A comparison of linear and orthogonal regression analysis for local INR determination in ECAA coagulometer studies

International sensitivity index calibrations based on the W.H.O. recommended method depend on orthogonal regression analysis. As this is not readily available in statistical packages, comparison has been made with simple linear regression analysis in a study of coagulometer effects on the International Normalized Ratio (INR) at 155 European centres. Sets of seven lyophilized normal and 20 lyophilized artificially depleted abnormal plasmas were provided with five coumarin test plasmas and two European Concerted Action on Anticoagulation reference thromboplastins (low International Sensitivity Index (ISI) human and high ISI rabbit). Local ISI based on the artificially depleted lyophilized plasmas using conventional orthogonal regression gave good correction for local coagulometer effects on the human reagent and minimal correction with the rabbit reagent INR. Results were considerably worse after attempts at correction using calibration based on linear regression analysis with both reagents. The results indicate that calibration of coagulometer prothrombin time systems using simple linear regression is not appropriate.     

Benign overfitting in linear regression

The phenomenon of benign overfitting is one of the key mysteries uncovered by deep learning methodology: deep neural networks seem to predict well, even with a perfect fit to noisy training data. Motivated by this phenomenon, we consider when a perfect fit to training data in linear regression is compatible with accurate prediction. We give a characterization of linear regression problems for which the minimum norm interpolating prediction rule has near-optimal prediction accuracy. The characterization is in terms of two notions of the effective rank of the data covariance. It shows that overparameterization is essential for benign overfitting in this setting: the number of directions in parameter space that are unimportant for prediction must significantly exceed the sample size. By studying examples of data covariance properties that this characterization shows are required for benign overfitting, we find an important role for finite-dimensional data: the accuracy of the minimum norm interpolating prediction rule approaches the best possible accuracy for a much narrower range of properties of the data distribution when the data lie in an infinite-dimensional space vs. when the data lie in a finite-dimensional space with dimension that grows faster than the sample size.     

Interpretation of commonly used statistical regression models

A review of some regression models commonly used in respiratory health applications is provided in this article. Simple linear regression, multiple linear regression, logistic regression and ordinal logistic regression are considered. The focus of this article is on the interpretation of the regression coefficients of each model, which are illustrated through the application of these models to a respiratory health research study.     

ICSH guidance for INR and D-dimer testing using point of care testing in primary care

This guideline has been written on behalf of the International Council for Standardisation in Haematology (ICSH) and focuses on two point of care haematology tests used within primary care, namely International Normalised Ratio (INR) and D-dimer. Primary care covers out of hospital settings and can include General Practice (GP), Pharmacy and other non-hospital settings (although these guidelines would also be applicable to hospital out-patient settings). The recommendations are based on published data in peer reviewed literature and expert opinion; they should supplement regional requirements, regulations or standards.     

Evaluation of the multiple linear regression method to monitor respiratory mechanics in ventilated neonates and young children.

A potentially useful method to monitor respiratory mechanics in artificially ventilated patients consists of analyzing the relationship between tracheal pressure (P), lung volume (V), and gas flow (V) by multiple linear regression (MLR) using a suitable model. Contrary to other methods, it does not require any particular flow waveform and, therefore, may be used with any ventilator. This approach was evaluated in three neonates and seven young children admitted into an intensive care unit for respiratory disorders of various etiologies. P and V were measured and digitized at a sampling rate of 40 Hz for periods of 20-48 s. After correction of P for the non-linear resistance of the endotracheal tube, the data were first analyzed with the usual linear monoalveolar model: P = PO + E.V + R.V where E and R are total respiratory elastance and resistance, and PO is the static recoil pressure at end-expiration. A good fit of the model to the data was seen in five of ten children. PO, E, and R were reproducible within cycles, and consistent with the patient's age and condition; the data obtained with two ventilatory modes were highly correlated. In the five instances in which the simple model did not fit the data well, they were reanalyzed with more sophisticated models allowing for mechanical non-homogeneity or for non-linearity of R or E. While several models substantially improved the fit, physiologically meaningful results were only obtained when R was allowed to change with lung volume. We conclude that the MLR method is adequate to monitor respiratory mechanics, even when the usual model is inadequate.     

HCV extinction analysis in district Gujrat,Pakistan by using SARIMA and linear regression models

Background:To investigate the track of Gujrat, a District of Pakistan is very essential, either it follow-up World Health Organization (WHO) Hepatitis C Virus (HCV) elimination plan or not. This study aimed to find out HCV extinction analysis by time series forecast from District Gujrat, Pakistan.Methods:From January 1, 2016 to December 31, 2020 total n-5,111 numbers of HCV real-time polymerase chain reaction (RT-PCR) tests were performed in Gujrat. For extinction analysis we used 2 different models, the first model was seasonal auto-regressive integrated moving average (SARIMA) and the second linear regression (LR) model. First, we fitted both models then these fitted and valid models were used to predict future HCV percentage in District Gujrat.Results:In District Gujrat, the men HCV infected ratio is high with a higher viral load as compared with women, from year 2016 to 2020 male to female ratio was (53.75:53.19), (45.67:43.84), (39.67:39.36), (41.94:35.88), (37.70:31.38) respectively. HCV percentage is decreasing from 2016 to 2020 with an average of 4.98%. Our both fitted models SARIMAX (0,1,1)(0,1,1,6) at 95% confidence intervals and LR model Y = –0.379 X + 53.378 at 99% confidence intervals (P-value = .00) revealed that in June 2029 and in August 2027 respectively HCV percentage will be 0 from district Gujrat, Pakistan.Conclusions:This study concluded that both SARIMA and LR models showed an effective modeling process for forecasting yearly HCV incidence. District Gujrat, Punjab, Pakistan is on track to achieve the WHO HCV elimination plan, before 2030 HCV will be extinct from this region.     

A milestone in prediction of the coronary artery dimensions from the multiple linear regression equation

Bachground /aimCoronary artery imaging is one of the most commonly used diagnostic methods. We aimed to investigate whether there is a correlation between left main coronary artery (LMCA), left anterior descending artery (LAD) and left circumflex artery (LCx) artery dimensions in normal cases and a possibility to express the coronary dimensions by multiple linear equations.Materials and methodsImages of coronary angiograms of 925 normal cases selected from 3855 cases made up the study population (515 men and 410 women; age range, 30–75 years). The mean age of the patients was 55.50 ± 6.49 years. The mean body mass index was 24.79 ± 1.45 kg/m² (range, 31.30–21.26 kg/m²). The mean dimensions of LMCA, LAD and LCx were 4.18 ± 0.65 mm, 3.22 ± 0.63 mm and 3.07 ± 0.65 mm, respectively. Correlation between LMCA, LAD and LCx diameters was investigated. Multiple linear regression analysis was used to develop a model to elucidate the relationship between LMCA, LAD and LCx diameters.ResultsThere was a strong correlation between LMCA dimensions and LAD and LCx dimensions (r = 0.526**, p < 0.001* and r = 0.469**, p < 0.001*, respectively). The positive correlation indicated that a regression analysis can be carried out by incorporating the measurements. Coronary artery dimensions were gender specific.ConclusionThe present study explored the possibility of explaining the relationship with the LMCA and its branches by multiple linear equations, which may then be used to estimate the reference diameter of a stenosed coronary artery when the other two arteries are normal.     

Reliability of delayed INR determination: implications for decentralized anticoagulant care with off-site blood sampling

In order to develop decentralized anticoagulant care by off-site blood sampling and transport of samples to a centralized laboratory for International Normalized Ratio (INR) determination we have performed a direct comparative study of INR stability. Analysis was performed daily for 5 d using nine thromboplastins. The overall mean difference of INR after 3 d was only 0.05 INR units for samples with a therapeutic INR. After 5 d there was a mean difference of 0.11 INR units with 'non-Manchester' reagents and 0.44 INR units with 'Manchester' reagents. With over-anticoagulated samples mean differences of 0.55–0.72 INR units were observed after 3 d and 1.16–2.46 INR units after 5 d. Although there was some variation in stability of results with different thromboplastins, the difference over time with each thromboplastin was much less than the difference between thromboplastins.
In conclusion, there is no clinically significant change in INR when analysis is delayed for up to 3 d. Off-site blood sampling can accommodate a large increase in patient workload without a major revenue increase in primary care and with continued total quality management and central expert advice.     

Reliability of delayed prothrombin time INR determinations in a central laboratory using off‐site blood sampling

A major concern of centralized anticoagulant measurements with off‐site sampling is the reliability of international normalized ratio (INR) determinations on blood that may have been taken from the patient hours before the analysis. We compared INR differences in the blood of patients receiving oral anticoagulants after 24 h storage in four conditions: centrifuged at room temperature, centrifuged at 4°C, uncentrifuged at room temperature and uncentrifuged at 4°C. The INR of centrifuged and uncentrifuged blood left at room temperature for 24 h consistently increased by 6% and, after adjustment, there were no misclassifications in the assessment of the adequacy of anticoagulant treatment. Inconsistent changes were noted in tests of refrigerated centrifuged blood. We conclude that storage of blood at room temperature for 24 h results in a consistent prolongation of the prothrombin time, which after correction can reliably be used to adjust the dose of oral anticoagulants.     

Monitoring Therapy with Vitamin K Antagonists in Patients with Lupus Anticoagulant: Effect on Different Tests for INR Determination

Background: Lupus anticoagulant (antiphospholipid antibodies) is associated with venous and arterial thrombosis in patients with and without autoimmune disorders. Vitamin K antagonists are the treatment of choice in patients with thrombosis, of which the dose is titrated by INR monitoring. Several recent reports suggest that the presence of the lupus anticoagulant disturbs the INR test and may lead to unreliable results with a large variation in INR values, dependent on the reagents used.Methods: We studied 11 lupus anticoagulant positive patients and 11 lupus anticoagulant negative patients, all using vitamin K antagonists. The INR value was determined using seven different tests and the variation in INR values was compared between the two groups. The amidolytic Factor X levels were used as an phospholipid independent measure for intensity of warfarin therapy. Factor VII and X activity were measured to assess the stability of warfarin therapy.Results: The variation of the results with different INR tests within one patient was minimal and comparable in the two groups for INR's in the therapeutic range. The coefficient of variation for the cases and control group was 10.43 and 9.35, respectively. Variation in both groups increased at supratherapeutic levels of anticoagulation and when the anticoagulation was unstable (measured with Factor X/Factor VII ratio). The relationship between INR values and Factor X analysis revealed no influence of the lupus anticoagulant.Conclusions: In this study, lupus anticoagulant antibodies do not disturb INR laboratory tests. Differences in INR measurements are seen in patients with a high intensity of anticoagulation and in patients who either just started or in whom no stable anticoagulation has been achieved.Abbreviated Abstract. This study investigates the influence of lupus anticoagulant on INR determination tests in patients treated with warfarin. Eleven cases and eleven lupus anticoagulant negative control patients, also on warfarin therapy, were included. Seven INR results per patient were obtained using different laboratory tests. A factor X assay was performed to obtain an independent measure for the intensity of warfarin therapy.The variation of INR results between the cases and controls revealed no difference in these groups. In addition, the relationship between INR values and Factor X analysis indicated no influence of the lupus anticoagulant. What was observed was an increased difference in INR values in patients with a high intensity of anticoagulation and in patients who either just started or in whom no stable anticoagulation has been achieved     

A capillary whole blood method for measuring the INR

This study describes a method of measuring the INR on native whole blood capillary samples using Innovin recombinant thromboplastin. Modification of the reagent was necessary to compensate for the nonoptimal level of calcium in the sample/reagent mixture. Ninety-five percent of results obtained by the capillary blood method were no more than 0.42 INR higher or 0.38 INR lower than the venous blood method. The effect of changes in haematocrit was minimal. Significant differences in results were found between the Innovin and Thrombotest capillary blood methods. Provided the reagent was properly stored, there was no reagent drift and satisfactory results were obtained on samples supplied by UKNEQAS (coagulation) from previous trials. The method described is a convenient, simple and accurate method of measuring the INR using native capillary whole blood and Innovin recombinant thromboplastin.     

A novel reliability-based regression model for medical modeling and forecasting

Background and aimsIn recent decades, modeling and forecasting have played a significant role in the diagnosis and treatment of different diseases. Various forecasting models have been developed to improve data-based decision-making processes in medical systems. Although these models differ in many aspects, they all originate from the assumption that more generalizable results are achieved by more accurate models. This means that accuracy is considered as the only prominent feature to evaluate the generalizability of forecasting models. On the other side, due to the changeable medical situations and even changeable models' results, making stable and reliable performance is necessary to adopt appropriate medical decisions. Hence, reliability and stability of models' performance is another effective factor on the model's generalizability that should be taken into consideration in developing medical forecasting models.MethodsIn this paper, a new reliability-based forecasting approach is developed to address this gap and achieve more consistent performance in making medical predictions. The proposed approach is implemented on the classic regression model which is a common accuracy-based statistical method in medical fields. To evaluate the effectiveness of the proposed model, it has been performed by using two medical benchmark datasets from UCI and obtained results are compared with the classic regression model.ResultsEmpirical results show that the proposed model has outperformed the classic regression model in terms of error criteria such as MSE and MAE. So, the presented model can be utilized as an appropriate alternative for the traditional regression model in making effective medical decisions.ConclusionsBased on the obtained results, the proposed model can be an appropriate alternative for traditional multiple linear regression for modeling in real-world applications, especially when more generalization and/or more reliability is needed.     

Comparison of three statistical methods for analysis of fall predictors in people with dementia: Negative binomial regression (NBR), regression tree (RT), and partial least squares regression (PLSR)

Searching for background factors associated with falls in people with dementia is difficult because the population is heterogeneous. The aim of this study was to compare the efficacies of three statistical methods for analysis of fall predictors in people with dementia. NBR, RT and PLSR analyses were compared. Data used for the comparison were from a prospective cohort study of 192 patients at a psychogeriatric ward, specializing in patients with cognitive impairment and related behavioral and psychological symptoms. Seventy-eight of these patients fell a total of 238 times. PLSR and RT analyses are directed at finding patterns among predictor variables related to outcome, whereas an NBR model is directed at finding predictor variables that, independent of other variables, are related to the outcome. The NBR analysis explained an additional 10–15% variation compared with the PLSR and RT analyses. The results of PLSR and RT show a similar plausible pattern of predictor variables. However, none of these techniques appears to be sufficient in itself. In order to gain patterns of explanatory variables, RT would be a good complement to NBR for analysis of fall predictors.