scholarly journals Development and external validation of prognostic models for COVID-19 to support risk stratification in secondary care

2021 ◽  
Author(s):  
Nicola J Adderley ◽  
Thomas Taverner ◽  
Malcolm Price ◽  
Christopher Sainsbury ◽  
David Greenwood ◽  
...  
Keyword(s):  
Secondary Care ◽  
External Validation ◽  
Model Performance ◽  
Demographic Variables ◽  
Prognostic Models ◽  
Physiological Measures ◽  
Good Discrimination ◽  
Improve Model ◽  
Wide Range ◽  
Improve Model Performance

AbstractObjectivesExisting UK prognostic models for patients admitted to hospital with COVID-19 are limited by reliance on comorbidities, which are under-recorded in secondary care, and lack of imaging data among the candidate predictors. Our aims were to develop and externally validate novel prognostic models for adverse outcomes (death, intensive therapy unit (ITU) admission) in UK secondary care; and externally validate the existing 4C score.DesignCandidate predictors included demographic variables, symptoms, physiological measures, imaging, laboratory tests. Final models used logistic regression with stepwise selection.SettingModel development was performed in data from University Hospitals Birmingham (UHB). External validation was performed in the CovidCollab dataset.ParticipantsPatients with COVID-19 admitted to UHB January-August 2020 were included.Main outcome measuresDeath and ITU admission within 28 days of admission.Results1040 patients with COVID-19 were included in the derivation cohort; 288 (28%) died and 183 (18%) were admitted to ITU within 28 days of admission. Area under the receiver operating curve (AUROC) for mortality was 0.791 (95%CI 0.761-0.822) in UHB and 0.767 (95%CI 0.754-0.780) in CovidCollab; AUROC for ITU admission was 0.906 (95%CI 0.883-0.929) in UHB and 0.811 (95%CI 0.795-0.828) in CovidCollab. Models showed good calibration. Addition of comorbidities to candidate predictors did not improve model performance. AUROC for the 4C score in the UHB dataset was 0.754 (95%CI 0.721-0.786).ConclusionsThe novel prognostic models showed good discrimination and calibration in derivation and external validation datasets, and outperformed the existing 4C score. The models can be integrated into electronic medical records systems to calculate each individual patient’s probability of death or ITU admission at the time of hospital admission. Implementation of the models and clinical utility should be evaluated.Article SummaryStrengths and limitations of this studyWe developed novel prognostic models predicting mortality and ITU admission within 28 days of admission for patients hospitalised with COVID-19, using a large routinely collected dataset gathered at admission with a wide range of possible predictors (demographic variables, symptoms, physiological measures, imaging, laboratory test results).These novel models showed good discrimination and calibration in both derivation and external validation cohorts, and outperformed the existing ISARIC model and 4C score in the derivation dataset. We found that addition of comorbidities to the set of candidate predictors included in model derivation did not improve model performance.If integrated into hospital electronic medical records systems, the model algorithms will provide a predicted probability of mortality or ITU admission for each patient based on their individual data at, or close to, the time of admission, which will support clinicians’ decision making with regard to appropriate patient care pathways and triage. This information might also assist clinicians in explaining complex prognostic assessments and decisions to patients and their relatives.A limitation of the study was that in the external validation cohort we were unable to examine all of the predictors included in the original full UHB model due to only a reduced set of candidate predictors being available in CovidCollab. Nevertheless, the reduced model performed well and the results suggest it may be applicable in a wide range of datasets where only a reduced set of predictor variables is available.Furthermore, it was not possible to carry out stratified analysis by ethnicity as the UHB dataset contained too few patients in most of the strata, and no ethnicity data was available in the CovidCollab dataset.

scholarly journals Towards Robust Representations of Spatial Networks Using Graph Neural Networks

2021 ◽  
Vol 11 (15) ◽  
pp. 6918
Author(s):  
Chidubem Iddianozie ◽  
Gavin McArdle
Keyword(s):  
Machine Learning ◽  
Model Performance ◽  
Network Models ◽  
Data Representation ◽  
Spatial Networks ◽  
Neural Network Models ◽  
Improve Model ◽  
Graph Neural Networks ◽  
Spatial Entities ◽  
Improve Model Performance

The effectiveness of a machine learning model is impacted by the data representation used. Consequently, it is crucial to investigate robust representations for efficient machine learning methods. In this paper, we explore the link between data representations and model performance for inference tasks on spatial networks. We argue that representations which explicitly encode the relations between spatial entities would improve model performance. Specifically, we consider homogeneous and heterogeneous representations of spatial networks. We recognise that the expressive nature of the heterogeneous representation may benefit spatial networks and could improve model performance on certain tasks. Thus, we carry out an empirical study using Graph Neural Network models for two inference tasks on spatial networks. Our results demonstrate that heterogeneous representations improves model performance for down-stream inference tasks on spatial networks.

scholarly journals Population Pharmacokinetics of Doxycycline in Children

2019 ◽  
Vol 63 (12) ◽  
Author(s):  
Elizabeth J. Thompson ◽  
Huali Wu ◽  
Chiara Melloni ◽  
Stephen Balevic ◽  
Janice E. Sullivan ◽  
...  
Keyword(s):  
Model Performance ◽  
Standard Of Care ◽  
Volume Of Distribution ◽  
Clinical Value ◽  
Final Model ◽  
Childhood Infections ◽  
Obesity Status ◽  
Improve Model ◽  
Using Data ◽  
Improve Model Performance

ABSTRACT Doxycycline is a tetracycline-class antimicrobial labeled by the U.S. Food and Drug Administration for children >8 years of age for many common childhood infections. Doxycycline is not labeled for children ≤8 years of age, due to the association between tetracycline-class antibiotics and tooth staining, although doxycycline may be used off-label under severe conditions. Accordingly, there is a paucity of pharmacokinetic (PK) data to guide dosing in children 8 years and younger. We leveraged opportunistically collected plasma samples after intravenous (i.v.) and oral doxycycline doses received per standard of care to characterize the PK of doxycycline in children of different ages and evaluated the effect of obesity and fasting status on PK parameters. We developed a population PK model of doxycycline using data collected from 47 patients 0 to 18 years of age, including 14 participants ≤8 years. We developed a 1-compartment PK model and found doxycycline clearance to be 3.32 liters/h/70 kg of body weight and volume to be 96.8 liters/70 kg for all patients, comparable to values reported in adults. We estimated a bioavailability of 89.6%, also consistent with adult data. Allometrically scaled clearance and volume of distribution did not differ between children 2 to ≤8 years of age and children >8 to ≤18 years of age, suggesting that younger children may be given the same per-kilogram dosing. Obesity status and fasting status were not selected for inclusion in the final model. Additional doxycycline PK samples collected in future studies may be used to improve model performance and maximize its clinical value.

scholarly journals Improving biomedical named entity recognition with syntactic information

2020 ◽  
Vol 21 (1) ◽  
Author(s):  
Yuanhe Tian ◽  
Wang Shen ◽  
Yan Song ◽  
Fei Xia ◽  
Min He ◽  
...  
Keyword(s):  
Named Entity Recognition ◽  
Model Performance ◽  
Entity Recognition ◽  
Input Word ◽  
Named Entity ◽  
Improve Model ◽  
Syntactic Information ◽  
Biomedical Texts ◽  
Improve Model Performance ◽  
Biomedical Named Entity Recognition

Abstract Background Biomedical named entity recognition (BioNER) is an important task for understanding biomedical texts, which can be challenging due to the lack of large-scale labeled training data and domain knowledge. To address the challenge, in addition to using powerful encoders (e.g., biLSTM and BioBERT), one possible method is to leverage extra knowledge that is easy to obtain. Previous studies have shown that auto-processed syntactic information can be a useful resource to improve model performance, but their approaches are limited to directly concatenating the embeddings of syntactic information to the input word embeddings. Therefore, such syntactic information is leveraged in an inflexible way, where inaccurate one may hurt model performance. Results In this paper, we propose BioKMNER, a BioNER model for biomedical texts with key-value memory networks (KVMN) to incorporate auto-processed syntactic information. We evaluate BioKMNER on six English biomedical datasets, where our method with KVMN outperforms the strong baseline method, namely, BioBERT, from the previous study on all datasets. Specifically, the F1 scores of our best performing model are 85.29% on BC2GM, 77.83% on JNLPBA, 94.22% on BC5CDR-chemical, 90.08% on NCBI-disease, 89.24% on LINNAEUS, and 76.33% on Species-800, where state-of-the-art performance is obtained on four of them (i.e., BC2GM, BC5CDR-chemical, NCBI-disease, and Species-800). Conclusion The experimental results on six English benchmark datasets demonstrate that auto-processed syntactic information can be a useful resource for BioNER and our method with KVMN can appropriately leverage such information to improve model performance.

Potential for developing independent daytime/nighttime LUR models based on short-term mobile monitoring to improve model performance

2021 ◽  
Vol 268 ◽  
pp. 115951
Author(s):  
Xiangyu Xu ◽  
Ning Qin ◽  
Zhenchun Yang ◽  
Yunwei Liu ◽  
Suzhen Cao ◽  
...  
Keyword(s):  
Model Performance ◽  
Mobile Monitoring ◽  
Short Term ◽  
Improve Model ◽  
Improve Model Performance

scholarly journals Probing Behavior of Diaphorina citri (Hemiptera: Liviidae) on Valencia Orange Influenced by Sex, Color, and Size

2020 ◽  
Vol 20 (2) ◽  
Author(s):  
Timothy A Ebert ◽  
Michael E Rogers
Keyword(s):  
Body Size ◽  
Model Performance ◽  
Diaphorina Citri ◽  
Probing Behavior ◽  
Valencia Orange ◽  
Improve Model ◽  
Candidatus Liberibacter ◽  
Citrus Huanglongbing ◽  
Liberibacter Asiaticus ◽  
Improve Model Performance

Abstract Candidatus Liberibacter asiaticus Jagoueix, Bové, and Garnier (Rhizobiales: Rhizobiaceae) is transmitted by the psyllid Diaphorina citri Kuwayama and putatively causes Huanglongbing disease in citrus. Huanglongbing has reduced yields by 68% relative to pre-disease yields in Florida. Disease management is partly through vector control. Understanding vector biology is essential in this endeavor. Our goal was to document differences in probing behavior linked to sex. Based on both a literature review and our results, we conclude that there is either no effect of sex or that identifying such an effect requires a sample size at least four times larger than standard methodologies. Including both color and sex in statistical models did not improve model performance. Both sex and color are correlated with body size, and body size has not been considered in previous studies on sex in D. citri in terms of probing behavior. An effect of body size was found wherein larger psyllids took longer to reach ingestion behaviors and larger individuals spent more time-ingesting phloem, but these relationships explained little of the variability in these data. We suggest that the effects of sex can be ignored when running EPG experiments on healthy psyllids.

scholarly journals Performance of African-ancestry-specific polygenic hazard score varies according to local ancestry in 8q24

2021 ◽  
Author(s):  
Roshan A. Karunamuni ◽  
Minh-Phuong Huynh-Le ◽  
Chun C. Fan ◽  
Wesley Thompson ◽  
Asona Lui ◽  
...  
Keyword(s):  
Proportional Hazards ◽  
African Ancestry ◽  
Prostate Cancer Risk ◽  
Model Performance ◽  
Principal Component ◽  
Cox Proportional Hazards ◽  
Risk Scores ◽  
Improve Model ◽  
Improve Model Performance ◽  
Improve Health

AbstractWe previously developed an African-ancestry-specific polygenic hazard score (PHS46+African) that substantially improved prostate cancer risk stratification in men with African ancestry. The model consists of 46 SNPs identified in Europeans and 3 SNPs from 8q24 shown to improve model performance in Africans. Herein, we used principal component (PC) analysis to uncover subpopulations of men with African ancestry for whom the utility of PHS46+African may differ. Genotypic data were obtained from PRACTICAL consortium for 6,253 men with African genetic ancestry. Genetic variation in a window spanning 3 African-specific 8q24 SNPs was estimated using 93 PCs. A Cox proportional hazards framework was used to identify the pair of PCs most strongly associated with performance of PHS46+African. A calibration factor (CF) was formulated using estimated Cox coefficients to quantify the extent to which the performance of PHS46+African varies with PC. CF of PHS46+African was strongly associated with the first and twentieth PCs. Predicted CF ranged from 0.41 to 2.94, suggesting that PHS46+African may be up to 7 times more beneficial to some African men than others. The explained relative risk for PHS46+African varied from 3.6% to 9.9% for individuals with low and high CF values, respectively. By cross-referencing our dataset with 1000 Genomes, we identified statistically significant associations between continental and calibration groupings. In conclusion, we identified PCs within 8q24 SNP window that were strongly associated with performance of PHS46+African. Further research to improve clinical utility of polygenic risk scores (or models) is needed to improve health outcomes for men of African ancestry

scholarly journals A Process-Oriented Model for Simulating Avalanche Danger

1980 ◽  
Vol 26 (94) ◽  
pp. 53-63
Author(s):  
Arthur Judson ◽  
Charles F. Leaf ◽  
Glen E. Brink
Keyword(s):  
Sensitivity Analysis ◽  
Snow Depth ◽  
Process Model ◽  
Model Performance ◽  
Simulation Process ◽  
Improve Model ◽  
Westerly Winds ◽  
Improve Model Performance ◽  
Snow Transport ◽  
Avalanche Danger

AbstractA simulation process model is developed for rating avalanche danger for twelve east-facing avalanche paths loaded by westerly winds. The model simulates layer age and densification, snow depth, snow transport and deposition, formation of melt crusts, snow temperatures, temperature gradient metamorphism, and avalanche danger on a 6 h basis. Conditioned on avalanches alone, the model predicted avalanche potential on 86% of the 175 avalanche days during an eight-year period. It indicated avalanche potential 50% of the time on non-avalanche days. A sensitivity analysis is under way to improve model performance, and simulation of danger from additional avalanche samples is planned.

scholarly journals Integrating a Measure of Social Vulnerability Into a Parsimonious Surgical Risk Calculator Does Not Improve Model Performance

2021 ◽  
Vol 233 (5) ◽  
pp. e48
Author(s):  
Adam R. Dyas ◽  
Heather Carmichael ◽  
Michael R. Bronsert ◽  
William G. Henderson ◽  
Helen J. Madsen ◽  
...  
Keyword(s):  
Social Vulnerability ◽  
Model Performance ◽  
Risk Calculator ◽  
Surgical Risk ◽  
Improve Model ◽  
Improve Model Performance

scholarly journals Empirical evidence of the impact of study characteristics on the performance of prediction models: a meta-epidemiological study

BMJ Open ◽  
2019 ◽  
Vol 9 (4) ◽  
pp. e026160 ◽  
Author(s):  
Johanna A A G Damen ◽  
Thomas P A Debray ◽  
Romin Pajouheshnia ◽  
Johannes B Reitsma ◽  
Rob J P M Scholten ◽  
...  
Keyword(s):  
Performance Measures ◽  
Systematic Reviews ◽  
Validation Studies ◽  
Prognostic Model ◽  
External Validation ◽  
Model Performance ◽  
Population Characteristics ◽  
Prognostic Models ◽  
C Statistic ◽  
Study Characteristics

ObjectivesTo empirically assess the relation between study characteristics and prognostic model performance in external validation studies of multivariable prognostic models.DesignMeta-epidemiological study.Data sources and study selectionOn 16 October 2018, we searched electronic databases for systematic reviews of prognostic models. Reviews from non-overlapping clinical fields were selected if they reported common performance measures (either the concordance (c)-statistic or the ratio of observed over expected number of events (OE ratio)) from 10 or more validations of the same prognostic model.Data extraction and analysesStudy design features, population characteristics, methods of predictor and outcome assessment, and the aforementioned performance measures were extracted from the included external validation studies. Random effects meta-regression was used to quantify the association between the study characteristics and model performance.ResultsWe included 10 systematic reviews, describing a total of 224 external validations, of which 221 reported c-statistics and 124 OE ratios. Associations between study characteristics and model performance were heterogeneous across systematic reviews. C-statistics were most associated with variation in population characteristics, outcome definitions and measurement and predictor substitution. For example, validations with eligibility criteria comparable to the development study were associated with higher c-statistics compared with narrower criteria (difference in logit c-statistic 0.21(95% CI 0.07 to 0.35), similar to an increase from 0.70 to 0.74). Using a case-control design was associated with higher OE ratios, compared with using data from a cohort (difference in log OE ratio 0.97(95% CI 0.38 to 1.55), similar to an increase in OE ratio from 1.00 to 2.63).ConclusionsVariation in performance of prognostic models across studies is mainly associated with variation in case-mix, study designs, outcome definitions and measurement methods and predictor substitution. Researchers developing and validating prognostic models should realise the potential influence of these study characteristics on the predictive performance of prognostic models.

Sign in / Sign up

Export Citation Format

Share Document