Risk factors and prediction of hypoglycaemia using the Hypo-RESOLVE cohort: a secondary analysis of pooled data from insulin clinical trials.

AIMS/HYPOTHESIS: The objective of the Hypoglycaemia REdefining SOLutions for better liVES (Hypo-RESOLVE) project is to use a dataset of pooled clinical trials across pharmaceutical and device companies in people with type 1 or type 2 diabetes to examine factors associated with incident hypoglycaemia events and to quantify the prediction of these events. METHODS: Data from 90 trials with 46,254 participants were pooled. Analyses were done for type 1 and type 2 diabetes separately. Poisson mixed models, adjusted for age, sex, diabetes duration and trial identifier were fitted to assess the association of clinical variables with hypoglycaemia event counts. Tree-based gradient-boosting algorithms (XGBoost) were fitted using training data and their predictive performance in terms of area under the receiver operating characteristic curve (AUC) evaluated on test data. Baseline models including age, sex and diabetes duration were compared with models that further included a score of hypoglycaemia in the first 6 weeks from study entry, and full models that included further clinical variables. The relative predictive importance of each covariate was assessed using XGBoost's importance procedure. Prediction across the entire trial duration for each trial (mean of 34.8 weeks for type 1 diabetes and 25.3 weeks for type 2 diabetes) was assessed. RESULTS: For both type 1 and type 2 diabetes, variables associated with more frequent hypoglycaemia included female sex, white ethnicity, longer diabetes duration, treatment with human as opposed to analogue-only insulin, higher glucose variability, higher score for hypoglycaemia across the 6 week baseline period, lower BP, lower lipid levels and treatment with psychoactive drugs. Prediction of any hypoglycaemia event of any severity was greater than prediction of hypoglycaemia requiring assistance (level 3 hypoglycaemia), for which events were sparser. For prediction of level 1 or worse hypoglycaemia during the whole follow-up period, the AUC was 0.835 (95% CI 0.826, 0.844) in type 1 diabetes and 0.840 (95% CI 0.831, 0.848) in type 2 diabetes. For level 3 hypoglycaemia, the AUC was lower at 0.689 (95% CI 0.667, 0.712) for type 1 diabetes and 0.705 (95% CI 0.662, 0.748) for type 2 diabetes. Compared with the baseline models, almost all the improvement in prediction could be captured by the individual's hypoglycaemia history, glucose variability and blood glucose over a 6 week baseline period. CONCLUSIONS/INTERPRETATION: Although hypoglycaemia rates show large variation according to sociodemographic and clinical characteristics and treatment history, looking at a 6 week period of hypoglycaemia events and glucose measurements predicts future hypoglycaemia risk.

RadSigBench: a framework for benchmarking functional genomics signatures of cancer cell radiosensitivity.

Multiple transcriptomic predictors of tumour cell radiosensitivity (RS) have been proposed, but they have not been benchmarked against one another or to control models. To address this, we present RadSigBench, a comprehensive benchmarking framework for RS signatures. The approach compares candidate models to those developed from randomly resampled control signatures and from cellular processes integral to the radiation response. Robust evaluation of signature accuracy, both overall and for individual tissues, is performed. The NCI60 and Cancer Cell Line Encyclopaedia datasets are integrated into our workflow. Prediction of two measures of RS is assessed: survival fraction after 2 Gy and mean inactivation dose. We apply the RadSigBench framework to seven prominent published signatures of radiation sensitivity and test for equivalence to control signatures. The mean out-of-sample R2 for the published models on test data was very poor at 0.01 (range: -0.05 to 0.09) for Cancer Cell Line Encyclopedia and 0.00 (range: -0.19 to 0.19) in the NCI60 data. The accuracy of both published and cellular process signatures investigated was equivalent to the resampled controls, suggesting that these signatures contain limited radiation-specific information. Enhanced modelling strategies are needed for effective prediction of intrinsic RS to inform clinical treatment regimes. We make recommendations for methodological improvements, for example the inclusion of perturbation data, multiomics, advanced machine learning and mechanistic modelling. Our validation framework provides for robust performance assessment of ongoing developments in intrinsic RS prediction.

Case-control study developing Scottish Epilepsy Deaths Study Score to predict epilepsy-related death.

This study aimed to develop a risk prediction model for epilepsy-related death in adults. In this age- and sex-matched case-control study, we compared adults (aged ≥16 years) who had epilepsy-related death between 2009 and 2016 to living adults with epilepsy in Scotland. Cases were identified from validated administrative national datasets linked to mortality records. ICD-10 cause-of-death coding was used to define epilepsy-related death. Controls were recruited from a research database and epilepsy clinics. Clinical data from medical records were abstracted and used to undertake univariable and multivariable conditional logistic regression to develop a risk prediction model consisting of four variables chosen a priori. A weighted sum of the factors present was taken to create a risk index-the Scottish Epilepsy Deaths Study Score. Odds ratios were estimated with 95% confidence intervals (CIs). Here, 224 deceased cases (mean age 48 years, 114 male) and 224 matched living controls were compared. In univariable analysis, predictors of epilepsy-related death were recent epilepsy-related accident and emergency attendance (odds ratio 5.1, 95% CI 3.2-8.3), living in deprived areas (odds ratio 2.5, 95% CI 1.6-4.0), developmental epilepsy (odds ratio 3.1, 95% CI 1.7-5.7), raised Charlson Comorbidity Index score (odds ratio 2.5, 95% CI 1.2-5.2), alcohol abuse (odds ratio 4.4, 95% CI 2.2-9.2), absent recent neurology review (odds ratio 3.8, 95% CI 2.4-6.1) and generalized epilepsy (odds ratio 1.9, 95% CI 1.2-3.0). Scottish Epilepsy Deaths Study Score model variables were derived from the first four listed before, with Charlson Comorbidity Index ≥2 given 1 point, living in the two most deprived areas given 2 points, having an inherited or congenital aetiology or risk factor for developing epilepsy given 2 points and recent epilepsy-related accident and emergency attendance given 3 points. Compared to having a Scottish Epilepsy Deaths Study Score of 0, those with a Scottish Epilepsy Deaths Study Score of 1 remained low risk, with odds ratio 1.6 (95% CI 0.5-4.8). Those with a Scottish Epilepsy Deaths Study Score of 2-3 had moderate risk, with odds ratio 2.8 (95% CI 1.3-6.2). Those with a Scottish Epilepsy Deaths Study Score of 4-5 and 6-8 were high risk, with odds ratio 14.4 (95% CI 5.9-35.2) and 24.0 (95% CI 8.1-71.2), respectively. The Scottish Epilepsy Deaths Study Score may be a helpful tool for identifying adults at high risk of epilepsy-related death and requires external validation.

The unity and diversity of verbal and visuospatial creativity: Dynamic changes in hemispheric lateralisation.

The investigation of similarities and differences in the mechanisms of verbal and visuospatial creative thinking has long been a controversial topic. Prior studies found that visuospatial creativity was primarily supported by the right hemisphere, whereas verbal creativity relied on the interaction between both hemispheres. However, creative thinking also involves abundant dynamic features that may have been ignored in the previous static view. Recently, a new method has been developed that measures hemispheric laterality from a dynamic perspective, providing new insight into the exploration of creative thinking. In the present study, dynamic lateralisation index was calculated with resting-state fMRI data. We combined the dynamic lateralisation index with sparse canonical correlation analysis to examine similarities and differences in the mechanisms of verbal and visuospatial creativity. Our results showed that the laterality reversal of the default mode network, fronto-parietal network, cingulo-opercular network and visual network contributed significantly to both verbal and visuospatial creativity and consequently could be considered the common neural mechanisms shared by these creative modes. In addition, we found that verbal creativity relied more on the language network, while visuospatial creativity relied more on the somatomotor network, which can be considered a difference in their mechanism. Collectively, these findings indicated that verbal and visuospatial creativity may have similar mechanisms to support the basic creative thinking process and different mechanisms to adapt to the specific task conditions. These findings may have significant implications for our understanding of the neural mechanisms of different types of creative thinking.

Predicting prognosis for adults with depression using individual symptom data: a comparison of modelling approaches.

BACKGROUND: This study aimed to develop, validate and compare the performance of models predicting post-treatment outcomes for depressed adults based on pre-treatment data. METHODS: Individual patient data from all six eligible randomised controlled trials were used to develop (k = 3, n = 1722) and test (k = 3, n = 918) nine models. Predictors included depressive and anxiety symptoms, social support, life events and alcohol use. Weighted sum scores were developed using coefficient weights derived from network centrality statistics (models 1-3) and factor loadings from a confirmatory factor analysis (model 4). Unweighted sum score models were tested using elastic net regularised (ENR) and ordinary least squares (OLS) regression (models 5 and 6). Individual items were then included in ENR and OLS (models 7 and 8). All models were compared to one another and to a null model (mean post-baseline Beck Depression Inventory Second Edition (BDI-II) score in the training data: model 9). Primary outcome: BDI-II scores at 3-4 months. RESULTS: Models 1-7 all outperformed the null model and model 8. Model performance was very similar across models 1-6, meaning that differential weights applied to the baseline sum scores had little impact. CONCLUSIONS: Any of the modelling techniques (models 1-7) could be used to inform prognostic predictions for depressed adults with differences in the proportions of patients reaching remission based on the predicted severity of depressive symptoms post-treatment. However, the majority of variance in prognosis remained unexplained. It may be necessary to include a broader range of biopsychosocial variables to better adjudicate between competing models, and to derive models with greater clinical utility for treatment-seeking adults with depression.

Measuring the performance of prediction models to personalize treatment choice.

When data are available from individual patients receiving either a treatment or a control intervention in a randomized trial, various statistical and machine learning methods can be used to develop models for predicting future outcomes under the two conditions, and thus to predict treatment effect at the patient level. These predictions can subsequently guide personalized treatment choices. Although several methods for validating prediction models are available, little attention has been given to measuring the performance of predictions of personalized treatment effect. In this article, we propose a range of measures that can be used to this end. We start by defining two dimensions of model accuracy for treatment effects, for a single outcome: discrimination for benefit and calibration for benefit. We then amalgamate these two dimensions into an additional concept, decision accuracy, which quantifies the model's ability to identify patients for whom the benefit from treatment exceeds a given threshold. Subsequently, we propose a series of performance measures related to these dimensions and discuss estimating procedures, focusing on randomized data. Our methods are applicable for continuous or binary outcomes, for any type of prediction model, as long as it uses baseline covariates to predict outcomes under treatment and control. We illustrate all methods using two simulated datasets and a real dataset from a trial in depression. We implement all methods in the R package predieval. Results suggest that the proposed measures can be useful in evaluating and comparing the performance of competing models in predicting individualized treatment effect.

Using metabolomics to predict severe traumatic brain injury outcome (GOSE) at 3 and 12 months.

BACKGROUND: Prognostication is very important to clinicians and families during the early management of severe traumatic brain injury (sTBI), however, there are no gold standard biomarkers to determine prognosis in sTBI. As has been demonstrated in several diseases, early measurement of serum metabolomic profiles can be used as sensitive and specific biomarkers to predict outcomes. METHODS: We prospectively enrolled 59 adults with sTBI (Glasgow coma scale, GCS ≤ 8) in a multicenter Canadian TBI (CanTBI) study. Serum samples were drawn for metabolomic profiling on the 1st and 4th days following injury. The Glasgow outcome scale extended (GOSE) was collected at 3- and 12-months post-injury. Targeted direct infusion liquid chromatography-tandem mass spectrometry (DI/LC-MS/MS) and untargeted proton nuclear magnetic resonance spectroscopy (1H-NMR) were used to profile serum metabolites. Multivariate analysis was used to determine the association between serum metabolomics and GOSE, dichotomized into favorable (GOSE 5-8) and unfavorable (GOSE 1-4), outcomes. RESULTS: Serum metabolic profiles on days 1 and 4 post-injury were highly predictive (Q2 > 0.4-0.5) and highly accurate (AUC > 0.99) to predict GOSE outcome at 3- and 12-months post-injury and mortality at 3 months. The metabolic profiles on day 4 were more predictive (Q2 > 0.55) than those measured on day 1 post-injury. Unfavorable outcomes were associated with considerable metabolite changes from day 1 to day 4 compared to favorable outcomes. Increased lysophosphatidylcholines, acylcarnitines, energy-related metabolites (glucose, lactate), aromatic amino acids, and glutamate were associated with poor outcomes and mortality. DISCUSSION: Metabolomic profiles were strongly associated with the prognosis of GOSE outcome at 3 and 12 months and mortality following sTBI in adults. The metabolic phenotypes on day 4 post-injury were more predictive and significant for predicting the sTBI outcome compared to the day 1 sample. This may reflect the larger contribution of secondary brain injury (day 4) to sTBI outcome. Patients with unfavorable outcomes demonstrated more metabolite changes from day 1 to day 4 post-injury. These findings highlighted increased concentration of neurobiomarkers such as N-acetylaspartate (NAA) and tyrosine, decreased concentrations of ketone bodies, and decreased urea cycle metabolites on day 4 presenting potential metabolites to predict the outcome. The current findings strongly support the use of serum metabolomics, that are shown to be better than clinical data, in determining prognosis in adults with sTBI in the early days post-injury. Our findings, however, require validation in a larger cohort of adults with sTBI to be used for clinical practice.

Decision curve analysis confirms higher clinical utility of multi-domain versus single-domain prediction models in patients with open abdomen treatment for peritonitis.

BACKGROUND: Prediction modelling increasingly becomes an important risk assessment tool in perioperative systems approaches, e.g. in complex patients with open abdomen treatment for peritonitis. In this population, combining predictors from multiple medical domains (i.e. demographical, physiological and surgical variables) outperforms the prediction capabilities of single-domain prediction models. However, the benefit of these prediction models for clinical decision-making remains to be investigated. We therefore examined the clinical utility of mortality prediction models in patients suffering from peritonitis with a decision curve analysis. METHODS: In this secondary analysis of a large dataset, a traditional logistic regression approach, three machine learning methods and a stacked ensemble were employed to examine the predictive capability of demographic, physiological and surgical variables in predicting mortality under open abdomen treatment for peritonitis. Calibration was examined with calibration belts and predictive performance was assessed with the area both under the receiver operating characteristic curve (AUROC) and under the precision recall curve (AUPRC) and with the Brier Score. Clinical utility of the prediction models was examined by means of a decision curve analysis (DCA) within a treatment threshold range of interest of 0-30%, where threshold probabilities are traditionally defined as the minimum probability of disease at which further intervention would be warranted. RESULTS: Machine learning methods supported available evidence of a higher prediction performance of a multi- versus single-domain prediction models. Interestingly, their prediction performance was similar to a logistic regression model. The DCA demonstrated that the overall net benefit is largest for a multi-domain prediction model and that this benefit is larger compared to the default "treat all" strategy only for treatment threshold probabilities above about 10%. Importantly, the net benefit for low threshold probabilities is dominated by physiological predictors: surgical and demographics predictors provide only secondary decision-analytic benefit. CONCLUSIONS: DCA provides a valuable tool to compare single-domain and multi-domain prediction models and demonstrates overall higher decision-analytic value of the latter. Importantly, DCA provides a means to clinically differentiate the risks associated with each of these domains in more depth than with traditional performance metrics and highlighted the importance of physiological predictors for conservative intervention strategies for low treatment thresholds. Further, machine learning methods did not add significant benefit either in prediction performance or decision-analytic utility compared to logistic regression in these data.

A probabilistic decision support tool for prediction and management of rainfall-related poor water quality events for a drinking water treatment plant.

A data-driven Bayesian Network (BN) model was developed for a large Australian drinking water treatment plant, whose raw water comes from a river into which a number of upstream dams outflow water and smaller tributaries flow. During wet weather events, the spatial distribution of rainfall has a crucial role on the incoming raw water quality, as runoff from specific sub-catchments usually causes significant turbidity and conductivity issues, as opposed to larger dam outflows which have typically better water quality. The BN relies on a conceptual model developed following expert consultation, as well as a combination of different types (e.g. water quality, flow, rainfall) and amount (e.g. high-frequency, daily, scarce depending on variable) of historical data. The validated model proved to have acceptable accuracy in predicting the probability of different incoming raw water quality ranges, and can be used to assess different scenarios (e.g. timing, flow) of dam water releases, for the purpose of achieving dilution of the tributary's poor-quality water and mitigate related drinking water treatment challenges.

Integrating human footprint with ensemble modelling identifies priority habitats for conservation: a case study in the distributional range of Arnebia euchroma, a vulnerable species.

Climate change-driven rapid alteration of ecosystems globally is further complicated by growing anthropogenic pressures, especially in the ecologically sensitive mountainous regions. However, these two major drivers of change have largely been considered separately in species distribution models, thus compromising their reliability. Here, we integrated ensemble modelling with the human pressure index for predicting distribution and mapping priority regions across a whole range of occurrences for vulnerable species, Arnebia euchroma. Our results identified 3.08% of the study area as 'highly suitable', 2.45% as 'moderately suitable', and 94.45% as 'not suitable' or 'least suitable'. Compared to current climatic conditions, future RCP scenarios of 2050 and 2070 showed a significant loss in habitat suitability and a slight shift in the distribution pattern of the target species. By excluding the high-pressure areas of the human footprint from the predicted suitable habitats, we were able to identify the unique areas (70% of the predicted suitable area) that need special attention for conservation and restoration. Such models, if well implemented, may play a pivotal role in achieving the effective targets under the aegis of the current UN decade on ecological restoration (2021-2030) in accordance with SDG 15.4.

New directions in research on childhood adversity.

Childhood adversities are major preventable risk factors for poor mental and physical health. Scientific advances in this area are not matched by clinical gains for affected individuals. We reflect on novel research directions that could accelerate clinical impact.

Seek COVER: using a disease proxy to rapidly develop and validate a personalized risk calculator for COVID-19 outcomes in an international network.

BACKGROUND: We investigated whether we could use influenza data to develop prediction models for COVID-19 to increase the speed at which prediction models can reliably be developed and validated early in a pandemic. We developed COVID-19 Estimated Risk (COVER) scores that quantify a patient's risk of hospital admission with pneumonia (COVER-H), hospitalization with pneumonia requiring intensive services or death (COVER-I), or fatality (COVER-F) in the 30-days following COVID-19 diagnosis using historical data from patients with influenza or flu-like symptoms and tested this in COVID-19 patients. METHODS: We analyzed a federated network of electronic medical records and administrative claims data from 14 data sources and 6 countries containing data collected on or before 4/27/2020. We used a 2-step process to develop 3 scores using historical data from patients with influenza or flu-like symptoms any time prior to 2020. The first step was to create a data-driven model using LASSO regularized logistic regression, the covariates of which were used to develop aggregate covariates for the second step where the COVER scores were developed using a smaller set of features. These 3 COVER scores were then externally validated on patients with 1) influenza or flu-like symptoms and 2) confirmed or suspected COVID-19 diagnosis across 5 databases from South Korea, Spain, and the United States. Outcomes included i) hospitalization with pneumonia, ii) hospitalization with pneumonia requiring intensive services or death, and iii) death in the 30 days after index date. RESULTS: Overall, 44,507 COVID-19 patients were included for model validation. We identified 7 predictors (history of cancer, chronic obstructive pulmonary disease, diabetes, heart disease, hypertension, hyperlipidemia, kidney disease) which combined with age and sex discriminated which patients would experience any of our three outcomes. The models achieved good performance in influenza and COVID-19 cohorts. For COVID-19 the AUC ranges were, COVER-H: 0.69-0.81, COVER-I: 0.73-0.91, and COVER-F: 0.72-0.90. Calibration varied across the validations with some of the COVID-19 validations being less well calibrated than the influenza validations. CONCLUSIONS: This research demonstrated the utility of using a proxy disease to develop a prediction model. The 3 COVER models with 9-predictors that were developed using influenza data perform well for COVID-19 patients for predicting hospitalization, intensive services, and fatality. The scores showed good discriminatory performance which transferred well to the COVID-19 population. There was some miscalibration in the COVID-19 validations, which is potentially due to the difference in symptom severity between the two diseases. A possible solution for this is to recalibrate the models in each location before use.

Prediction of acute appendicitis among patients with undifferentiated abdominal pain at emergency department.

BACKGROUND: Early screening and accurately identifying Acute Appendicitis (AA) among patients with undifferentiated symptoms associated with appendicitis during their emergency visit will improve patient safety and health care quality. The aim of the study was to compare models that predict AA among patients with undifferentiated symptoms at emergency visits using both structured data and free-text data from a national survey. METHODS: We performed a secondary data analysis on the 2005-2017 United States National Hospital Ambulatory Medical Care Survey (NHAMCS) data to estimate the association between emergency department (ED) patients with the diagnosis of AA, and the demographic and clinical factors present at ED visits during a patient's ED stay. We used binary logistic regression (LR) and random forest (RF) models incorporating natural language processing (NLP) to predict AA diagnosis among patients with undifferentiated symptoms. RESULTS: Among the 40,441 ED patients with assigned International Classification of Diseases (ICD) codes of AA and appendicitis-related symptoms between 2005 and 2017, 655 adults (2.3%) and 256 children (2.2%) had AA. For the LR model identifying AA diagnosis among adult ED patients, the c-statistic was 0.72 (95% CI: 0.69-0.75) for structured variables only, 0.72 (95% CI: 0.69-0.75) for unstructured variables only, and 0.78 (95% CI: 0.76-0.80) when including both structured and unstructured variables. For the LR model identifying AA diagnosis among pediatric ED patients, the c-statistic was 0.84 (95% CI: 0.79-0.89) for including structured variables only, 0.78 (95% CI: 0.72-0.84) for unstructured variables, and 0.87 (95% CI: 0.83-0.91) when including both structured and unstructured variables. The RF method showed similar c-statistic to the corresponding LR model. CONCLUSIONS: We developed predictive models that can predict the AA diagnosis for adult and pediatric ED patients, and the predictive accuracy was improved with the inclusion of NLP elements and approaches.

Factors associated with accessing long-term adult social care in people aged 75 and over: a retrospective cohort study.

BACKGROUND: An ageing population and limited resources have put strain on state provision of adult social care (ASC) in England. With social care needs predicted to double over the next 20 years, there is a need for new approaches to inform service planning and development, including through predictive models of demand. OBJECTIVE: Describe risk factors for long-term ASC in two inner London boroughs and develop a risk prediction model for long-term ASC. METHODS: Pseudonymised person-level data from an integrated care dataset were analysed. We used multivariable logistic regression to model associations of demographic factors, and baseline aspects of health status and health service use, with accessing long-term ASC over 12 months. RESULTS: The cohort comprised 13,394 residents, aged ≥75 years with no prior history of ASC at baseline. Of these, 1.7% became ASC clients over 12 months. Residents were more likely to access ASC if they were older or living in areas with high socioeconomic deprivation. Those with preexisting mental health or neurological conditions, or more intense prior health service use during the baseline period, were also more likely to access ASC. A prognostic model derived from risk factors had limited predictive power. CONCLUSIONS: Our findings reinforce evidence on known risk factors for residents aged 75 or over, yet even with linked routinely collected health and social care data, it was not possible to make accurate predictions of long-term ASC use for individuals. We propose that a paradigm shift towards more relational, personalised approaches, is needed.

Retinal venular tortuosity and fractal dimension predict incident retinopathy in adults with type 2 diabetes: the Edinburgh Type 2 Diabetes Study.

AIMS/HYPOTHESIS: Our aim was to determine whether a range of prespecified retinal vessel traits were associated with incident diabetic retinopathy in adults with type 2 diabetes. METHODS: In the prospective observational cohort Edinburgh Type 2 Diabetes Study of 1066 adults with type 2 diabetes, aged 60-75 years at recruitment, 718 were free from diabetic retinopathy at baseline. Baseline retinal traits including vessel widths, tortuosity (curvature) and fractal dimensions (network complexity), were quantified using fundus camera images and semiautomated software, and analysed using logistic regression for their association with incident diabetic retinopathy over 10 years. RESULTS: The incidence of diabetic retinopathy was 11.4% (n = 82) over 10 years. After adjustment for a range of vascular and diabetes-related risk factors, both increased venular tortuosity (OR 1.51; 95% CI 1.15, 1.98; p = 0.003) and decreased fractal dimension (OR 0.75; 95% CI 0.58, 0.96; p = 0.025) were associated with incident retinopathy. There was no evidence of an association with arterial tortuosity, and associations between measurements of vessel widths and retinopathy lost statistical significance after adjustment for diabetes-related factors and vascular disease. Adding venular tortuosity to a model including established risk factors for diabetic retinopathy (HbA1c, BP and kidney function) improved the discriminative ability (C statistic increased from 0.624 to 0.640, p = 0.013), but no such benefit was found with fractal dimension. CONCLUSIONS/INTERPRETATION: Increased retinal venular tortuosity and decreased fractal dimension are associated with incident diabetic retinopathy, independent of classical risk factors. There is some evidence that venular tortuosity may be a useful biomarker to improve the predictive ability of models based on established retinopathy risk factors, and its inclusion in further risk prediction modelling is warranted.

Cause-specific mortality prediction in older residents of São Paulo, Brazil: a machine learning approach.

BACKGROUND: Populational ageing has been increasing in a remarkable rate in developing countries. In this scenario, preventive strategies could help to decrease the burden of higher demands for healthcare services. Machine learning algorithms have been increasingly applied for identifying priority candidates for preventive actions, presenting a better predictive performance than traditional parsimonious models. METHODS: Data were collected from the Health, Well Being and Aging (SABE) Study, a representative sample of older residents of São Paulo, Brazil. Machine learning algorithms were applied to predict death by diseases of respiratory system (DRS), diseases of circulatory system (DCS), neoplasms and other specific causes within 5 years, using socioeconomic, demographic and health features. The algorithms were trained in a random sample of 70% of subjects, and then tested in the other 30% unseen data. RESULTS: The outcome with highest predictive performance was death by DRS (AUC-ROC = 0.89), followed by the other specific causes (AUC-ROC = 0.87), DCS (AUC-ROC = 0.67) and neoplasms (AUC-ROC = 0.52). Among only the 25% of individuals with the highest predicted risk of mortality from DRS were included 100% of the actual cases. The machine learning algorithms with the highest predictive performance were light gradient boosted machine and extreme gradient boosting. CONCLUSION: The algorithms had a high predictive performance for DRS, but lower for DCS and neoplasms. Mortality prediction with machine learning can improve clinical decisions especially regarding targeted preventive measures for older individuals.

Have you had bleeding from your gums? Self-report to identify giNGival inflammation (The SING diagnostic accuracy and diagnostic model development study).

AIM: To assess the diagnostic performance of self-reported oral health questions and develop a diagnostic model with additional risk factors to predict clinical gingival inflammation in systemically healthy adults in the United Kingdom. METHODS: Gingival inflammation was measured by trained staff and defined as bleeding on probing (present if bleeding sites ≥ 30%). Sensitivity and specificity of self-reported questions were calculated; a diagnostic model to predict gingival inflammation was developed and its performance (calibration and discrimination) assessed. RESULTS: We included 2853 participants. Self-reported questions about bleeding gums had the best performance: the highest sensitivity was 0.73 (95% CI 0.70, 0.75) for a Likert item and the highest specificity 0.89 (95% CI 0.87, 0.90) for a binary question. The final diagnostic model included self-reported bleeding, oral health behaviour, smoking status, previous scale and polish received. Its area under the curve was 0.65 (95% CI 0.63-0.67). CONCLUSION: This is the largest assessment of diagnostic performance of self-reported oral health questions and the first diagnostic model developed to diagnose gingival inflammation. A self-reported bleeding question or our model could be used to rule in gingival inflammation since they showed good sensitivity, but are limited in identifying healthy individuals and should be externally validated.

Exploring the possibility of predicting human head hair greying from DNA using whole-exome and targeted NGS data.

BACKGROUND: Greying of the hair is an obvious sign of human aging. In addition to age, sex- and ancestry-specific patterns of hair greying are also observed and the progression of greying may be affected by environmental factors. However, little is known about the genetic control of this process. This study aimed to assess the potential of genetic data to predict hair greying in a population of nearly 1000 individuals from Poland. RESULTS: The study involved whole-exome sequencing followed by targeted analysis of 378 exome-wide and literature-based selected SNPs. For the selection of predictors, the minimum redundancy maximum relevance (mRMRe) method was used, and then two prediction models were developed. The models included age, sex and 13 unique SNPs. Two SNPs of the highest mRMRe score included whole-exome identified KIF1A rs59733750 and previously linked with hair loss FGF5 rs7680591. The model for greying vs. no greying prediction achieved accuracy of cross-validated AUC = 0.873. In the 3-grade classification cross-validated AUC equalled 0.864 for no greying, 0.791 for mild greying and 0.875 for severe greying. Although these values present fairly accurate prediction, most of the prediction information was brought by age alone. Genetic variants explained < 10% of hair greying variation and the impact of particular SNPs on prediction accuracy was found to be small. CONCLUSIONS: The rate of changes in human progressive traits shows inter-individual variation, therefore they are perceived as biomarkers of the biological age of the organism. The knowledge on the mechanisms underlying phenotypic aging can be of special interest to the medicine, cosmetics industry and forensics. Our study improves the knowledge on the genetics underlying hair greying processes, presents prototype models for prediction and proves hair greying being genetically a very complex trait. Finally, we propose a four-step approach based on genetic and epigenetic data analysis allowing for i) sex determination; ii) genetic ancestry inference; iii) greying-associated SNPs assignment and iv) epigenetic age estimation, all needed for a final prediction of greying.

Dynamic interventions to control COVID-19 pandemic: a multivariate prediction modelling study comparing 16 worldwide countries.

To date, non-pharmacological interventions (NPI) have been the mainstay for controlling the coronavirus disease-2019 (COVID-19) pandemic. While NPIs are effective in preventing health systems overload, these long-term measures are likely to have significant adverse economic consequences. Therefore, many countries are currently considering to lift the NPIs-increasing the likelihood of disease resurgence. In this regard, dynamic NPIs, with intervals of relaxed social distancing, may provide a more suitable alternative. However, the ideal frequency and duration of intermittent NPIs, and the ideal "break" when interventions can be temporarily relaxed, remain uncertain, especially in resource-poor settings. We employed a multivariate prediction model, based on up-to-date transmission and clinical parameters, to simulate outbreak trajectories in 16 countries, from diverse regions and economic categories. In each country, we then modelled the impacts on intensive care unit (ICU) admissions and deaths over an 18-month period for following scenarios: (1) no intervention, (2) consecutive cycles of mitigation measures followed by a relaxation period, and (3) consecutive cycles of suppression measures followed by a relaxation period. We defined these dynamic interventions based on reduction of the mean reproduction number during each cycle, assuming a basic reproduction number (R0) of 2.2 for no intervention, and subsequent effective reproduction numbers (R) of 0.8 and 0.5 for illustrative dynamic mitigation and suppression interventions, respectively. We found that dynamic cycles of 50-day mitigation followed by a 30-day relaxation reduced transmission, however, were unsuccessful in lowering ICU hospitalizations below manageable limits. By contrast, dynamic cycles of 50-day suppression followed by a 30-day relaxation kept the ICU demands below the national capacities. Additionally, we estimated that a significant number of new infections and deaths, especially in resource-poor countries, would be averted if these dynamic suppression measures were kept in place over an 18-month period. This multi-country analysis demonstrates that intermittent reductions of R below 1 through a potential combination of suppression interventions and relaxation can be an effective strategy for COVID-19 pandemic control. Such a "schedule" of social distancing might be particularly relevant to low-income countries, where a single, prolonged suppression intervention is unsustainable. Efficient implementation of dynamic suppression interventions, therefore, confers a pragmatic option to: (1) prevent critical care overload and deaths, (2) gain time to develop preventive and clinical measures, and (3) reduce economic hardship globally.

Predicting influenza-like illness-related emergency department visits by modelling spatio-temporal syndromic surveillance data.

Predicting the magnitude of the annual seasonal peak in influenza-like illness (ILI)-related emergency department (ED) visit volumes can inform the decision to open influenza care clinics (ICCs), which can mitigate pressure at the ED. Using ILI-related ED visit data from the Alberta Real Time Syndromic Surveillance Net for Edmonton, Alberta, Canada, we developed (training data, 1 August 2004-31 July 2008) and tested (testing data, 1 August 2008-19 February 2014) spatio-temporal statistical prediction models of daily ILI-related ED visits to estimate high visit volumes 3 days in advance. Our Main Model, based on a generalised linear mixed model with random intercept, incorporated prediction residuals over 14 days and captured increases in observed volume ahead of peaks. During seasonal influenza periods, our Main Model predicted volumes within ±30% of observed volumes for 67%-82% of high-volume days and within 0.3%-21% of observed seasonal peak volumes. Model predictions were not as successful during the 2009 H1N1 pandemic. Our model can provide early warning of increases in ILI-related ED visit volumes during seasonal influenza periods of differing intensities. These predictions may be used to support public health decisions, such as if and when to open ICCs, during seasonal influenza epidemics.