Prospective evaluation of data-driven models to predict daily risk of Clostridioides difficile infection at 2 large academic health centers.

Many data-driven patient risk stratification models have not been evaluated prospectively. We performed and compared the prospective and retrospective evaluations of 2 Clostridioides difficile infection (CDI) risk-prediction models at 2 large academic health centers, and we discuss the models' robustness to data-set shifts.

Aging Alters the Aortic Proteome in Health and Thoracic Aortic Aneurysm.

BACKGROUND: Aging enhances most chronic diseases but its impact on human aortic tissue in health and in thoracic aortic aneurysms (TAA) remains unclear. METHODS: We employed a human aortic biorepository of healthy specimens (n=17) and those that underwent surgical repair for TAA (n=20). First, we performed proteomics comparing aortas of healthy donors to aneurysmal specimens, in young (ie, <60 years of age) and old (ie, ≥60 years of age) subjects. Second, we measured proteins, via immunoblotting, involved in mitophagy (ie, Parkin) and also mitochondrial-induced inflammatory pathways, specifically TLR (toll-like receptor) 9, STING (stimulator of interferon genes), and IFN (interferon)-ß. RESULTS: Proteomics revealed that aging transformed the aorta both quantitatively and qualitatively from health to TAA. Whereas young aortas exhibited an enrichment of immunologic processes, older aortas exhibited an enrichment of metabolic processes. Immunoblotting revealed that the expression of Parkin directly correlated to subject age in health but inversely to subject age in TAA. In TAA, but not in health, phosphorylation of STING and the expression of IFN-ß was impacted by aging regardless of whether subjects had bicuspid or tricuspid valves. In subjects with bicuspid valves and TAAs, TLR9 expression positively correlated with subject age. Interestingly, whereas phosphorylation of STING was inversely correlated with subject age, IFN-ß positively correlated with subject age. CONCLUSIONS: Aging transforms the human aortic proteome from health to TAA, leading to a differential regulation of biological processes. Our results suggest that the development of therapies to mitigate vascular diseases including TAA may need to be modified depending on subject age.

Early identification of patients admitted to hospital for covid-19 at risk of clinical deterioration: model development and multisite external validation study.

OBJECTIVE: To create and validate a simple and transferable machine learning model from electronic health record data to accurately predict clinical deterioration in patients with covid-19 across institutions, through use of a novel paradigm for model development and code sharing. DESIGN: Retrospective cohort study. SETTING: One US hospital during 2015-21 was used for model training and internal validation. External validation was conducted on patients admitted to hospital with covid-19 at 12 other US medical centers during 2020-21. PARTICIPANTS: 33 119 adults (≥18 years) admitted to hospital with respiratory distress or covid-19. MAIN OUTCOME MEASURES: An ensemble of linear models was trained on the development cohort to predict a composite outcome of clinical deterioration within the first five days of hospital admission, defined as in-hospital mortality or any of three treatments indicating severe illness: mechanical ventilation, heated high flow nasal cannula, or intravenous vasopressors. The model was based on nine clinical and personal characteristic variables selected from 2686 variables available in the electronic health record. Internal and external validation performance was measured using the area under the receiver operating characteristic curve (AUROC) and the expected calibration error-the difference between predicted risk and actual risk. Potential bed day savings were estimated by calculating how many bed days hospitals could save per patient if low risk patients identified by the model were discharged early. RESULTS: 9291 covid-19 related hospital admissions at 13 medical centers were used for model validation, of which 1510 (16.3%) were related to the primary outcome. When the model was applied to the internal validation cohort, it achieved an AUROC of 0.80 (95% confidence interval 0.77 to 0.84) and an expected calibration error of 0.01 (95% confidence interval 0.00 to 0.02). Performance was consistent when validated in the 12 external medical centers (AUROC range 0.77-0.84), across subgroups of sex, age, race, and ethnicity (AUROC range 0.78-0.84), and across quarters (AUROC range 0.73-0.83). Using the model to triage low risk patients could potentially save up to 7.8 bed days per patient resulting from early discharge. CONCLUSION: A model to predict clinical deterioration was developed rapidly in response to the covid-19 pandemic at a single hospital, was applied externally without the sharing of data, and performed well across multiple medical centers, patient subgroups, and time periods, showing its potential as a tool for use in optimizing healthcare resources.

Evaluating a Widely Implemented Proprietary Deterioration Index Model among Hospitalized Patients with COVID-19.

Rationale: The Epic Deterioration Index (EDI) is a proprietary prediction model implemented in over 100 U.S. hospitals that was widely used to support medical decision-making during the coronavirus disease (COVID-19) pandemic. The EDI has not been independently evaluated, and other proprietary models have been shown to be biased against vulnerable populations. Objectives: To independently evaluate the EDI in hospitalized patients with COVID-19 overall and in disproportionately affected subgroups. Methods: We studied adult patients admitted with COVID-19 to units other than the intensive care unit at a large academic medical center from March 9 through May 20, 2020. We used the EDI, calculated at 15-minute intervals, to predict a composite outcome of intensive care unit-level care, mechanical ventilation, or in-hospital death. In a subset of patients hospitalized for at least 48 hours, we also evaluated the ability of the EDI to identify patients at low risk of experiencing this composite outcome during their remaining hospitalization. Results: Among 392 COVID-19 hospitalizations meeting inclusion criteria, 103 (26%) met the composite outcome. The median age of the cohort was 64 (interquartile range, 53-75) with 168 (43%) Black patients and 169 (43%) women. The area under the receiver-operating characteristic curve of the EDI was 0.79 (95% confidence interval, 0.74-0.84). EDI predictions did not differ by race or sex. When exploring clinically relevant thresholds of the EDI, we found patients who met or exceeded an EDI of 68.8 made up 14% of the study cohort and had a 74% probability of experiencing the composite outcome during their hospitalization with a sensitivity of 39% and a median lead time of 24 hours from when this threshold was first exceeded. Among the 286 patients hospitalized for at least 48 hours who had not experienced the composite outcome, 14 (13%) never exceeded an EDI of 37.9, with a negative predictive value of 90% and a sensitivity above this threshold of 91%. Conclusions: We found the EDI identifies small subsets of high-risk and low-risk patients with COVID-19 with good discrimination, although its clinical use as an early warning system is limited by low sensitivity. These findings highlight the importance of independent evaluation of proprietary models before widespread operational use among patients with COVID-19.

Comparison of readmission and early revision rates as a quality metric in total knee arthroplasty using the Nationwide Readmission Database.

BACKGROUND: After release of the Comprehensive Care for Joint Replacement bundle, there has been increased emphasis on reducing readmission rates for total knee arthroplasty (TKA). The potential for a separate, clinically-relevant metric, TKA revision rates within a year following surgery, has not been fully explored. Based on this, we compared rates and payments for TKA readmission and revision procedures as metrics for improving quality and cost. METHODS: We utilized the 2013 Nationwide Readmission Database (NRD) to examine national readmission and revision rates, the reasons for revision procedures, and associated costs for elective TKA procedures. As data are not linked across years, we examined revision rates for TKA completed in the month of January by capturing revision procedures in the subsequent following 11-month period to approximate a 1-year revision rate. Diagnosis and procedure codes for revision procedures were collected. Average readmission and revision procedure costs were then calculated, and the cost distributed across the entire TKA population. RESULTS: We identified 20,851 patients having TKA surgery. The mean unadjusted 30- and 90-day TKA readmission rates were 3.4% and 5.8%, respectively. In contrast, the mean unadjusted 3-month and approximate 1-year reoperation rates were 1.0% and 1.6%, respectively. The most common cause for revision was periprosthetic joint infection, which accounting for 62% of all reported revision procedures. The mean payment for 90-day readmission was roughly half ($10,589±$11,084) of the mean inpatient payment for single reoperation procedure at 90 days ($20,222±$17,799). Importantly, nearly half (46%) of all 90-day readmissions were associated with a reoperation event within the first year. CONCLUSIONS: Readmission following TKA is associated with a 1-year reoperation in approximately half of patients. These reoperations represent a significant patient burden and have a higher per episode cost. Early reoperation may represent a more clinically relevant target for quality improvement and cost containment.

Evaluating a Widely Implemented Proprietary Deterioration Index Model Among Hospitalized COVID-19 Patients.

INTRODUCTION: The Epic Deterioration Index (EDI) is a proprietary prediction model implemented in over 100 U.S. hospitals that was widely used to support medical decision-making during the COVID-19 pandemic. The EDI has not been independently evaluated, and other proprietary models have been shown to be biased against vulnerable populations. METHODS: We studied adult patients admitted with COVID-19 to non-ICU care at a large academic medical center from March 9 through May 20, 2020. We used the EDI, calculated at 15-minute intervals, to predict a composite outcome of ICU-level care, mechanical ventilation, or in-hospital death. In a subset of patients hospitalized for at least 48 hours, we also evaluated the ability of the EDI to identify patients at low risk of experiencing this composite outcome during their remaining hospitalization. RESULTS: Among 392 COVID-19 hospitalizations meeting inclusion criteria, 103 (26%) met the composite outcome. Median age of the cohort was 64 (IQR 53-75) with 168 (43%) African Americans and 169 (43%) women. Area under the receiver-operating-characteristic curve (AUC) of the EDI was 0.79 (95% CI 0.74-0.84). EDI predictions did not differ by race or sex. When exploring clinically-relevant thresholds of the EDI, we found patients who met or exceeded an EDI of 68.8 made up 14% of the study cohort and had a 74% probability of experiencing the composite outcome during their hospitalization with a median lead time of 24 hours from when this threshold was first exceeded. Among the 286 patients hospitalized for at least 48 hours who had not experienced the composite outcome, 14 (13%) never exceeded an EDI of 37.9, with a negative predictive value of 90% and a sensitivity above this threshold of 91%. CONCLUSION: We found the EDI identifies small subsets of high- and low-risk COVID-19 patients with fair discrimination. We did not find evidence of bias by race or sex. These findings highlight the importance of independent evaluation of proprietary models before widespread operational use among COVID-19 patients.

Dynamic readmission prediction using routine postoperative laboratory results after radical cystectomy.

OBJECTIVE: To determine if the addition of electronic health record data enables better risk stratification and readmission prediction after radical cystectomy. Despite efforts to reduce their frequency and severity, complications and readmissions following radical cystectomy remain common. Leveraging readily available, dynamic information such as laboratory results may allow for improved prediction and targeted interventions for patients at risk of readmission. METHODS: We used an institutional electronic medical records database to obtain demographic, clinical, and laboratory data for patients undergoing radical cystectomy. We characterized the trajectory of common postoperative laboratory values during the index hospital stay using support vector machine learning techniques. We compared models with and without laboratory results to assess predictive ability for readmission. RESULTS: Among 996 patients who underwent radical cystectomy, 259 patients (26%) experienced a readmission within 30 days. During the first week after surgery, median daily values for white blood cell count, urea nitrogen, bicarbonate, and creatinine differentiated readmitted and nonreadmitted patients. Inclusion of laboratory results greatly increased the ability of models to predict 30-day readmissions after cystectomy. CONCLUSIONS: Common postoperative laboratory values may have discriminatory power to help identify patients at higher risk of readmission after radical cystectomy. Dynamic sources of physiological data such as laboratory values could enable more accurate identification and targeting of patients at greatest readmission risk after cystectomy. This is a proof of concept study that suggests further exploration of these techniques is warranted.

Inaugural Readmission Penalties for Total Hip and Total Knee Arthroplasty Procedures Under the Hospital Readmissions Reduction Program.

Importance: The Hospital Readmissions Reduction Program (HRRP) is a Centers for Medicare and Medicaid Services policy that levies hospital reimbursement penalties based on excess readmissions of patients with 4 medical conditions and 3 surgical procedures. A greater understanding of factors associated with the 3 surgical reimbursement penalties is needed for clinicians in surgical practice. Objective: To investigate the first year of HRRP readmission penalties applied to 2 surgical procedures-elective total hip arthroplasty (THA) and total knee arthroplasty (TKA)-in the context of hospital and patient characteristics. Design, Setting, and Participants: Fiscal year 2015 HRRP penalization data from Hospital Compare were linked with the American Hospital Association Annual Survey and with the Healthcare Cost and Utilization Project State Inpatient Database for hospitals in the state of Florida. By using a case-control framework, those hospitals were separated based on HRRP penalty severity, as measured with the HRRP THA and TKA excess readmission ratio, and compared according to orthopedic volume as well as hospital-level and patient-level characteristics. The first year of HRRP readmission penalties applied to surgery in Florida Medicare subsection (d) hospitals was examined, identifying 60 663 Medicare patients who underwent elective THA or TKA in 143 Florida hospitals. The data analysis was conducted from February 2016 to January 2017. Exposures: Annual hospital THA and TKA volume, other hospital-level characteristics, and patient factors used in HRRP risk adjustment. Main Outcomes and Measures: The HRRP penalties with HRRP excess readmission ratios were measured, and their association with annual THA and TKA volume, a common measure of surgical quality, was evaluated. The HRRP penalties for surgical care according to hospital and readmitted patient characteristics were then examined. Results: Among 143 Florida hospitals, 2991 of 60 663 Medicare patients (4.9%) who underwent THA or TKA were readmitted within 30 days. Annual hospital arthroplasty volume seemed to follow an inverse association with both unadjusted readmission rates (r = -0.16, P = .06) and HRRP risk-adjusted readmission penalties (r = -0.12, P = .14), but these associations were not statistically significant. Other hospital characteristics and readmitted patient characteristics were similar across HRRP orthopedic penalty severity. Conclusions and Relevance: This study's findings suggest that higher-volume hospitals had less severe, but not significantly different, rates of readmission and HRRP penalties, without systematic differences across readmitted patients.

Using Machine Learning and the Electronic Health Record to Predict Complicated Clostridium difficile Infection.

BACKGROUND: Clostridium (Clostridioides) difficile infection (CDI) is a health care-associated infection that can lead to serious complications. Potential complications include intensive care unit (ICU) admission, development of toxic megacolon, need for colectomy, and death. However, identifying the patients most likely to develop complicated CDI is challenging. To this end, we explored the utility of a machine learning (ML) approach for patient risk stratification for complications using electronic health record (EHR) data. METHODS: We considered adult patients diagnosed with CDI between October 2010 and January 2013 at the University of Michigan hospitals. Cases were labeled complicated if the infection resulted in ICU admission, colectomy, or 30-day mortality. Leveraging EHR data, we trained a model to predict subsequent complications on each of the 3 days after diagnosis. We compared our EHR-based model to one based on a small set of manually curated features. We evaluated model performance using a held-out data set in terms of the area under the receiver operating characteristic curve (AUROC). RESULTS: Of 1118 cases of CDI, 8% became complicated. On the day of diagnosis, the model achieved an AUROC of 0.69 (95% confidence interval [CI], 0.55-0.83). Using data extracted 2 days after CDI diagnosis, performance increased (AUROC, 0.90; 95% CI, 0.83-0.95), outperforming a model based on a curated set of features (AUROC, 0.84; 95% CI, 0.75-0.91). CONCLUSIONS: Using EHR data, we can accurately stratify CDI cases according to their risk of developing complications. Such an approach could be used to guide future clinical studies investigating interventions that could prevent or mitigate complicated CDI.

Role of Post-Acute Care on Hospital Readmission After High-Risk Surgery.

BACKGROUND: Payment models, including the Hospital Readmissions Reduction Program and bundled payments, place pressures on hospitals to limit readmissions. Against this backdrop, we sought to investigate the association of post-acute care after major surgery and readmission rates. METHODS: We identified patients undergoing high-risk surgery (abdominal aortic aneurysm repair, coronary bypass grafting, aortic valve replacement, carotid endarterectomy, esophagectomy, pancreatectomy, lung resection, and cystectomy) from 2005 to 2010 using the Healthcare Cost and Utilization Project's State Inpatient Database. The primary outcome was readmission rates after major surgery. Secondary outcome was readmission length of stay. RESULTS: We identified 135,523 patients of whom 56,720 (42%) received post-acute care. Patients receiving post-acute care had higher readmission rates than those who were discharged home (16% versus 10%, respectively; P < 0.001). The risk-adjusted readmission length of stay was greatest for patients who received care from a skilled nursing facility, followed by those who received home care, and lowest for those who did not receive post-acute care (7.1 versus 5.4 versus 4.8 d, respectively; P < 0.001). CONCLUSIONS: The use of post-acute care was associated with higher readmission rates and higher readmission lengths of stay. Improving the support of patients in post-acute care settings may help reduce readmissions and readmission intensity.

Predictors and Cost of Readmission in Total Knee Arthroplasty.

BACKGROUND: The Comprehensive Care for Joint Replacement bundle was created to decrease total knee arthroplasty (TKA) cost. To help accomplish this, there is a focus on reducing TKA readmissions. However, there is a lack of national representative sample of all-payer hospital admissions to direct strategy, identify risk factors for readmission, and understand actual readmission cost. METHODS: We used the Nationwide Readmission Database to examine national readmission rates, predictors of readmission, and associated readmission costs for elective TKA procedures. We fit a multivariable logistic regression model to examine factors associated with readmission. Then, we determined mean readmission costs and calculated the readmission cost when distributed across the entire TKA population. RESULTS: We identified 224,465 patients having TKA across all states participating in the Nationwide Readmission Database. The mean unadjusted 30-day TKA readmission rate was 4%. The greatest predictors of readmission were congestive heart failure (odds ratio [OR] 2.51, 95% confidence interval [CI] 2.62-2.80), renal disease (OR 2.19, 95% CI 2.03-2.37), and length of stay greater than 4 days (OR 2.4, 95% CI 2.25-2.61). The overall median cost for each readmission was $6753 ± 175. Extrapolating the readmission cost for the entire TKA population resulted in the readmission cost being 2% of the overall 30-day procedure cost. CONCLUSIONS: A major focus of the Comprehensive Care for Joint Replacement bundle is improving cost and quality by limiting readmission rates. TKA readmissions are low and comprise a small percentage of total TKA cost, suggesting that they may not be the optimal measure of quality care or a significant driver of overall cost.

Variation in readmission expenditures after high-risk surgery.

BACKGROUND: The Hospital Readmissions Reduction Program reduces payments to hospitals with excess readmissions for three common medical conditions and recently extended its readmission program to surgical patients. We sought to investigate readmission intensity as measured by readmission cost for high-risk surgeries and examine predictors of higher readmission costs. MATERIALS AND METHODS: We used the Healthcare Cost and Utilization Project's State Inpatient Database to perform a retrospective cohort study of patients undergoing major chest (aortic valve replacement, coronary artery bypass grafting, lung resection) and major abdominal (abdominal aortic aneurysm repair [open approach], cystectomy, esophagectomy, pancreatectomy) surgery in 2009 and 2010. We fit a multivariable logistic regression model with generalized estimation equations to examine patient and index admission factors associated with readmission costs. RESULTS: The 30-d readmission rate was 16% for major chest and 22% for major abdominal surgery (P < 0.001). Discharge to a skilled nursing facility was associated with higher readmission costs for both chest (odds ratio [OR]: 1.99; 95% confidence interval [CI]: 1.60-2.48) and abdominal surgeries (OR: 1.86; 95% CI: 1.24-2.78). Comorbidities, length of stay, and receipt of blood or imaging was associated with higher readmission costs for chest surgery patients. Readmission >3 wk after discharge was associated with lower costs among abdominal surgery patients. CONCLUSIONS: Readmissions after high-risk surgery are common, affecting about one in six patients. Predictors of higher readmission costs differ among major chest and abdominal surgeries. Better identifying patients susceptible to higher readmission costs may inform future interventions to either reduce the intensity of these readmissions or eliminate them altogether.

No Differences in Population-based Readmissions After Open and Robotic-assisted Radical Cystectomy: Implications for Post-discharge Care.

OBJECTIVE: To inform whether readmission reduction strategies should consider surgical approach, we examined readmission differences between open and robotic-assisted radical cystectomy (RARC) using population-based data. METHODS: We identified patients who underwent cystectomy between January 2010 and September 2013 based on International Classification of Diseases-9th edition codes and administrative claims from a large, national US health insurer (Clinformatics Data Mart Database, OptumInsight, Eden Prairie, MN). We assessed post-discharge health system utilization and tested for differences in readmissions after the 2 surgical approaches. RESULTS: We identified 935 patients treated with cystectomy: open = 785 (84%) and RARC = 150 (16%). Patients undergoing RARC were slightly older, male, had more ileal conduit urinary reconstruction, and less need for intensive care. Index length of stay was shorter for RARC than for open surgery (7 days vs 8 days, P < .001). However, we found no differences in 30-day readmission rates (24% open vs 29% RARC, P = .26) or other readmission parameters, including readmission length of stay (5 days open vs 4 days RARC, P = .32), emergency department use (22% open vs 24% RARC, P = .86), reasons for readmission, or timing of first outpatient visits (11.5 days open vs 9 days RARC, P = .41). For both approaches, the majority of patients were readmitted within 2 weeks. CONCLUSION: The surgical approach to cystectomy does not appear to impact readmissions. Strategies to reduce the readmission burden after cystectomy do not need to consider surgical approach but should focus on timing of medical contacts.

All spiking, sustained ON displaced amacrine cells receive gap-junction input from melanopsin ganglion cells.

Retinal neurons exhibit sustained versus transient light responses, which are thought to encode low- and high-frequency stimuli, respectively. This dichotomy has been recognized since the earliest intracellular recordings from the 1960s, but the underlying mechanisms are not yet fully understood. We report that in the ganglion cell layer of rat retinas, all spiking amacrine interneurons with sustained ON photoresponses receive gap-junction input from intrinsically photosensitive retinal ganglion cells (ipRGCs), recently discovered photoreceptors that specialize in prolonged irradiance detection. This input presumably allows ipRGCs to regulate the secretion of neuromodulators from these interneurons. We have identified three morphological varieties of such ipRGC-driven displaced amacrine cells: (1) monostratified cells with dendrites terminating exclusively in sublamina S5 of the inner plexiform layer, (2) bistratified cells with dendrites in both S1 and S5, and (3) polyaxonal cells with dendrites and axons stratifying in S5. Most of these amacrine cells are wide field, although some are medium field. The three classes respond to light differently, suggesting that they probably perform diverse functions. These results demonstrate that ipRGCs are a major source of tonic visual information within the retina and exert widespread intraretinal influence. They also add to recent evidence that ganglion cells signal not only to the brain.

Melatonin Suppression by Light in Humans Is More Sensitive Than Previously Reported.

The retina drives various non-image-forming photoresponses, including circadian photoentrainment and pupil constriction. Previous investigators showed that in humans, photic suppression of the clock-controlled hormone melatonin is most sensitive to 460-nm blue light, with a threshold of ~12 log photons cm(-2) s(-1). This threshold is surprising because non-image-forming vision is mediated by intrinsically photosensitive retinal ganglion cells, which receive rod-driven synaptic input and can respond to light levels as low as ~7 log photons cm(-2) s(-1). Using a protocol that enhances data precision, we have found the threshold for human melatonin suppression to be ~10 log photons cm(-2) s(-1) at 460 nm. This finding has far-reaching implications since there is mounting evidence that nocturnal activation of the circadian system can be harmful.

The rat retina has five types of ganglion-cell photoreceptors.

Intrinsically photosensitive retinal ganglion cells (ipRGCs) are inner retinal photoreceptors that mediate non-image-forming visual functions, e.g. pupillary constriction, regulation of pineal melatonin release, and circadian photoentrainment. Five types of ipRGCs were recently discovered in mouse, but whether they exist in other mammals remained unknown. We report that the rat also has five types of ipRGCs, whose morphologies match those of mouse ipRGCs; this is the first demonstration of all five cell types in a non-mouse species. Through immunostaining and λmax measurements, we showed that melanopsin is likely the photopigment of all rat ipRGCs. The various cell types exhibited diverse spontaneous spike rates, with the M1 type spiking the least and M4 spiking the most, just like we had observed for their mouse counterparts. Also similar to mouse, all ipRGCs in rat generated not only sluggish intrinsic photoresponses but also fast, synaptically driven ones. However, we noticed two significant differences between these species. First, whereas we learned previously that all mouse ipRGCs had equally sustained synaptic light responses, rat M1 cells' synaptic photoresponses were far more transient than those of M2-M5. Since M1 cells provide all input to the circadian clock, this rat-versus-mouse discrepancy could explain the difference in photoentrainment threshold between mouse and other species. Second, rat ipRGCs' melanopsin-based spiking photoresponses could be classified into three varieties, but only two were discerned for mouse ipRGCs. This correlation of spiking photoresponses with cell types will help researchers classify ipRGCs in multielectrode-array (MEA) spike recordings.

A low dimensional approach on network characterization.

In many applications, one may need to characterize a given network among a large set of base networks, and these networks are large in size and diverse in structure over the search space. In addition, the characterization algorithms are required to have low volatility and with a small circle of uncertainty. For large datasets, these algorithms are computationally intensive and inefficient. However, under the context of network mining, a major concern of some applications is speed. Hence, we are motivated to develop a fast characterization algorithm, which can be used to quickly construct a graph space for analysis purpose. Our approach is to transform a network characterization measure, commonly formulated based on similarity matrices, into simple vector form signatures. We shall show that the [Formula: see text] similarity matrix can be represented by a dyadic product of two N-dimensional signature vectors; thus the network alignment process, which is usually solved as an assignment problem, can be reduced into a simple alignment problem based on separate signature vectors.