imagery (1986–2021), including basal area, canopy cover, quadratic mean diameter of dominant trees, stand height, and the density of large diameter trees. We made extrapolations from models fit to a 10-year reference epoch to both earlier periods (2001–2006 hindcast) and to later period (2011–2016 update) and quantified prediction accuracies relative to models based on the full data (2001–2016). To evaluate the influence of spatial scale on hindcasting and updating, we compared full and extrapolation model predictions at pixel-level (0.09 ha) and hexagon-level (780 ha).At the plot-level, we found no strong differences between the full and extrapolation model predictions for R2 and mean error nor among predicted vs. observed regression coefficients. At the pixel-level, average differences due to hindcasting and updating were near zero, though differences varied up to 20 % across pixels. At the hexagon-level, the range in map differences was small (+/- 5 %), but hindcasting resulted in lesser forest attribute predictions. We observed greater variability in pixel-level and hexagon-level prediction differences when hindcasting or updating was temporally further away from the reference period. Using 2001 hindcast and 2016 updated maps as a case study, we found that with hindcasting and updating map differences were spatially aggregated across the study region. Our results support Landsat-based hindcasting and updating of forest attribute mapping beyond the time period covered by forest plot data. Our results suggest aggregating data to coarse spatial resolutions may minimize differences due to hindcasting and updating. Further research is needed to identify the key drivers for prediction differences to improve the accuracy of both hindcasting and updating as a basis for forest monitoring.

Sub-sampling graph neural networks for genomic prediction of quantitative phenotypes.

In genomics, use of deep learning (DL) is rapidly growing and DL has successfully demonstrated its ability to uncover complex relationships in large biological and biomedical data sets. With the development of high-throughput sequencing techniques, genomic markers can now be allocated to large sections of a genome. By analyzing allele sharing between individuals, one may calculate realized genomic relationships from single-nucleotide polymorphisms (SNPs) data rather than relying on known pedigree relationships under polygenic model. The traditional approaches in genome-wide prediction (GWP) of quantitative phenotypes utilize genomic relationships in fixed global covariance modeling, possibly with some nonlinear kernel mapping (for example Gaussian processes). On the other hand, the DL approaches proposed so far for GWP fail to take into account the non-Euclidean graph structure of relationships between individuals over several generations. In this paper, we propose one global convolutional neural network (GCN) and one local sub-sampling architecture (GCN-RS) that are specifically designed to perform regression analysis based on genomic relationship information. A GCN is tailored to non-Euclidean spaces and consists of several layers of graph convolutions. The GCN-RS architecture is designed to further improve the GCN's performance by sub-sampling the graph to reduce the dimensionality of the input data. Through these graph convolutional layers, the GCN maps input genomic markers to their quantitative phenotype values. The graphs are constructed using an iterative nearest neighbor approach. Comparisons show that the GCN-RS outperforms the popular Genomic Best Linear Unbiased Predictor method on one simulated and three real datasets from wheat, mice and pig with a predictive improvement of 4.4% to 49.4% in terms of test mean squared error. This indicates that GCN-RS is a promising tool for genomic predictions in plants and animals. Furthermore, GCN-RS is computationally efficient, making it a viable option for large-scale applications.

Scattering matrix similarity metric optimization for improved defect characterisation based on dynamic graph attention networks

Ultrasonic scattering matrices contain rich defect information and have great potential for characterising small crack-like defects. However, experimentally measured scattering matrices often exhibit some level of distortions compared to those of the idealised defects, posing challenges for accurate defect characterisation. In this paper, defect characterisation was performed by adopting a nearest neighbour approach based on a scattering matrix database of reference defects, and the test data were contaminated by coherent measurement noise of varying amplitudes. The performance of different similarity metrics on characterisation accuracy was studied, including the Euclidean similarity, cosine similarity, Pearson correlation coefficient, and the structural similarity index. Based on a comprehensive analysis of the strengths and weaknesses of different similarity metrics, we propose a defect characterisation framework by constructing similarity graphs and leveraging advanced graph neural networks. Within the proposed approach, multiple metrics were adopted to quantify the similarity between the scattering matrices of different defects, and an improved dynamic graph attention network was developed based on a customised neighbour sampling strategy to learn the optimal metric from the graph-structured data. Experimental results show that compared to the conventional approach which adopted a globally optimal similarity metric, the proposed method can reduce the root mean squared error for the length and angle predictions by 60.5% and 67.1%, respectively.

Energy-Aware 3D Path Planning by Autonomous Ground Vehicle in Wireless Sensor Networks

Wireless sensor networks are used to monitor the environment, to detect anomalies or any other problems and risks in the system. If used in the transportation network, they can monitor traffic and detect traffic risks. In wireless sensor networks, energy constraints must be handled to enable continuous environmental monitoring and surveillance data gathering and communication. Energy-aware path planning of autonomous ground vehicle charging for sensor nodes can solve energy and battery replacement problems. This paper uses the Nearest Neighbour algorithm for the energy-aware path planning problem with an autonomous ground vehicle. Path planning simulations show that the Nearest Neighbour algorithm converges faster and produces a better solution than the genetic algorithm. We offer robust and energy-efficient path planning algorithms to swiftly collect sensor data with less energy, allowing the monitoring system to respond faster to anomalies. Positioning communicating sensors closer minimizes their energy usage and improves the network lifetime. This study also considers the scenario in which it is recommended to avoid taking direct travelling pathways between particular node pairs for a variety of different reasons. To address this more challenging scenario, we provide an Obstacle-Avoided Nearest Neighbour-based approach that has been adapted from the Nearest Neighbour approach. Within the context of this technique, the direct paths that connect the nodes are restricted. Even in this case, the Obstacle-Avoided Nearest Neighbour-based approach achieves almost the same performance as the the Neighbour-based approach.

Toxicity reference values for force health protection: Provisional occupational exposure guidelines

Force Health Protection programs in the U.S. Air Force endeavor to sustain the operational readiness of the warfighters. We have previously identified hundreds of chemical substances of interest and toxicity reference value (TRV) knowledge gaps that constrain risk based-decision-making for potential exposures. Multiple approaches to occupational TRV estimation were used to generate possible guideline values for 84 compounds (18% of the substances of interest). These candidate TRVs included values from international databases, chemical similarity (nearest neighbor) approaches, empirical adjustments to account for duration differences, quantitative activity relationships, and thresholds of toxicological concern. This present work describes derivation of provisional TRVs from these candidate values. Rodent bioassay-derived long-term worker Derived No-Effect Levels (DNELs) were deemed presumptively the most reliable, but only 19 such DNELs were available for the 84 substances with TRV gaps. In the absence of DNELs, the quality of the approaches and consistency among candidate values were key elements of the weight of evidence used to select the most suitable guideline values. The use of novel nearest-neighbor approaches, empirical adjustment of short term TRVs, and occupational exposure bands were found to be options that would allow occupational TRV estimation with reasonable confidence for nearly all substances evaluated.

Toxicity reference values (TRVs) for force health protection: Gap identification and TRV prediction

The mission of the Force Health Protection (FHP) program of the U.S. Air Force (USAF), sustaining the readiness of warfighters, relies on determinations of acceptable levels of exposure to a wide array of substances that USAF personnel may encounter. In many cases, exposure details are limited or authoritative toxicity reference values (TRVs) are unavailable. To address some of the TRV gaps, we are integrating several approaches to generate health protective exposure guidelines. Descriptions are provided for identification of chemicals of interest for USAF FHP (467 to date), synthesis of multiple TRVs to derive Operational Exposure Limits (OpELs), and strategies for identifying and developing candidate values for provisional OpELs when authoritative TRVs are lacking. Rodent bioassay-derived long-term Derived No Effect Levels (DNELs) for workers were available only for a minority of the substances with occupational TRV gaps (19 of 84). Additional occupational TRV estimation approaches were found to be straightforward to implement: Tier 1 Occupational Exposure Bands, cheminformatics approaches (multiple linear regression and novel nearest-neighbor approaches), and empirical adjustment of short term TRVs. Risk assessors working in similar contexts may benefit from application of the resources referenced and developed in this work.

Spatial effects of infiltrating T cells on neighbouring cancer cells and prognosis in stage III CRC patients.

Colorectal cancer (CRC) is one of the most frequently occurring cancers, but prognostic biomarkers identifying patients at risk of recurrence are still lacking. In this study, we aimed to investigate in more detail the spatial relationship between intratumoural T cells, cancer cells, and cancer cell hallmarks as prognostic biomarkers in stage III colorectal cancer patients. We conducted multiplexed imaging of 56 protein markers at single-cell resolution on resected fixed tissue from stage III CRC patients who received adjuvant 5-fluorouracil (5FU)-based chemotherapy. Images underwent segmentation for tumour, stroma, and immune cells, and cancer cell 'state' protein marker expression was quantified at a cellular level. We developed a Python package for estimation of spatial proximity, nearest neighbour analysis focusing on cancer cell-T-cell interactions at single-cell level. In our discovery cohort (Memorial Sloan Kettering samples), we processed 462 core samples (total number of cells: 1,669,228) from 221 adjuvant 5FU-treated stage III patients. The validation cohort (Huntsville Clearview Cancer Center samples) consisted of 272 samples (total number of cells: 853,398) from 98 stage III CRC patients. While there were trends for an association between the percentage of cytotoxic T cells (across the whole cancer core), it did not reach significance (discovery cohort: p = 0.07; validation cohort: p = 0.19). We next utilised our region-based nearest neighbour approach to determine the spatial relationships between cytotoxic T cells, helper T cells, and cancer cell clusters. In both cohorts, we found that shorter distance between cytotoxic T cells, T helper cells, and cancer cells was significantly associated with increased disease-free survival. An unsupervised trained model that clustered patients based on the median distance between immune cells and cancer cells, as well as protein expression profiles, successfully classified patients into low-risk and high-risk groups (discovery cohort: p = 0.01; validation cohort: p = 0.003). © 2024 The Author(s). The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Navigating uncertainty: A dynamic Bayesian network-based risk assessment framework for maritime trade routes

Maritime safety is crucial for international seaborne trade and the global economy. Acknowledging the inevitable and multifaceted risks present in maritime navigation, we introduce a novel approach to evaluate the time-dependent risk performance of various routes, with the objective to mitigate vulnerabilities and fortify resilience. To achieve this goal, we have developed an effective framework using a Dynamic Bayesian network (DBN) model to capture the interactive relationships among the influential factors in probabilistic terms. This framework employs a combination of the synthetic minority over-sampling technique and edited nearest neighbor approach to address the issue of imbalanced maritime accident reports. Furthermore, the gradient descent algorithm is utilized to calculate the conditional probabilities among the influential factors, while the Markov chain model is applied to develop the DBN model. The framework enables dynamical consideration of the risk performance of the maritime trade routes and identifies the key influential factors during different time intervals. Our research scientifically identifies the devastating impact of terrorism on shipping routes, followed by other risk factors such as piracy, military conflicts, and various induced risks, ranked in descending order of severity. The case study on the Indian Ocean shipping route, which applies the proposed methodology, reveals an initial phase of fluctuating increases, succeeded by a prevailing downward trend over the span of 2009–2019. The methodology and results provide valuable support for maritime stakeholders in making informed decisions.

Thermal and visual comforts of occupants for a naturally ventilated educational building in low-income economies: A machine learning approach

Thermal and visual comfort (TVC) significantly impact the health and productivity of occupants as well as the indoor environmental quality in educational buildings. Machine learning approaches have currently gained popularity for TVC prediction; however, accurately predicting TVC faces challenges due to instrumental errors, environmental noise, and imbalanced data. Moreover, studies have revealed the unreliability of TVC predictions, particularly in naturally ventilated buildings. Therefore, this study develops three machine learning models, random forest (RF), decision tree (DT), and XGBoost, to accurately predict TVC using 1310 sets of field survey data collected from a naturally ventilated educational building (NVEB) in Dhaka, Bangladesh. Recursive feature elimination with cross-validation is employed to identify optimal features, while two resampling methods are applied to manage imbalanced classification. The RF model achieves the highest accuracy of 96 % for thermal comfort prediction, while the XGBoost model attains the highest accuracy of 95 % for visual comfort prediction. Meanwhile, the DT model achieves an accuracy of 94 % and 93 % for TVC prediction, respectively. Moreover, both RF and XGBoost outperform DT in class prediction performance, with an average area under the curve (AUC) scores above 0.95, while DT achieves an average AUC score above 0.80. In terms of two resampling methods, the results show that the models developed using the synthetic minority over-sampling technique (SMOTE) + edited nearest neighbour approach leads to less misclassification compared to the SMOTE + Tomek link. These findings demonstrate the capability of machine learning models in accurately predicting TVC in NVEBs and handling imbalanced datasets.

Quantifying the features of earthquake clusters in north-Central Iran, based on nearest-neighbor distances and network analysis

The statistical features of earthquake clusters in North-Central Iran (Tehran Region) are investigated, with the aim of quantitatively characterizing the properties of earthquake triggering and allow exploring their possible relations with the tectonic setting of the study area.The nearest-neighbor approach is used for the identification of the earthquake clusters in the space-time-energy domain. This approach permits for a data-driven identification of clusters so that, within multi-event clusters, the features of secondary and higher orders dependent events can be explored. The study is based on a revised dataset that is extracted from the catalog compiled by the Iranian Seismological Center (IRSC) for the period of 1996–2022. In order to exclude the effect of non-tectonic events, which turn out quite numerous within the study region, explosions within quarry-rich areas are removed; the identification of non-tectonic events is performed by considering the normalized ratios of daytime to nighttime events in an iterative removal procedure. According to preliminary analysis of the resulting catalog, an area is selected, within which a satisfactory completeness level is assessed for events with magnitude >2.0. Robust values of the scaling parameters, namely the b-value and the fractal dimension of epicenters, are also computed and are used to calculate the nearest-neighbor distances and to identify the earthquake clusters.The nearest-neighbor method also permits to investigate the internal structure of earthquake sequences, and to differentiate the spatial properties of seismicity according to the different topological features of the clusters structure. The obtained results allow us identifying two macro-areas, approximately separated by the 52°E meridian, which are characterized by different clustering features, namely: high complexity indexes, indicating simple (burst-like) structure of clusters, to the East; low complexity index, corresponding to complex multi-level (swarm-like) structure of clusters, to the West. The complexity measures, borrowed from network theory (i.e. the Closeness and Outdegree Centralization indexes), consistently capture the complexity of the identified clusters, and confirm that the cluster structures have distinct preferred geographic locations. The territorial heterogeneity of the examined clustering properties can be related with the spatial variability of tectonic, structural and geophysical features of the Alborz region, in good agreement with findings from the Alps-Dinarides junction (Northeastern Italy), a region also characterized by a contractional structural setting, mainly including reverse and strike-slip faulting systems, and by moderate to high seismic activity.

Monte Carlo simulation study of the structure of calcium and potassium chlorides ternary solutions

ABSTRACT Monte Carlo simulations were conducted on pure water and aqueous calciumand potassium chloride ternary solutions at different salts concentrations. To assess the local order within the solutions, the Monte Carlo simulations utilised the Madrid-2019 force field, usingtheTIP4P/2005 water model and scaled charges. The results demonstrate that as the salt concentrations increase, there is a notable decrease in the contribution of the structured tetrahedral arrangement of water molecules, favouring an unstructured configuration instead. The use of the nearest neighbour approach, provided unambiguous information regarding the radial and orientation characteristics of the unstructured configuration of water. Structures of the solvation shells around ions are also studied. The Monte Carlo simulations indicate that as the salt concentration increases, the hydration number around the cation decreases. This effect is likely due to the occasional substitution of some water molecules in the first hydration shell by chlorine ions.

Design and implementation of duty cycle-based futuristic clustering technique in WSN

In recent times, wireless sensor networks (WSNs) and their applications have exhibited a remarkable surge. These networks strive to devise and implement strategies that optimize network energy utilization, thereby extending their operational lifespan. An energy efficient network can be achieved using renewable source of energy and by controlling the duty cycle of nodes. The pivotal role of duty cycle in curtailing energy consumption in WSNs cannot be overstated. In this work, we introduce a novel duty cycle based futuristic clustering technique (DCBFCT) employing a nearest neighbor approach. This technique selectively induces sleep and awake modes in nodes, effectively minimizing the network’s overall energy consumption and, consequently, prolonging its lifespan. It calculates optimal node duty cycle values based on distance. Results demonstrate a substantial reduction in energy consumption, exhibiting an improved network lifetime. Empirical results presented in this study not only affirm the effectiveness of DCBFCT but also contribute valuable insights toward the development of sustainable and resilient WSNs in the era of burgeoning sensor network applications. The experimentation is conducted using the MATLAB/Simulink tool, considering diverse cases. The scalability and versatility of DCBFCT make it suitable for deployment in real-world applications, ranging from environmental monitoring to industrial automation.

Toward Cost-Effective Adaptive Random Testing: An Approximate Nearest Neighbor Approach

Toward Cost-Effective Adaptive Random Testing: An Approximate Nearest Neighbor Approach

Childhood neighbourhoods and life‐time fertility in twentieth‐century Southern Sweden: A k‐nearest neighbour approach

AbstractDespite a large literature on the importance of childhood neighbourhoods for life course transitions, there is a lack of fertility studies combining a life‐course perspective with detailed neighbourhood measures. Addressing this gap, we use longitudinal data in which the entire population of a Swedish town is geocoded at the address‐level, 1939–1967, and linked to national registers from 1968 to 2015. We study how social neighbourhoods in childhood influence fertility outcomes by constructing individual neighbourhoods at the address level to measure the social class of nearby childhood neighbours. We analyse the age at first and last birth, children ever born, birth spacing and childlessness. Growing up in higher‐status neighbourhoods is associated with delayed fertility for both men and women, but no association is found for the number of children ever born or for childlessness. Associations are stable over time, and later ages of neighbourhood exposure matter more, especially for men. Contrary to prior literature's focus on the lower classes, our results are driven by higher‐status individuals growing up in distinctly white‐collar neighbourhoods.

Bridging the feedback implementation gap: A comparison of empirical and rational decision rules in naturalistic psychotherapy

Objective: Previous research indicates positive effects of feedback based on rational or empirical decision rules in psychotherapy. The implementation of these usually session-to-session-based feedback systems into clinical practice, however, remains challenging. This study aims to evaluate decision rules based on routine outcome monitoring with reduced assessment frequency. Method: Data routinely collected every 5–20 sessions of N = 3758 patients treated with CBT in an outpatient clinic (M sessions = 42.8, SD = 15.4) were used to develop feedback decision rules based on the expected treatment response and nearest neighbors approach, the reliable change index, and method of percental improvement. The detection of patients at risk of treatment failure served as primary endpoint. Results: Significantly lower reliable improvement, higher reliable deterioration rates, and smaller effect sizes were found for patients identified at risk of treatment failure by all rules. The nearest neighbors-based approach showed the highest sensitivity regarding the detection of reliably deteriorated cases. Conclusion: Consistent with previous research, the empirical models outperformed the rational rules. Still, the first-time used percental improvement-based rule also showed satisfactory results. Overall, the results point to the potential of basic feedback systems that might be easier to implement in practice than session-to-session based systems.

A118 THE IMPACT OF HOMELESSNESS ON ENDOSCOPY UTILIZATION, SURGICAL TREATMENT, AND POST-OPERATIVE COMPLICATIONS AMONG HOSPITALIZED PATIENTS WITH CROHN’S DISEASE

Abstract Background Homelessness is an important public health issue with a prevalence around 4.2% in the United States. Homelessness has been associated with lack of access to healthcare services, poor quality of care, and negative health outcomes. However, there is limited data on clinical outcomes among homeless individuals with Crohn’s disease (CD). Aims In this study, we examine inpatient outcomes including endoscopy utilization, surgical treatment, and post-operative complications among homeless patients with CD. Methods The Healthcare Utilization Project (HCUP) State Inpatient Databases from New York, Arizona, Massachusetts and Florida for 2013 and 2014 were used to identify adults ≥18 years admitted with CD identified by International Classification of Diseases, Ninth Revision codes. Homeless patients were then identified, and propensity score matched using a one-to-ten greedy nearest-neighbor approach to non-homeless patients to balance the distribution of baseline covariates. Differences in outcomes including endoscopy utilization, surgical treatment and post-operative complications were reported as odds ratios (ORs) and 95% confidence intervals (CI). Results A total of 70,457 CD hospitalizations were identified, of which 469 (0.6%) were associated with homelessness. We matched 440 homeless patients with 4,400 housed patients with CD. Homeless individuals had a higher proportion of males (p=0.002), tended to be older (p=0.035), and had more comorbidities than non-homeless individuals. There was no difference in mortality between non-homeless and homeless patients (OR 1.6, 95% CI 0.72-3.57). Endoscopy utilization was comparable between non-homeless and homeless patients with CD including gastroscopy (OR 0.71, 95% CI 0.40-1.18), colonoscopy (OR 0.80, 95% CI 0.31-1.69), and sigmoidoscopy (OR 0.37, 95% CI 0.02-1.74) [Figure 1]. Surgical treatment was also comparable, except for lower rates of bowel resection among homeless patients (OR 0.55, 95% CI: 0.29-0.93). The two groups also had similar rates of post-operative complications including pulmonary, cardiovascular, gastrointestinal, and infectious complications (Figure 1). Conclusions Homeless patients with CD have comparable endoscopy utilization and surgical treatment as non-homeless patients, except for lower bowel resection rates. Postoperative complications are also similar between the two groups. Future studies are needed to better understand the impact of homelessness on health outcomes and healthcare service utilization among patients with CD. Figure 1. Endoscopy utilization, surgical treatment, and post-operative complications among homeless and non-homeless individuals using 1:10 propensity matching Funding Agencies None