Prediction Of Glucose Concentration Research Articles

Comparison of multivariate data analysis techniques to improve glucose concentration prediction in mammalian cell cultivations by Raman spectroscopy

In-situ Raman spectroscopy is frequently applied to monitor and even control the glucose concentration of monoclonal antibody producing mammalian cell cultivations. Previous studies used the PLSR algorithm only, however other multivariate algorithms were applied successfully for different protein production processes. In this study, four mammalian cell cultivation runs were followed with Raman spectroscopy and the spectra were analysed quantitatively and qualitatively as well. The PCA analysis showed that one of the most dominant factors in the Raman spectra were the concentration of glucose, which strongly correlated with the score values of the eighth principal component. This observation further substantiated that Raman spectroscopy is an excellent tool for bioprocess monitoring and induced the test of the Multivariate Linear Regression (MLR), Principal Component Regression (PCR) and Partial Least Squares Regression (PLSR) algorithms, using the results of the PCA as one of the variable selection techniques, to determine the glucose concentration during cultivation. However, the novel variable selection technique of PCA correlation enhanced only the model accuracy when it was applied with MLR and only model robustness was increased when it was used with PCR and PLSR because the relatively strong Raman signal of glucose concentration seemed to be enough to build an accurate model on. Therefore, PLSR, the most advanced algorithm of the three, delivered the lowest 2.21 mM RMSEP but it was demonstrated that in certain cases PCR could also produce satisfactorily results.

Implementation of an integrating sphere for the enhancement of noninvasive glucose detection using quantum cascade laser spectroscopy

An integrating sphere is used to enhance the collection of backscattered light in a noninvasive glucose sensor based on quantum cascade laser spectroscopy. The sphere enhances signal stability by roughly an order of magnitude, allowing us to use a thermoelectrically (TE) cooled detector while maintaining comparable glucose prediction accuracy levels. Using a smaller TE-cooled detector reduces form factor, creating a mobile sensor. Principal component analysis has predicted principal components of spectra taken from human subjects that closely match the absorption peaks of glucose. These principal components are used as regressors in a linear regression algorithm to make glucose concentration predictions, over 75% of which are clinically accurate.

Model-fusion-based online glucose concentration predictions in people with type 1 diabetes

Accurate predictions of glucose concentrations are necessary to develop an artificial pancreas (AP) system for people with type 1 diabetes (T1D). In this work, a novel glucose forecasting paradigm based on a model fusion strategy is developed to accurately characterize the variability and transient dynamics of glycemic measurements. To this end, four different adaptive filters and a fusion mechanism are proposed for use in the online prediction of future glucose trajectories. The filter fusion mechanism is developed based on various prediction performance indexes to guide the overall output of the forecasting paradigm. The efficiency of the proposed model fusion based forecasting method is evaluated using simulated and clinical datasets, and the results demonstrate the capability and prediction accuracy of the data-based fusion filters, especially in the case of limited data availability. The model fusion framework may be used in the development of an AP system for glucose regulation in patients with T1D.

Data Based Prediction of Blood Glucose Concentrations Using Evolutionary Methods.

Predicting glucose values on the basis of insulin and food intakes is a difficult task that people with diabetes need to do daily. This is necessary as it is important to maintain glucose levels at appropriate values to avoid not only short-term, but also long-term complications of the illness. Artificial intelligence in general and machine learning techniques in particular have already lead to promising results in modeling and predicting glucose concentrations. In this work, several machine learning techniques are used for the modeling and prediction of glucose concentrations using as inputs the values measured by a continuous monitoring glucose system as well as also previous and estimated future carbohydrate intakes and insulin injections. In particular, we use the following four techniques: genetic programming, random forests, k-nearest neighbors, and grammatical evolution. We propose two new enhanced modeling algorithms for glucose prediction, namely (i) a variant of grammatical evolution which uses an optimized grammar, and (ii) a variant of tree-based genetic programming which uses a three-compartment model for carbohydrate and insulin dynamics. The predictors were trained and tested using data of ten patients from a public hospital in Spain. We analyze our experimental results using the Clarke error grid metric and see that 90% of the forecasts are correct (i.e., Clarke error categories A and B), but still even the best methods produce 5 to 10% of serious errors (category D) and approximately 0.5% of very serious errors (category E). We also propose an enhanced genetic programming algorithm that incorporates a three-compartment model into symbolic regression models to create smoothed time series of the original carbohydrate and insulin time series.

Frequency-range optimized preprocessing methods for quantitative analysis of glucose in blood serum from broadband dielectric spectra.

This paper presents the results of combining scatter correction and spectral derivation preprocessing methods with frequency-range optimization to improve the accuracy of blood glucose concentration prediction when using a Partial Least Square Regression model. Using broadband dielectric spectroscopy, absorption spectrums in the frequency range of 500 MHz to 50 GHz were gathered from blood serums. Partial Least Square Regression models were trained using data gathered from samples of varying glucose concentrations and temperatures, and the quality of the predictions were evaluated by performing leave-one-out cross validation. Potential improvements in prediction accuracy were assessed by finding the optimal frequency ranges of dielectric absorption spectrums through iteration, in addition to treatments using common preprocessing methods. The most effective combination of preprocessing method and its corresponding characteristic frequency range was determined from validation using several samples. Finally, among all the preprocessing methods and the frequency ranges explored, performing Savitzky-Golay filtering without derivations lowered the average root mean squared error amongst all the samples from 172 mg/dL without preprocessing, to 143 mg/dL. Additionally, by focusing on a specific frequency range, the root mean squared error dropped to 80.2 mg/dL.

Continuous Glucose Monitoring Sensors: Past, Present and Future Algorithmic Challenges.

Continuous glucose monitoring (CGM) sensors are portable devices that allow measuring and visualizing the glucose concentration in real time almost continuously for several days and are provided with hypo/hyperglycemic alerts and glucose trend information. CGM sensors have revolutionized Type 1 diabetes (T1D) management, improving glucose control when used adjunctively to self-monitoring blood glucose systems. Furthermore, CGM devices have stimulated the development of applications that were impossible to create without a continuous-time glucose signal, e.g., real-time predictive alerts of hypo/hyperglycemic episodes based on the prediction of future glucose concentration, automatic basal insulin attenuation methods for hypoglycemia prevention, and the artificial pancreas. However, CGM sensors’ lack of accuracy and reliability limited their usability in the clinical practice, calling upon the academic community for the development of suitable signal processing methods to improve CGM performance. The aim of this paper is to review the past and present algorithmic challenges of CGM sensors, to show how they have been tackled by our research group, and to identify the possible future ones.

Fluorometric In Situ Monitoring of an Escherichia coli Cell Factory with Cytosolic Expression of Human Glycosyltransferase GalNAcT2: Prospects and Limitations.

The glycosyltransferase HisDapGalNAcT2 is the key protein of the Escherichia coli (E. coli) SHuffle® T7 cell factory which was genetically engineered to allow glycosylation of a protein substrate in vivo. The specific activity of the glycosyltransferase requires time-intensive analytics, but is a critical process parameter. Therefore, it has to be monitored closely. This study evaluates fluorometric in situ monitoring as option to access this critical process parameter during complex E. coli fermentations. Partial least square regression (PLS) models were built based on the fluorometric data recorded during the EnPresso® B fermentations. Capable models for the prediction of glucose and acetate concentrations were built for these fermentations with rout mean squared errors for prediction (RMSEP) of 0.19 g·L−1 and 0.08 g·L−1, as well as for the prediction of the optical density (RMSEP 0.24). In situ monitoring of soluble enzyme to cell dry weight ratios (RMSEP 5.5 × 10−4 µg w/w) and specific activity of the glycosyltransferase (RMSEP 33.5 pmol·min−1·µg−1) proved to be challenging, since HisDapGalNAcT2 had to be extracted from the cells and purified. However, fluorescence spectroscopy, in combination with PLS modeling, proved to be feasible for in situ monitoring of complex expression systems.

Exploring the capabilities of fluorometric online monitoring on chinese hamster ovary cell cultivations producing a monoclonal antibody.

Online monitoring of Chinese hamster ovary fed-batch cell cultures via two-dimensional fluorescence spectroscopy (2DFS) was evaluated in this work. Particular attention was directed toward different process strategies regarding the use of nutrient-rich feed media and temperature shifts. These intentionally performed process manipulations broadened the variances in the obtained fluorescence spectra and this was suspected to hamper the generation of reliable soft sensors. Principal component analysis of the obtained fluorescence data showed that temperature shift and feeding strategy had a considerable impact on the fluorescence signals. Partial least square regression models were calculated for the prediction of glucose, lactate, monoclonal antibody (mAb), and viable cell concentrations (VCC). It was aimed to integrate all 2DFS datasets in the respective calibration models regardless of the process-strategy-dependent diversity. Contrary to the expectations, it was feasible to calibrate soft sensors for the online prediction of glucose (7 latent variables (LVs), Rcal2 = 0.97, rout mean squared error of prediction (RMSEP) = 1.1 g L-1 ), lactate (5 LV; Rcal2 = 0.96; RMSEP = 0.5 g L-1 ) and mAb concentrations (4 LV; Rcal2 = 0.99; RMSEP = 11.4 mg L-1 ). Feeding and temperature shifts had the highest impact on the VCC model (3 LV; Rcal2 = 0.94; RMSEP 3.8 × 105 mL-1 ), nevertheless the prediction of VCC from the fed-batch 2DFS data was feasible. The results strongly indicate that variances in the datasets due to the process strategy can be tolerated to some extent by the respective soft sensors. © 2016 American Institute of Chemical Engineers Biotechnol. Prog., 32:1592-1600, 2016.

Prediction of Glucose Concentration in a Glucose-Lactose Mixture Based on the Reflective Optical Power at Dual Probe Wavelengths

Department of Electronic Engineering, Kwangwoon University, 20 Kwangwoon-ro, Nowon-gu, Seoul 01897, South Korea(Received September 17, 2015 : revised November 11, 2015 : accepted December 8, 2015)An enzyme-free optical method is proposed for estimating high concentrations of glucose in a glucose-lactose mixture, based on a predictive equation that takes advantage of the reflective optical power observed at two discrete wavelengths. Compared to the conventional absorption spectroscopy method based on Beer’s Law, which is mainly valid for concentrations below hundreds of mg/dL, the proposed scheme, which relies on reflection signals, can be applied to measure higher glucose concentrations, of even several g/dL in a glucose-lactose mixture. Two probe wavelengths of 1160 and 1300 nm were selected to provide a linear relationship between the reflective power and pure glucose/lactose concentration, where the relevant linear coefficients were derived to complete the predictive equation. Glucose concentrations from 2 to 7 g/dL in a glucose-lactose mixture were efficiently estimated, using the established predictive equation based on monitored reflective powers. The standard error of prediction was 1.17 g/dL.Keywords : High glucose concentration, Correlation coefficient, Glucose-lactose mixture, Reflective optical powerOCIS codes : (280.4788) Optical sensing and sensors; (120.5700) Reflection; (330.1880) Detection; (350.4800) Optical standards and testing

Process analytical technology tools for perfusion cell culture

During cell cultivation processes for the production of biopharmaceuticals, good process performance and good product quality can be ensured by online monitoring of critical process parameters (e.g. temperature, pH, or dissolved oxygen). These data can be used in real‐time for process control, as suggested by the process analytical technology (PAT) initiative. Today, solutions for real‐time monitoring of parameters such as concentrations of cells, main nutrients, and metabolism by‐products are developing, but applications of these more complex tools in industrial settings are still limited. Here, we evaluated the use of dielectric spectroscopy (DS) and near‐infrared spectroscopy (NIRS) as PAT tools for a perfusion PER.C6® cultivation process. We showed that DS enabled predictions of viable cell density from the cultivation vessel, with a root mean square error of prediction (RMSEP) of 4.4% of the calibration range. Additionally, predictions of glucose and lactate concentrations from the cultivation vessel (RMSEP of 10 and 14%, respectively) and from the perfusion stream (RMSEP of 12 and 10%, respectively) were achieved with NIRS. We also showed that the perfusion stream offers great opportunities for noninvasive, yet frequent process monitoring. Accurate online monitoring of critical process parameters with PAT tools is the essential first step toward increased control of process output.

Online prediction of glucose concentration in type 1 diabetes using extreme learning machines.

We propose an online machine-learning solution to the problem of nonlinear glucose time series prediction in type 1 diabetes. Recently, extreme learning machine (ELM) has been proposed for training single hidden layer feed-forward neural networks. The high accuracy and fast learning speed of ELM drive us to investigate its applicability to the glucose prediction problem. Given that diabetes self-monitoring data are received sequentially, we focus on online sequential ELM (OS-ELM) and online sequential ELM kernels (KOS-ELM). A multivariate feature set is utilized concerning subcutaneous glucose, insulin therapy, carbohydrates intake and physical activity. The dataset comes from the continuous multi-day recordings of 15 type 1 patients in free-living conditions. Assuming stationarity and evaluating the performance of the proposed method by 10-fold cross- validation, KOS-ELM were found to perform better than OS-ELM in terms of prediction error, temporal gain and regularity of predictions for a 30-min prediction horizon.

A single-loop fiber attenuated total reflection sensor enhanced by silver nanoparticles for continuous glucose monitoring

Currently, the implantable enzyme electrode sensor is the only method for continuous glucose monitoring in clinical applications, but the significant drift caused by bioelectricity in body and the effect of electrochemical reactions under hypoxia reduce the accuracy of the glucose measurements. Therefore, finger-prick blood corrections are often required to calibrate enzyme-based glucose sensors several times each day. In this paper, we proposed an implantable miniaturized fiber Attenuated Total Reflection (ATR) sensor based on mid-infrared spectroscopy to overcome the two drawbacks. A single-loop structure of fiber ATR sensor was used to increase the effective optical length to enhance the sensitivity. By growing silver nanoparticles (AgNPs) on the cylindrical surface of the fiber ATR sensor with a chemical silver mirror method, the sensitivity of the infrared absorption signal was further enhanced. In the experiment, the fiber ATR sensor enhanced by AgNPs combined with a CO2 laser showed an intense absorption signal from glucose, with enhancement factors of two to six folds at different absorption wavenumbers compared with the bare fiber ATR sensors. The prediction of glucose concentration with high accuracy was achieved by using the five-variable partial least-squares model yielding a root-mean-square error of prediction as small as 4.45mg/dL.

Prediction of Glycemia Based on Diabetes Self-Monitoring Data

This paper deals with the application of self-monitoring diabetes data that are supplemented by the continuous glucose monitoring for the blood glucose concentration prediction. The short-term predictor is designed and evaluated on three different datasets. A diabetes-specific metrics is used to evaluate the predictors. Standard Least Squares identification as well as an alternative identification method with constraints is considered.

Lactate in cord blood and its relation to fetal gluconeogenesis in at term deliveries

BackgroundIn the human fetus, an increased lactate and glucose level can be anticipated when hypoxia and stress are present and is likely to be a function of both anaerobic metabolism and catecholamine-mediated glycogenolysis/glycolysis. AimWe assessed if measurement of lactate in cord artery blood after vaginal and cesarean delivery may predict glucose concentration. Study designUmbilical artery cord blood lactacidemia, acidemia, and glucose concentration was tested by ‘mini-lab’ Radiometer ABL90 FLEX analyzers (Radiometer®, Copenhagen, Denmark) after vaginal delivery (VD), spontaneous (n=493) and by vacuum extractor (n=41) or by cesarean delivery (CD), elective (n=120) and emergency (n=68) in at term, vigorous neonates delivered from March to December 2012 at the 2nd level maternity ward of Policlinico Abano Terme, Abano Terme (Italy). ResultsCord blood lactacidemia and glucose levels were significantly higher in VD by vacuum extractor than in all other groups (5.32±1.96mmol/L, p=0.050 and 103.6±30.5mg/dL, p<0.001, respectively) and significantly lower in elective CD group (1.77±0.99mmol/L, p<0.001 and 69.8±13.0mg/dL, p<0.001). The cord blood lactate concentration was significantly and positively correlated with glucose levels (r=0.434, p<0.001), but significantly and negatively correlated with pH (r=−0,662, p<0.001), NaHCO3− (r=−0,802, p<0.001), and base excess (BE) (r=−0,698, p<0.001). However, in multivariate linear regression analysis, only BE, PaCO2 and cord blood lactate were significant predictive variables (R2=0.410; p<0.001) of glucose levels at birth. ConclusionCord blood artery lactate and glucose concentration are significantly and positively correlated at birth in healthy, at term vaginally and cesarean delivered neonates, but BE is the best indicator of activated fetal gluconeogenesis.

The Utilization of Ultrasonic Standing Waves for Predicting Glucose Concentration Levels in Dextrose Mixed Intralipid Based Tissue Phantom

This paper represents the effect of amplitude modulated ultrasonic standing waves in determining the glucose concentration in dextrose mixed intralipid based tissue phantom. The intralipid exhibits biological tissue related light absorption, reflection, scattering properties. The intralipid act as a suitable substitute medium for tissue related lab based experimentations. The ultrasonic waves were produced here using two sinusoidal waves. Phase of the ultrasonic waves were controllable in nature. The piezoelectric crystal based ultrasonic transmitter and receiver of 40 kHz were used here. The ultrasound produced here causes series of vibrating pattern on the area of its applications. Molecules usually vibrate specifically based on their structural orientations and physical properties. The vibration patterns of glucose molecule were observed here using infrared LED of 940nm. The experimental design and resultant output in Fast Fourier Transform Domain for glucose prediction in dextrose mixed intralipid phantoms had been described and discussed in this paper. This technology can be utilized to design and improve noninvasive blood glucose monitoring devices.

Jump neural network for real-time prediction of glucose concentration.

Prediction of the future value of a variable is of central importance in a wide variety of fields, including economy and finance, meteorology, informatics, and, last but not least important, medicine. For example, in the therapy of Type 1 Diabetes (T1D), in which, for patient safety, glucose concentration in the blood should be maintained in a defined normoglycemic range, the ability to forecast glucose concentration in the short-term (with a prediction horizon of around 30 min) might be sufficient to reduce the incidence of hypoglycemic and hyperglycemic events. Neural Network (NN) approaches are suitable for prediction purposes because of their ability to model nonlinear dynamics and handle in their inputs signals coming from different domains. In this chapter we illustrate the design of a jump NN glucose prediction algorithm that exploits past glucose concentration data, measured in real-time by a minimally invasive continuous glucose monitoring (CGM) sensor, and information on ingested carbohydrates, supplied by the patient himself or herself. The methodology is assessed by tuning the NN on data of ten T1D individuals and then testing it on a dataset of ten different subjects. Results with a prediction horizon of 30 min show that prediction of glucose concentration in T1D via NN is feasible and sufficiently accurate. The average time anticipation obtained is compatible with the generation of preventive hypoglycemic and hyperglycemic alerts and the improvement of artificial pancreas performance.

Development of a Glucose Measuring System Using a Confocal Optical System

SUMMARYOur research team has developed a confocal optical system using a near‐infrared laser that is capable of reducing the effects of scattering when measuring glucose concentration. We have measured the light reflection of our samples, and because our system has the depth resolution offered by a confocal optical system, it is able to limit the length of the light path and to detect light absorption at each depth level of the samples, thereby restricting the point of measurement. This report details the use of this system for a basic experiment involving measuring blood samples and predicting glucose concentration levels. The results show a high correlation between the glucose concentration and near‐infrared light absorption. We also demonstrate the potential for calculating glucose concentrations in living organisms using our system.

Glucose Adulteration in Saudi Honey with Visible and Near Infrared Spectroscopy

This article reports on the implementation of visible and near infrared spectroscopy for the detection of glucose concentration in a mixture of Saudi and imported honey samples adulterated by glucose syrup of five concentrations: 0, 5, 12, 19, and 33 g/100 g. Honey samples were scanned in trans-reflectance mode with an AgroSpec mobile, fibre type, visible and near infrared spectrophotometer (tec5 Technology for Spectroscopy, Germany), with a measurement range of 305–2200 nm. The entire data set of 345 spectra was randomly divided into calibration (70%) and prediction (30%) sets. The first group was subjected to a partial least squares regression analysis with a leave-one-out cross-validation to establish a calibration model for the prediction of glucose concentration, whereas the second group was used to validate the partial least squares model. For the cross-validation, the values for root mean square error of prediction, coefficient of determination, and ratio of prediction deviation, which is the standard deviation divided by root mean square error of prediction were 4.52 g/100 g, 0.85, and 2.53, respectively. A slightly lower range of accuracy was obtained in the prediction set, with root mean square error of prediction, coefficient of determination, and ratio of prediction deviation values of 5.56 g/100 g, 0.78 and 2.06, respectively. The results achieved suggest that the visible and near infrared spectroscopy is a powerful technique for the quantification of glucose adulteration in Saudi honey.

Using Experimental Data Designs and Multivariate Modeling to Assess the Effect of Glycated Serum Protein Concentration on Glucose Prediction from Near-Infrared Spectra of Human Serum

Near-infrared (NIR) spectra of human blood serum consist of overlapping strong absorption bands of water and serum proteins, which affect the ability of multivariate calibration models to predict glucose. Furthermore, serum proteins such as albumin and globulins undergo a glycation reaction by forming covalent bonds with freely available glucose molecules in the serum. In diabetic individuals with poor glucose control, more and more serum protein molecules react with glucose, resulting in a high glycated protein concentration. The glucose molecules covalently bonded to serum proteins might contribute to the overall glucose signal acquired by NIR spectroscopy. This might affect the prediction ability of multivariate calibration models such as partial least squares regression (PLSR). In this study, we investigated the effect of total protein concentration and the glycated protein concentration in blood serum on the prediction ability of PLSR calibration models. Serum samples were subjected to ultra-filtration, and the PLSR model was built using NIR spectra of filtered serum solutions. Prediction performance was found to improve by 39-42% in absence of serum protein molecules. Various experimental data set designs were generated by carefully varying the glycated serum protein concentration in calibration and test sets of PLSR models. This investigation revealed that the impact of varying glycated protein concentration on the root mean square error of prediction was not drastic. To test the statistical significance of the prediction results, a multiple linear regression model was built. The glycated serum protein concentration was found to be statistically insignificant (p = 0.86) in predicting glucose concentration. Overall, it was concluded that the glycated serum proteins do not affect the glucose prediction accuracy of PLSR models using NIR spectra of human serum.

Multivariable Adaptive Identification and Control for Artificial Pancreas Systems

A constrained weighted recursive least squares method is proposed to provide recursive models with guaranteed stability and better performance than models based on regular identification methods in predicting the variations of blood glucose concentration in patients with Type 1 Diabetes. Use of physiological information from a sports armband improves glucose concentration prediction and enables earlier recognition of the effects of physical activity on glucose concentration. Generalized predictive controllers (GPC) based on these recursive models are developed. The performance of GPC for artificial pancreas systems is illustrated by simulations with UVa-Padova simulator and clinical studies. The controllers developed are good candidates for artificial pancreas systems with no announcements from patients.