A strategy based on integer programming for optimal dosing and timing of preventive hypoglycemic treatments in type 1 diabetes management

BACKGROUND AND OBJECTIVES

One of the major problems related to type 1 diabetes (T1D) management is hypoglycemia, a condition characterized by low blood glucose levels and responsible for reduced quality of life and increased mortality. Fast-acting carbohydrates, also known as hypoglycemic treatments (HT), can counteract this event. In the literature, dosage and timing of HT are usually based on heuristic rules. In the present work, we propose an algorithm for mitigating hypoglycemia by suggesting preventive HT consumption, with dosages and timing determined by solving an optimization problem.

METHODS

By leveraging integer programming and linear inequality constraints, the algorithm can bind the amount of suggested carbohydrates to standardized quantities (i.e., those available in "off-the-shelf" HT) and the minimal distance between consecutive suggestions (to reduce the nuisance for patients).

RESULTS

The proposed method was tested in silico and compared with competitor algorithms using the UVa/Padova T1D simulator. At the cost of a slight increase of HT consumed per day, the proposed algorithm produces the lowest median and interquartile range of the time spent in hypoglycemia, with a statistically significant improvement over most competitor algorithms. Also, the average number of hypoglycemic events per day is reduced to 0 in median.

CONCLUSIONS

Thanks to its positive performances and reduced computational burden, the proposed algorithm could be a candidate tool for integration in a DSS aimed at improving T1D management.

Emotional intelligence: a comparison between patients after first episode mania and those suffering from chronic bipolar disorder type I

BACKGROUND

Deficits in emotional intelligence (EI) were detected in patients with bipolar disorder (BD), but little is known about whether these deficits are already present in patients after presenting a first episode mania (FEM). We sought (i) to compare EI in patients after a FEM, chronic BD and healthy controls (HC); (ii) to examine the effect exerted on EI by socio-demographic, clinical and neurocognitive variables in FEM patients.

METHODS

The Emotional Intelligence Quotient (EIQ) was calculated with the Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT). Performance on MSCEIT was compared among the three groups using generalized linear models. In patients after a FEM, the influence of socio-demographic, clinical and neurocognitive variables on the EIQ was examined using a linear regression model.

RESULTS

In total, 184 subjects were included (FEM n = 48, euthymic chronic BD type I n = 75, HC n = 61). BD patients performed significantly worse than HC on the EIQ [mean difference (MD) = 10.09, standard error (s.e.) = 3.14, p = 0.004] and on the understanding emotions branch (MD = 7.46, s.e. = 2.53, p = 0.010). FEM patients did not differ from HC and BD on other measures of MSCEIT. In patients after a FEM, EIQ was positively associated with female sex (β = -0.293, p = 0.034) and verbal memory performance (β = 0.374, p = 0.008). FEM patients performed worse than HC but better than BD on few neurocognitive domains.

CONCLUSIONS

Patients after a FEM showed preserved EI, while patients in later stages of BD presented lower EIQ, suggesting that impairments in EI might result from the burden of disease and neurocognitive decline, associated with the chronicity of the illness.

Prodromal phase: Differences in prodromal symptoms, risk factors and markers of vulnerability in first episode mania versus first episode psychosis with onset in late adolescence or adulthood

OBJECTIVE

This study was aimed at identifying differences in the prodromal symptoms and their duration, risk factors and markers of vulnerability in patients presenting a first episode mania (FEM) or psychosis (FEP) with onset in late adolescence or adulthood in order to guide tailored treatment strategies.

METHODS

Patients with a FEM or FEP underwent a clinical assessment. Prodromes were evaluated with the Bipolar Prodrome Symptom Scale-Retrospective (BPSS-R). Chi-squared tests were conducted to assess specific prodromal symptoms, risk factors or markers of vulnerability between groups. Significant prodromal symptoms were entered in a stepwise forward logistic regression model. The probabilities of a gradual versus rapid onset pattern of the prodromes were computed with logistic regression models.

RESULTS

The total sample included 108 patients (FEM = 72, FEP = 36). Social isolation was associated with the prodromal stage of a FEP whilst Increased energy or goal-directed activity with the prodrome to a FEM. Physically slowed down presented the most gradual onset whilst Increased energy presented the most rapid. The presence of obstetric complications and difficulties in writing and reading during childhood were risk factors for FEP. As for markers of vulnerability, impairment in premorbid adjustment was characteristic of FEP patients. No specific risk factor or marker of vulnerability was identified for FEM.

CONCLUSION

Early characteristics differentiating FEP from FEM were identified. These findings might help shape early identification and preventive intervention programmes.

Shaped before birth: Obstetric complications identify a more severe clinical phenotype among patients presenting a first affective or non-affective episode of psychosis

Obstetric complications (OCs) may contribute to the heterogeneity that characterizes psychiatric illness, particularly the phenotypic presentation of first episode psychoses (FEP). Our aim was to examine the relationship between OCs and socio-demographic, clinical, functioning and neuropsychological characteristics in affective and non-affective FEP. We performed a cross-sectional,study where we recruited participants with FEP between 2011 and 2021, and retrospectively assessed OCs using the Lewis-Murray scale. OCs were used as a dichotomous variable and further stratified into three subtypes: complications of pregnancy, abnormal fetal growth and development, and difficulties in delivery. We performed a logistic stepwise forward regression analysis to examine variables associated with the presence of OCs. Of the 104 participants (67 affective FEP and 37 non-affective FEP), 31.7% (n = 33) had experienced OCs. Subjects with OCs showed a more gradual emergence of prodromal symptoms as well as higher negative and total Positive and Negative Syndrome Scale (PANSS) scores. In the multivariate analysis, the presence of OCs was independently associated with a younger age at first episode of any type (OR = 0.904, p = 0.003) and slower emergence of prodromal symptoms (OR = 0.274, p = 0.011). When considering specific types of OCs, those related with fetal growth were associated with worse neuropsychological performance, while OCs at delivery were related to earlier onset of illness and more severe symptoms. In conclusion, OCs signaled a specific FEP phenotype characterized by earlier and more protracted onset of illness as well as more burdensome symptoms, independently of FEP type (i.e., affective vs non-affective). These results indicate a potential target of early intervention in FEP.

Super-twisting-based meal detector for type 1 diabetes management: Improvement and assessment in a real-life scenario

BACKGROUND AND OBJECTIVE

Hybrid automated insulin delivery systems rely on carbohydrate counting to improve postprandial control in type 1 diabetes. However, this is an extra burden on subjects, and it introduces a source of potential errors that could impact control performances. In fact, carbohydrates estimation is challenging, prone to errors, and it is known that subjects sometimes struggle to adhere to this requirement, forgetting to perform this task. A possible solution is the use of automated meal detection algorithms. In this work, we extended a super-twisting-based meal detector suggested in the literature and assessed it on real-life data.

METHODS

To reduce the false detections in the original meal detector, we implemented an implicit discretization of the super-twisting and replaced the Euler approximation of the glucose derivative with a Kalman filter. The modified meal detector is retrospectively evaluated in a challenging real-life dataset corresponding to a 2-week trial with 30 subjects using sensor-augmented pump control. The assessment includes an analysis of the nature and riskiness of false detections.

RESULTS

The proposed algorithm achieved a recall of 70 [13] % (median [interquartile range]), a precision of 73 [26] %, and had 1.4 [1.4] false positives-per-day. False positives were related to rising glucose conditions, whereas false negatives occurred after calibrations, missing samples, or hypoglycemia treatments.

CONCLUSIONS

The proposed algorithm achieves encouraging performance. Although false positives and false negatives were not avoided, they are related to situations with a low risk of hypoglycemia and hyperglycemia, respectively.

Generation of post-meal insulin correction boluses in type 1 diabetes simulation models for in-silico clinical trials: More realistic scenarios obtained using a decision tree approach

BACKGROUND AND OBJECTIVE

In type 1 diabetes (T1D) research, in-silico clinical trials (ISCTs) notably facilitate the design/testing of new therapies. Published simulation tools embed mathematical models of blood glucose (BG) and insulin dynamics, continuous glucose monitoring (CGM) sensors, and insulin treatments, but lack a realistic description of some aspects of patient lifestyle impacting on glucose control. Specifically, to effectively simulate insulin correction boluses, required to treat post-meal hyperglycemia (BG > 180 mg/dL), the timing of the bolus may be influenced by subjects' behavioral attitudes. In this work, we develop an easily interpretable model of the variability of correction bolus timing observed in real data, and embed it into a popular simulation tool for ISCTs.

METHODS

Using data collected in 196 adults with T1D monitored in free-living conditions, we trained a decision tree (DT) model to classify whether a correction bolus is injected in a future time window, based on predictors collected back in time, related to CGM data, previous insulin boluses and subject's characteristics. The performance was compared to that of a logistic regression classifier with LASSO regularization (LC), trained on the same dataset. After validation, the DT was embedded within a popular T1D simulation tool and an ISCT was performed to compare the simulated correction boluses against those observed in a subset of data not used for model training.

RESULTS

The DT provided better classification performance (accuracy: 0.792, sensitivity: 0.430, specificity: 0.878, precision: 0.455) than the LC and presented good interpretability. The most predictive features were related to CGM (and its temporal variations), time since the last insulin bolus, and time of the day. The correction boluses simulated by the DT, after implementation in the simulation tool, showed a good agreement with real-world data.

CONCLUSIONS

The DT developed in this work represents a simple set of rules to mimic the same timing of correction boluses observed on real data. The inclusion of the model in simulation tools allows investigators to perform ISCTs that more realistically represent the patient behavior in taking correction boluses and the post-prandial BG response. In the future, more complex models can be investigated.

A Mathematical Formula to Determine the Minimum Continuous Glucose Monitoring Duration to Assess Time-in-ranges: Sensitivity Analysis Over the Parameters

In diabetes management, the fraction of time spent with glucose concentration within the physiological range of [70-180] mg/dL, namely time in range (TIR) is often computed by clinicians to assess glycemic control using a continuous glucose monitoring sensor. However, a sufficiently long monitoring period is required to reliably estimate this index. A mathematical equation derived by our group provides the minimum trial duration granting a desired uncertainty around the estimated TIR. The equation involves two parameters, p_r and α, related to the population under analysis, which should be set based on the clinician's experience. In this work, we evaluated the sensitivity of the formula to the parameters.Considering two independent datasets, we predicted the uncertainty of TIR estimate for a population, using the parameters of the formula estimated for a different population. We also stressed the robustness of the formula by testing wider ranges of parameters, thus assessing the impact of large errors in the parameters' estimates.Plausible errors on the α estimate impact very slightly on the prediction (relative discrepancy < 5%), thus we suggest using a fixed value for α independently on the population being analyzed. Instead, p_r should be adjusted to the TIR expected in the population, considering that errors around 20% result in a relative discrepancy of ~10%.In conclusion, the proposed formula is sufficiently robust to parameters setting and can be used by investigators to determine a suitable duration of the study.

Nonlinear Machine Learning Models for Insulin Bolus Estimation in Type 1 Diabetes Therapy

Type 1 diabetes (T1D) therapy requires multiple daily insulin injections to compensate the lack of endogenous insulin production due to β-cells destruction. An empirical standard formula (SF) is commonly used for such a task. Unfortunately, SF does not include information on glucose dynamics, e.g. the glucose rate-of-change (ROC) provided by continuous glucose monitoring (CGM) sensor. Hence, SF can sometimes lead to under/overestimations that can cause critical hypo/hyperglycemic episodes during/after the meal. Recently, to overcome this limitation, we proposed new linear regression models, integrating ROC information and personalized features. Despite the first encouraging results, the nonlinear nature of the problem calls for the application of nonlinear models. In this work, random forest (RF) and gradient boosting tree (GBT), nonlinear machine learning methodologies, were investigated. A dataset of 100 virtual subjects, opportunely divided into training and testing sets, was used. For each individual, a single-meal scenario with different meal conditions (preprandial ROC, BG and meal amounts) was simulated. The assessment was performed both in terms of accuracy in estimating the optimal bolus and glycemic control. Results were compared to the best performing linear model previously developed. The two tree-based models proposed lead to a statistically significant improvement of glycemic control compared to the linear approach, reducing the time spent in hypoglycemia (from 32.49% to 27.57-25.20% for RF and GBT, respectively). These results represent a preliminary step to prove that nonlinear machine learning techniques can improve the estimation of insulin bolus in T1D therapy. Particularly, RF and GBT were shown to outperform the previously linear models proposed.Clinical Relevance- Insulin bolus estimation with nonlinear machine learning techniques reduces the risk of adverse events in T1D therapy.

Non-Invasive Continuous-Time Blood Pressure Estimation from a Single Channel PPG Signal using Regularized ARX Models

Continuous blood pressure (BP) monitoring can help in preventing hypertension and other cardiovascular diseases. In principle, an indirect non-invasive continuous-time measurement of BP is possible by exploiting the photoplethysmography (PPG) signal, which can be obtained through wearable optical sensor devices. However, a model of the PPG-to-BP dynamical system is needed. In this study, we investigate if autoregressive with exogenous input (ARX) models with kernel-based regularization are suited for the scope. We analyzed 10 PPG time-series acquired on different individuals by a wearable optical sensor and correspondent BP reference values to evaluate feasibility of continuous BP estimation from a single PPG source. This first proof-of-concept study shows promising results in continuous BP estimation during resting states.

Patient decision-making of CGM sensor driven insulin therapies in type 1 diabetes: In silico assessment

In type 1 diabetes (T1D) therapy, continuous glucose monitoring (CGM) sensors, which provide glucose concentration in the subcutis every 1-5 min for 7 consecutive days, should allow in principle a more efficient insulin dosing than that based on the conventional 3-4 self-monitoring of blood glucose (SMBG) measurements per day. However, CGM, at variance with SMBG, is still not approved for insulin dosing in T1D management because regulatory agencies, e.g. FDA, are looking for more factual evidence on its safety. An in silico assessment of SMBG- vs CGM-driven insulin therapy can be a first step. Here we present a simulation model of T1D patient decision-making obtained by interconnecting models of glucose-insulin dynamics, SMBG and CGM measurement errors, carbohydrates-counting errors, insulin boluses time variability and forgetfulness, and subcutaneous insulin pump delivery. Inter- and intra- patient variability of model parameters are considered. The T1D patient decision-making model allows to run realistic multi-day simulations scenarios in a population of virtual subjects. We present the first results of simulations run in 20 virtual subjects over a 7-day period, which demonstrates that additional information brought by CGM (trend and hypo/hyperglycemic warnings) with respect to SMBG produces a statistically significant increment (about of 9%) of time spent by the patient in the euglycemic range (70-180 mg/dl).

Accuracy of devices for self-monitoring of blood glucose: A stochastic error model

Self-monitoring of blood glucose (SMBG) devices are portable systems that allow measuring glucose concentration in a small drop of blood obtained via finger-prick. SMBG measurements are key in type 1 diabetes (T1D) management, e.g. for tuning insulin dosing. A reliable model of SMBG accuracy would be important in several applications, e.g. in in silico design and optimization of insulin therapy. In the literature, the most used model to describe SMBG error is the Gaussian distribution, which however is simplistic to properly account for the observed variability. Here, a methodology to derive a stochastic model of SMBG accuracy is presented. The method consists in dividing the glucose range into zones in which absolute/relative error presents constant standard deviation (SD) and, then, fitting by maximum-likelihood a skew-normal distribution model to absolute/relative error distribution in each zone. The method was tested on a database of SMBG measurements collected by the One Touch Ultra 2 (Lifescan Inc., Milpitas, CA). In particular, two zones were identified: zone 1 (BG≤75 mg/dl) with constant-SD absolute error and zone 2 (BG>75mg/dl) with constant-SD relative error. Mean and SD of the identified skew-normal distributions are, respectively, 2.03 and 6.51 in zone 1, 4.78% and 10.09% in zone 2. Visual predictive check validation showed that the derived two-zone model accurately reproduces SMBG measurement error distribution, performing significantly better than the single-zone Gaussian model used previously in the literature. This stochastic model allows a more realistic SMBG scenario for in silico design and optimization of T1D insulin therapy.

Automated analysis of local field potentials evoked by mechanical whisker stimulation in rat barrel cortex

Local field potentials (LFPs) recorded in the barrel cortex in rats and mice are important to investigate somatosensory systems, the final aim being to start to understand mechanisms of brain representation of sensory stimuli in humans. Parameters extracted from LFP of particular interest include spike timing and transmembrane current flow. Recent improvements in microelectrodes technology have enabled neuroscientists to acquire a great amount of LFP signals during the same experimental session, calling for the development of algorithms for their quantitative automatic analysis. In the present work, an algorithm based on Phillips-Tikhonov regularization is presented to automatically detect the main features (in terms of amplitude and latency) of LFP waveforms recorded after whisker stimulation in rat. The accuracy of the algorithm is first assessed in a Monte Carlo simulation mimicking the acquisition of LFP in three different conditions of SNR. Then, the algorithm is tested by analyzing a set of 100 LFP recorded in the primary somatosensory (S1) cortex, i.e., the region involved in the cortical representation of touch in mammals.

Decrease of EEG Coherence during hypoglycemia in type 1 diabetic subjects

Hypoglycemic events have been proven to be associated with measurable EEG changes. Several works in the literature have evaluated these changes by considering approaches at the single EEG channel level, but multivariate analyses have been scarcely investigated in Type 1 diabetes (T1D) subjects. The aim of the present work is to assess if and how hypoglycemia affects EEG coherence in a subset of EEG channels acquired in a hospital setting where eye- and muscle activation-induced artifacts are virtually absent. In particular, EEG multichannel data, acquired in 19 T1D hospitalized subjects undertaken to an insulin-induced hypoglycemia experiment, are considered. Computation of Partial Directed Coherence (PDC) through multivariate autoregressive models of P3-A1A2, P4-A1A2, C3-A1A2 and C4-A1A2 EEG channels shows that a decrease in the value of coherence, most likely related to the progressive loss of cognitive function and altered cerebral activity, occurs when passing from eu- to hypoglycemia, in both theta ([4, 8] Hz) and alpha ([8, 13] Hz) bands.

An EEGLAB plugin to analyze individual EEG alpha rhythms using the "channel reactivity-based method"

A recent paper [1] proposed a new technique, termed the channel reactivity-based method (CRB), for characterizing EEG alpha rhythms using individual (IAFs) and channel (CAFs) alpha frequencies. These frequencies were obtained by identifying the frequencies at which the power of the alpha rhythms decreases. In the present study, we present a graphical interactive toolbox that can be plugged into the popular open source environment EEGLAB, making it easy to use CRB. In particular, we illustrate the major functionalities of the software and discuss the advantages of this toolbox for common EEG investigations. The CRB analysis plugin, along with extended documentation and the sample dataset utilized in this study, is freely available on the web at http://bio.dei.unipd.it/crb/.

Sparse Principal Component Analysis for the parsimonious description of glucose variability in diabetes

Abnormal glucose variability (GV) is considered to be a risk factor for the development of diabetes complications. For its quantification from continuous glucose monitoring (CGM) data, tens of different indices have been proposed in the literature, but the information carried by them is highly redundant. In the present work, the Sparse Principal Component Analysis (SPCA) technique is used to select, from a wide pool of GV metrics, a smaller subset of indices that preserves the majority of the total original variance, providing a parsimonious but still comprehensive description of GV. In detail, SPCA is applied to a set of 25 literature GV indices evaluated on CGM time-series collected in 17 type 1 (T1D) and 13 type 2 (T2D) diabetic subjects. Results show that the 10 GV indices selected by SPCA preserve more than the 75% of the variance of the original set of 25 indices, both in T1D and T2D. Moreover, 6 indices of the parsimonious set are shared by T1D and T2D.

Jump neural network for online short-time prediction of blood glucose from continuous monitoring sensors and meal information

Several real-time short-term prediction methods, based on time-series modeling of past continuous glucose monitoring (CGM) sensor data have been proposed with the aim of allowing the patient, on the basis of predicted glucose concentration, to anticipate therapeutic decisions and improve therapy of type 1 diabetes. In this field, neural network (NN) approaches could improve prediction performance handling in their inputs additional information. In this contribution we propose a jump NN prediction algorithm (horizon 30 min) that exploits not only past CGM data but also ingested carbohydrates information. The NN is tuned on data of 10 type 1 diabetics and then assessed on 10 different subjects. Results show that predictions of glucose concentration are accurate and comparable to those obtained by a recently proposed NN approach (Zecchin et al. (2012) [26]) having higher structural and algorithmical complexity and requiring the patient to announce the meals. This strengthen the potential practical usefulness of the new jump NN approach.

Insight into the relationship between brain/behavioral speed and variability in patients with minimal hepatic encephalopathy

OBJECTIVE

Intra-individual variability (IIV) of response reaction times (RTs) and psychomotor slowing were proposed as markers of brain dysfunction in patients with minimal hepatic encephalopathy (MHE), a subclinical disorder of the central nervous system frequently detectable in patients with liver cirrhosis. However, behavioral measures alone do not enable investigations into the neural correlates of these phenomena. The aim of this study was to investigate the electrophysiological correlates of psychomotor slowing and increased IIV of RTs in patients with MHE.

METHODS

Event-related potentials (ERPs), evoked by a stimulus-response (S-R) conflict task, were recorded from a sample of patients with liver cirrhosis, with and without MHE, and a group of healthy controls. A recently presented Bayesian approach was used to estimate single-trial P300 parameters.

RESULTS

Patients with MHE, with both psychomotor slowing and higher IIV of RTs, showed higher P300 latency jittering and lower single-trial P300 amplitude compared to healthy controls. In healthy controls, distribution analysis revealed that single-trial P300 latency increased and amplitude decreased as RTs became longer; however, in patients with MHE the linkage between P300 and RTs was weaker or even absent.

CONCLUSIONS

These findings suggest that in patients with MHE, the loss of the relationship between P300 parameters and RTs is related to both higher IIV of RTs and psychomotor slowing.

SIGNIFICANCE

This study highlights the utility of investigating the relationship between single-trial ERPs parameters along with RT distributions to explore brain functioning in normal or pathological conditions.

A new neural network approach for short-term glucose prediction using continuous glucose monitoring time-series and meal information

In the last decade, improvements in diabetes daily management have become possible thanks to the development of minimally-invasive portable sensors which allow continuous glucose monitoring (CGM) for several days. In particular, hypo and hyperglycemia can be promptly detected when glucose exceeds the normal range thresholds, and even avoided through the use of on-line glucose prediction algorithms. Several algorithms with prediction horizon (PH) of 15-30-45 min have been proposed in the literature, e.g. including AR/ARMA time-series modeling and neural networks. Most of them are fed by CGM signals only. The purpose of this work is to develop a new short-term glucose prediction algorithm based on a neural network that, in addition to past CGM readings, also exploits information on carbohydrates intakes quantitatively described through a physiological model. Results on simulated data quantitatively show that the new method outperforms other published algorithms. Qualitative preliminary results on a real diabetic subject confirm the potentialities of the new approach.

A methodology to improve estimation of stimulus-evoked hemodynamic response from fNIRS measurements

Functional near-infrared spectroscopy (fNIRS) is a non-invasive optical neuroimaging method used to investigate functional activity of the cerebral cortex evoked by cognitive, visual, auditory and motor tasks, detecting regional changes of oxy- and deoxy-hemoglobin concentration. Accurate estimation of the stimulus-evoked hemodynamic response (HR) from fNIRS signals in order to quantitatively investigate cognitive functions requires to cope with several noise components. Some of them appear as random disturbances (typically tackled through averaging techniques), while others are due to physiological sources, such as heart beat, respiration, vasomotor waves, and are particularly challenging to be dealt with because they lie in the same frequency band of HR. In this work we present a new two-steps methodology for the HR estimation from fNIRS data. The first step is a pre-processing stage where physiological trends in fNIRS data are reduced by exploiting a mathematical model identified from the signal of a reference channel. In the second step, the pre-processed data of the other channels are filtered with a recently presented non-parametric Bayesian approach (Scarpa et al., Optics Express, 2010). The presented method for HR estimation is compared with widely used methods: conventional averaging, band-pass filtering and principal component analysis (PCA). Results on simulated data reveal the ability of the proposed method to improve the accuracy of the estimates of the functional hemodynamic response, as well as the estimate of peak amplitude and latency. Encouraging preliminary results in a representative real data set showing an improvement of contrast to noise ratio are also reported.

Bayesian filtering of human brain hemodynamic activity elicited by visual short-term maintenance recorded through functional near-infrared spectroscopy (fNIRS)

Functional near-infrared spectroscopy (fNIRS) is a neuroimaging technique that measures changes in oxy-hemoglobin (ΔHbO) and deoxy-hemoglobin (ΔHbR) concentration associated with brain activity. The signal acquired with fNIRS is naturally affected by disturbances engendering from ongoing physiological activity (e.g., cardiac, respiratory, Mayer wave) and random measurement noise. Despite its several drawbacks, the so-called conventional averaging (CA) is still widely used to estimate the hemodynamic response function (HRF) from noisy signal. One such drawback is related to the number of trials necessary to derive stable HRF functions adopting the CA approach, which must be substantial (N >> 50). In this work, a pre-processing procedure to remove artifacts followed by the application of a non-parametric Bayesian approach is proposed that capitalizes on a priori available knowledge about HRF and noise. Results with the proposed Bayesian approach were compared with CA and with a straightforward band-pass filtering approach. On simulated data, a five times lower estimation error on HRF was obtained with respect to that obtained by CA, and 2.5 times lower than that obtained by band pass filtering. On real data, the improvement achieved by the present method was attested by an increase in the contrast to noise ratio (CNR) and by a reduced variability in single trial estimation. An application of the present Bayesian approach is illustrated that was optimized to monitor changes in hemodynamic activity reflecting variations in visual short-term memory load in humans, which are notoriously hard to detect using functional magnetic resonance imaging (fMRI). In particular, statistical analyses of HRFs recorded during a memory task established with high reliability the crucial role of the intraparietal sulcus and the intra-occipital sulcus in posterior areas of the human brain in visual short-term memory maintenance.

Artificial neural network algorithm for online glucose prediction from continuous glucose monitoring

BACKGROUND AND AIMS

Continuous glucose monitoring (CGM) devices could be useful for real-time management of diabetes therapy. In particular, CGM information could be used in real time to predict future glucose levels in order to prevent hypo-/hyperglycemic events. This article proposes a new online method for predicting future glucose concentration levels from CGM data.

METHODS

The predictor is implemented with an artificial neural network model (NNM). The inputs of the NNM are the values provided by the CGM sensor during the preceding 20 min, while the output is the prediction of glucose concentration at the chosen prediction horizon (PH) time. The method performance is assessed using datasets from two different CGM systems (nine subjects using the Medtronic [Northridge, CA] Guardian and six subjects using the Abbott [Abbott Park, IL] Navigator. Three different PHs are used: 15, 30, and 45 min. The NNM accuracy has been estimated by using the root mean square error (RMSE) and prediction delay.

RESULTS

The RMSE is around 10, 18, and 27 mg/dL for 15, 30, and 45 min of PH, respectively. The prediction delay is around 4, 9, and 14 min for upward trends and 5, 15, and 26 min for downward trends, respectively. A comparison with a previously published technique, based on an autoregressive model (ARM), has been performed. The comparison shows that the proposed NNM is more accurate than the ARM, with no significant deterioration in the prediction delay.

CONCLUSIONS

The proposed NNM is a reliable solution for the online prediction of future glucose concentrations from CGM data.

Reconstructing by deconvolution plasma glucose from continuous glucose monitoring sensor data

In the recent past, several sensors have been developed which allow continuous glucose monitoring (CGM) for several days. CGM can improve diabetes management and in particular decrease the risk of hypoglycemic events. However, CGM sensors measure glucose concentration in the interstitial fluid (ISF) rather than in plasma and ISF lags plasma glucose. The purpose of this work is to investigate if plasma glucose can be reconstructed from ISF CGM data by using a deconvolution approach, based on the knowledge of the model of plasma-interstitium kinetics. Results obtained in 6 volunteers monitored for 2 days with simultaneous plasma and ISF glucose by the Freestyle Navigator CGM sensor measurements show that calibration is a critical component for a reliable reconstruction of plasma glucose from ISF data.

Method for the deconvolution of auditory steady-state responses

The potential evoked by a 'train' of N equally spaced auditory clicks, with an inter-click period shorter than the duration of the response to an isolated click, is said to be a steady-state response (SSR). Extracting the individual responses evoked by the clicks of the train during steady state can be key to understanding of the neurophysiological mechanisms underlying SSR generation. In the literature, this task has been dealt with only under the (unwarranted) assumption that the response of the system does not vary during the presentation of the clicks, i.e. no neurophysiological adaptation is present. In this work, a new, non-parametric algorithm is proposed that, relaxing the time-invariance hypothesis, allows the extraction from the SSR of the N waveforms individually evoked by the N clicks of the train. The performance of the approach is evaluated on simulated SSRs and on real data recorded from the temporal cortex of awake rats. Results show that the method is able to detect and assess possible adaptation of the neurophysiological system in the generation of SSRs.

Reconstructing insulin secretion rate after a glucose stimulus by an improved stochastic deconvolution method

Reconstructing insulin secretion rate (ISR) after a glucose stimulus by deconvolution is difficult because of its biphasic pattern, i.e., a rapid secretion peak is followed by a slower release. Here, we refine a recently proposed stochastic deconvolution method by modeling ISR as the multiple integration of a white noise process with time-varying statistics. The unknown parameters are estimated from the data by employing a maximum likelihood criterion. A fast computational scheme implementing the method is presented. Monte Carlo simulation results are developed which numerically show a more reliable ISR profile reconstructed by the new method.

Electrocochleography potentials evoked by condensation and rarefaction clicks independently derived by a new numerical filtering approach

The cochlear microphonic potential (CM) and the compound action potential (CAP) cannot be measured separately but only in combination. In the literature their individual estimates are conventionally recovered by the so-called CM cancellation technique. This method averages the potential obtained in response to rarefaction and condensation clicks under the assumption that changing the polarity of the clicks only affects the CM sign and does not alter the CAP in any way. However, both theory and evidence suggest that these hypotheses can be critical. In addition, recent contributions in the electrocochleography (ECochG) literature suggested that assessing the influence of stimulus polarity on the evoked CAP may constitute an indicator of clinical usefulness which the CM cancellation method cannot supply. In this work we propose a new algorithm to estimate the cochlear potentials evoked from positive clicks, CAP+ and CM+, and those evoked from negative clicks, CAP- and CM-, by processing the same kind and amount of data employed in the CM cancellation method. The application to real data taken from 3 subjects exhibiting quantitatively and qualitatively different ECochG responses at various levels of stimulation intensity is presented. In addition, simulated problems where the true CAP and CM are known are studied to permit a fair assessment of the proposed technique. Results suggest that the new algorithm is potentially able to point out small differences between CAP+ and CAP-. This encourages its further employment on a larger scale.

Bayesian identification of a population compartmental model of C-peptide kinetics

When models are used to measure or predict physiological variables and parameters in a given individual, the experiments needed are often complex and costly. A valuable solution for improving their cost effectiveness is represented by population models. A widely used population model in insulin secretion studies is the one proposed by Van Cauter et al. (Diabetes 41:368-377, 1992), which determines the parameters of the two compartment model of C-peptide kinetics in a given individual from the knowledge of his/her age, sex, body surface area, and health condition (i.e., normal, obese, diabetic). This population model was identified from the data of a large training set (more than 200 subjects) via a deterministic approach. This approach, while sound in terms of providing a point estimate of C-peptide kinetic parameters in a given individual, does not provide a measure of their precision. In this paper, by employing the same training set of Van Cauter et al., we show that the identification of the population model into a Bayesian framework (by using Markov chain Monte Carlo) allows, at the individual level, the estimation of point values of the C-peptide kinetic parameters together with their precision. A successful application of the methodology is illustrated in the estimation of C-peptide kinetic parameters of seven subjects (not belonging to the training set used for the identification of the population model) for which reference values were available thanks to an independent identification experiment.

Approximate entropy studies of hormone pulsatility from plasma concentration time series: influence of the kinetics assessed by simulation

Approximate entropy (ApEn) is a method developed in the early nineties to quantify the "regularity" of a time series. In recent years, it has been vigorously employed to study the oscillatory/pulsatile secretory behavior of many hormones and found capable of successfully identifying pathological or prepathological states characterized by an enhanced secretion irregularity. Since hormone secretion rate is nonaccessible to direct measurement, ApEn is usually calculated from the time series of the hormone concentrations in plasma. However, the plasma concentration time course also reflects the whole-body kinetics of the hormone and can thus only provide a distorted portrait of the secretion rate at the gland level. In this paper, we investigate by simulation whether and how this distortion can influence the study of the regularity of hormone pulsatility by ApEn. Pulsatile secretion time series with different degrees of irregularity are simulated by varying the statistics of the random parameters which describe the secretory pulses. Then, plasma concentration time series are obtained by convolution with the hormone impulse response. Different degrees of impulse response smoothness are also considered in order to vary the amount of the distortion introduced. Results show that ApEn computed from secretion time series consistently discriminated better than ApEn calculated from plasma concentration time series among processes with different degrees of regularity. In addition, smoother impulse responses decreased the ApEn differences between plasma concentration time series corresponding to different degrees of secretion regularity. Therefore, the power of the ApEn index in the study of hormone pulsatility can potentially be enhanced by applying it to the hormone secretion time series.

Maximum-likelihood versus maximum a posteriori parameter estimation of physiological system models: the C-peptide impulse response case study

Maximum-likelihood (ML), also given its connection to least-squares (LS), is widely adopted in parameter estimation of physiological system models, i.e., assigning numerical values to the unknown model parameters from the experimental data. A more sophisticated but less used approach is maximum a posteriori (MAP) estimation. Conceptually, while ML adopts a Fisherian approach, i.e., only experimental measurements are supplied to the estimator, MAP estimation is a Bayesian approach, i.e., a priori available statistical information on the unknown parameters is also exploited for their estimation. In this paper, after a brief review of the theory behind ML and MAP estimators, we compare their performance in the solution of a case study concerning the determination of the parameters of a sum of exponential model which describes the impulse response of C-peptide (CP), a key substance for reconstructing insulin secretion. The results show that MAP estimation always leads to parameter estimates with a precision (sometimes significantly) higher than that obtained through ML, at the cost of only a slightly worse fit. Thus, a three exponential model can be adopted to describe the CP impulse response model in place of the two exponential model usually identified in the literature by the ML/LS approach. Simulated case studies are also reported to evidence the importance of taking into account a priori information in a data poor situation, e.g., when a few or too noisy measurements are available. In conclusion, our results show that, when a priori information on the unknown model parameters is available, Bayes estimation can be of relevant interest, since it can significantly improve the precision of parameter estimates with respect to Fisher estimation. This may also allow the adoption of more complex models than those determinable by a Fisherian approach.

Compound action potential and cochlear microphonic extracted from electrocochleographic responses to condensation or rarefaction clicks

In electrocochleography (ECochG) compound action potential (CAP) and summation potential (SP) are usually separated from the cochlear microphonic (CM) by the CM cancellation technique consisting in averaging the responses evoked by rarefaction and condensation clicks. With the aim of analysing the ECochG responses evoked by monophasic clicks, we developed a numerical method based on the theory of optimal filtering, which makes no assumptions about the unknown potentials. The application of the filtering technique to the ECochG recordings obtained from 6 normally hearing children and 10 children with cochlear hearing loss allowed us to perform CAP extraction in cases where CM was not cancelled by the conventional method. Differences in SP amplitude and polarity were found between rarefaction and condensation click-evoked responses in cochlear hearing losses.

Estimation of organ transport function for recirculating indicator dilution curves

The transport function of an indicator through an organ allows the calculation of important physiological parameters, but its estimation, especially in the presence of recirculation, can be difficult. In this paper, we estimate the transport function of 3H-mannitol (an extracellular tracer of glucose) in the human leg skeletal muscle. To do so, an indicator bolus is administered into the femoral artery and its recirculating dilution curves are nonuniformly sampled in both the femoral artery and the femoral vein. A new deconvolution-based method is used to simultaneously estimate the indicator transport function and the organ plasma flow. Subsequently, the indicator mean transit time and distribution volume are calculated. The reliability of the method is assessed by Monte Carlo simulation. The ability to estimate parameters, like mean transit time and extracellular distribution volume, is critical to the study of pathophysiologic states such as diabetes, insulin resistance, and hypertension.

Removal of catheter distortion in multiple indicator dilution studies: a deconvolution-based method and case studies on glucose blood-tissue exchange

The study of blood-tissue exchange by the multiple indicator dilution technique often needs frequent sampling in the blood of the indicator dilution curves (IDC). Usually, this requires the use of a catheter supported by a pump. This causes a distortion in the IDC, which must be removed for proper interpretation of the data. A deconvolution-based methodology to remove IDC distortion is presented. First, the catheter impulse response is modelled by means of data obtained from a suitable experiment. Then the reconstruction of the blood IDC is tackled by a new nonparametric deconvolution algorithm, which provides (quasi) time-continuous signals and exploits statistically based criteria for the choice of the regularisation parameter. The methodology is applied to the removal of catheter distortion in studies of glucose blood-tissue exchange in the human forearm and myocardium.

Estimation of endogenous glucose production after a glucose perturbation by nonparametric stochastic deconvolution

The knowledge of the time course of endogenous glucose production (EGP) after a glucose perturbation is crucially important for understanding the glucose regulation system in both healthy and disease (e.g. diabetes) states. EGP is not directly accessible, and thus an indirect measurement approach is required. The estimation of EGP during an intravenous glucose tolerance test (IVGTT) can be posed as an input estimation problem solvable as a Fredholm integral equation of the first kind (A. Caumo and C. Cobelli, Am. J. Physiol., 264 (1993) E829-E841). The time-varying model of the kernel of the glucose system was identified from a concomitant tracer experiment, and EGP was reconstructed by employing the Phillips-Tikhonov regularization (deconvolution) algorithm. However, the proposed deconvolution approach left some issues open, e.g. how to choose the amount of regularization and how to deal with nonuniform/infrequent sampling. Here, a solution to these problems is provided by resorting to a new deconvolution algorithm. Thanks to the stochastic embedding into which the new deconvolution method is stated, the amount of regularization is determined in a statistically sound manner. In addition, in face of infrequent sampling, a time continuous profile of EGP is obtained. The method is shown to work reliably for reconstructing EGP in different IVGTT experimental protocols, both in normal and disease states.

Impulse response model in reconstruction of insulin secretion by deconvolution: role of input design in the identification experiment

Insulin secretion rate (ISR) in vivo can be reconstructed by deconvolution of plasma concentration of C-peptide (CP), a peptide co-secreted with insulin but not extracted by the liver and exhibiting linear kinetics. Deconvolution requires the knowledge of the CP impulse response. A two exponential model is usually chosen to describe the CP impulse response but three exponential and one exponential models have also been used. The purpose of this paper is to investigate the role of the CP impulse response model order in reconstructing ISR by deconvolution in three standard physiological/clinical situations: ultradian oscillations, rapid pulses, and biphasic response to a glucose stimulus. By resorting to simulation, we first show that, in each situation, the validity of impulse response models with different orders depends on the input chosen in the impulse response identification experiment. Real data are then used which support the simulation results.

A stochastic deconvolution method to reconstruct insulin secretion rate after a glucose stimulus

Insulin secretion rate (ISR) is not directly measurable in man but it can be reconstructed from C-peptide (CP) concentration measurements by solving an input estimation problem by deconvolution. The major difficulties posed by the estimation of ISR after a glucose stimulus, e.g., during an intravenous glucose tolerance test (IVGTT), are the ill-conditioning of the problem, the nonstationary pattern of the secretion rate, and the nonuniform/infrequent sampling schedule. In this work, a nonparametric method based on the classic Phillips-Tikhonov regularization approach is presented. The problem of nonuniform/infrequent sampling is addressed by a novel formulation of the regularization method which allows the estimation of quasi time-continuous input profiles. The input estimation problem is stated into a Bayesian context, where the a priori known nonstationary characteristics of ISR after the glucose stimulus are described by a stochastic model. Deconvolution is tackled by linear minimum variance estimation, thus allowing the derivation of new statistically based regularization criteria. Finally, a Monte-Carlo strategy is implemented to assess the uncertainty of the estimated ISR arising from CP measurement error and impulse response parameters uncertainty.

Reconstruction of insulin secretion rate by deconvolution: domain of validity of a monoexponential C-peptide impulse response model

Insulin secretion rate (ISR) in vivo is reconstructed by deconvolution from plasma concentration of C-peptide (CP), a peptide with linear kinetics which is co-secreted with insulin but is not extracted by the liver. Deconvolution requires the knowledge of the CP impulse response. A two-exponential (2E) model is usually chosen to describe the CP impulse response but a one-exponential (1E) model is also used in the literature. The purpose here is to discuss the domain of validity of the 1E model in reconstructing the ISR by deconvolution. In particular, we show that the 1E model can be reliably used only if the ISR spectrum is concentrated in a narrow frequency band and a suitable input is designed for its identification.

The subjective, behavioral and cognitive effects of subanesthetic concentrations of isoflurane and nitrous oxide in healthy volunteers

A prospective, crossover, double-blind trial was conducted in nine healthy volunteers in which the subjective, psychomotor and memory effects of isoflurane (0.0, 0.3 and 0.6%) and nitrous oxide (N2O) (0, 20 and 40%) were examined. Dependent measures included visual analog scales and a standardized drug effects inventory (subjective effects), reaction time and eye-hand coordination (e.g., psychomotor performance), and immediate and delayed free recall (memory). There were some similarities in subjective effects between the two inhaled drugs (e.g., increased ratings of "drunk" and "spaced out"), but isoflurane had effects which N2O did not have. Isoflurane but not N2O increased visual analog scale ratings of "confused," "sedated," and "carefree," and decreased ratings of "in control of thoughts" and "in control of body." An odor was detected with isoflurane and it was disliked. Psychomotor performance was more grossly impaired during isoflurane inhalation than during N2O inhalation. Psychomotor recovery from both agents was rapid and complete so that 5 min after the inhalation period had ceased, performance had returned to baseline levels. Both isoflurane and nitrous oxide impaired immediate and delayed free recall. The feasibility of using isoflurane in conscious sedation procedures is discussed.

Role of intraoperative ultrasound in the screening of liver metastases from colorectal carcinoma: initial experiences

The aim of this study was to assess the utility of intraoperative ultrasound (IOUS) in the diagnosis and management of liver metastases from colorectal carcinoma. IOUS was performed on a consecutive series of 70 patients undergoing surgery for colorectal carcinoma, with follow-up ranging from 6 to 24 months. In ten cases (14.3%), 13 metastatic tumours were diagnosed; only six of these had been found by preoperative workup and/or surgical inspection. Seven (53.9%) small metastatic liver lesions were identified only by IOUS. None of the lesions diagnosed by IOUS was palpable, and they were all extremely small--ranging from 4 x 6 to 12 x 16 mm. Seventy-three locations were examined in order to compare the results of IOUS with those of other methods. The sensitivity of the former proved to be higher (P less than .05) than that of conventional pre- and intraoperative screening.

Carcinoma of the lung, stage III. Experience with the new TNM-AJCC classification

The records of 228 patients who underwent surgery for primary lung cancer in 1970-1986 were reviewed. In 115 cases (94 men, 21 women) the disease was in stage III according to the 1978 classification of the American Joint Committee on Cancer (AJCC). These 115 cases were retrospectively reassessed, using a recently proposed new TNM classification with subdivision of stage III into IIIa, in which the patients may benefit from surgery, and IIIb, in which surgery is not advisable. Stage IIIa disease was present at operation in 87 cases and stage IIIb in 28. Actuarial analysis of survival rates showed that the new subclassification permits identification of those stage III patients who may benefit from surgical therapy.

Leukocyte migration test (LMT) in patients with thyroid disease: the response to human thyroid subcellular fractions

The response of circulating leukocytes to thyroid subcellular fractions was investigated in 19 patients with Graves' disease, 15 patients with Hashimoto's thyroiditis, 7 patients with toxic adenoma, 19 patients with nontoxic goiter and in 10 healthy students as control subjects. For this purpose, the leukocyte migration test of Soborg and Bendixen was performed against human crude thyroid extract (CTE), cell plasma membranes, nuclei, ribosomes, mitochondria and microsomes. Our results show positive LMT against: 1) CTE in patients with Graves' disease (61 +/- 13, p less than 0.001) and Hashimoto's thyroiditis (65 +/- 11, p less than 0.001) compared to controls (90 +/- 11); 2) cell plasma membranes in patients with Graves' disease (41 +/- 14, p less than 0.001) and Hashimoto's thyroiditis (64 +/- 21, p less than 0.05) compared to controls (88 +/- 19); 3) nuclei in patients with Graves' disease (53 +/- 25, p less than 0.001) and Hashimoto's thyroiditis (53 +/- 23, p less than 0.001) compared to controls (83 +/- 11). Our findings of circulating leukocytes sensitized to cell plasma membranes and nuclear fraction in patients with Graves' disease and Hashimoto's thyroiditis provide the additional information that these patients have a specific defect in immune-surveillance.