Electroencephalographic markers from routine sleep discriminate individuals who are vulnerable or resilient to sleep loss.

Sleep loss impairs cognition; however, individuals differ in their response to sleep loss. Current methods to identify an individual's vulnerability to sleep loss involve time-consuming sleep-loss challenges and neurobehavioural tests. Here, we sought to identify electroencephalographic markers of sleep-loss vulnerability obtained from routine night sleep. We retrospectively analysed four studies in which 50 healthy young adults (21 women) completed a laboratory baseline-sleep phase followed by a sleep-loss challenge. After classifying subjects as resilient or vulnerable to sleep loss, we extracted three electroencephalographic features from four channels during the baseline nights, evaluated the discriminatory power of these features using the first two studies (discovery), and assessed reproducibility of the results using the remaining two studies (reproducibility). In the discovery analysis, we found that, compared to resilient subjects, vulnerable subjects exhibited: (1) higher slow-wave activity power in channel O1 (p < 0.0042, corrected for multiple comparisons) and in channels O2 and C3 (p < 0.05, uncorrected); (2) higher slow-wave activity rise rate in channels O1 and O2 (p < 0.05, uncorrected); and (3) lower sleep spindle frequency in channels C3 and C4 (p < 0.05, uncorrected). Our reproducibility analysis confirmed the discovery results on slow-wave activity power and slow-wave activity rise rate, and for these two electroencephalographic features we observed consistent group-difference trends across all four channels in both analyses. The higher slow-wave activity power and slow-wave activity rise rate in vulnerable individuals suggest that they have a persistently higher sleep pressure under normal rested conditions.

APPRAISE-HRI: AN ARTIFICIAL INTELLIGENCE ALGORITHM FOR TRIAGE OF HEMORRHAGE CASUALTIES.

ABSTRACT: Background: Hemorrhage remains the leading cause of death on the battlefield. This study aims to assess the ability of an artificial intelligence triage algorithm to automatically analyze vital-sign data and stratify hemorrhage risk in trauma patients. Methods: Here, we developed the APPRAISE-Hemorrhage Risk Index (HRI) algorithm, which uses three routinely measured vital signs (heart rate and diastolic and systolic blood pressures) to identify trauma patients at greatest risk of hemorrhage. The algorithm preprocesses the vital signs to discard unreliable data, analyzes reliable data using an artificial intelligence-based linear regression model, and stratifies hemorrhage risk into low (HRI:I), average (HRI:II), and high (HRI:III). Results: To train and test the algorithm, we used 540 h of continuous vital-sign data collected from 1,659 trauma patients in prehospital and hospital (i.e., emergency department) settings. We defined hemorrhage cases (n = 198) as those patients who received ≥1 unit of packed red blood cells within 24 h of hospital admission and had documented hemorrhagic injuries. The APPRAISE-HRI stratification yielded a hemorrhage likelihood ratio (95% confidence interval) of 0.28 (0.13-0.43) for HRI:I, 1.00 (0.85-1.15) for HRI:II, and 5.75 (3.57-7.93) for HRI:III, suggesting that patients categorized in the low-risk (high-risk) category were at least 3-fold less (more) likely to have hemorrhage than those in the average trauma population. We obtained similar results in a cross-validation analysis. Conclusions: The APPRAISE-HRI algorithm provides a new capability to evaluate routine vital signs and alert medics to specific casualties who have the highest risk of hemorrhage, to optimize decision-making for triage, treatment, and evacuation.

Prospective Validation of 2B-Cool : Integrating Wearables and Individualized Predictive Analytics to Reduce Heat Injuries.

INTRODUCTION: An uncontrollably rising core body temperature (T C ) is an indicator of an impending exertional heat illness. However, measuring T C invasively in field settings is challenging. By contrast, wearable sensors combined with machine-learning algorithms can continuously monitor T C nonintrusively. Here, we prospectively validated 2B-Cool , a hardware/software system that automatically learns how individuals respond to heat stress and provides individualized estimates of T C , 20-min ahead predictions, and early warning of a rising T C . METHODS: We performed a crossover heat stress study in an environmental chamber, involving 11 men and 11 women (mean ± SD age = 20 ± 2 yr) who performed three bouts of varying physical activities on a treadmill over a 7.5-h trial, each under four different clothing and environmental conditions. Subjects wore the 2B-Cool system, consisting of a smartwatch, which collected vital signs, and a paired smartphone, which housed machine-learning algorithms and used the vital sign data to make individualized real-time forecasts. Subjects also wore a chest strap heart rate sensor and a rectal probe for comparison purposes. RESULTS: We observed very good agreement between the 2B-Cool forecasts and the measured T C , with a mean bias of 0.16°C for T C estimates and nearly 75% of measurements falling within the 95% prediction intervals of ±0.62°C for the 20-min predictions. The early-warning system results for a 38.50°C threshold yielded a 98% sensitivity, an 81% specificity, a prediction horizon of 35 min, and a false alarm rate of 0.12 events per hour. We observed no sex differences in the measured or predicted peak T C . CONCLUSION: 2B-Cool provides early warning of a rising T C with a sufficient lead time to enable clinical interventions and to help reduce the risk of exertional heat illness.

Efficacy of two intermittent cooling strategies during prolonged work-rest intervals in the heat with personal protective gear compared with a control condition.

INTRODUCTION: Personal protective equipment (PPE) inhibits heat dissipation and elevates heat strain. Impaired cooling with PPE warrants investigation into practical strategies to improve work capacity and mitigate exertional heat illness. PURPOSE: Examine physiological and subjective effects of forearm immersion (FC), fan mist (MC), and passive cooling (PC) following three intermittent treadmill bouts while wearing PPE. METHODS: Twelve males (27 ± 6 years; 57.6 ± 6.2 ml/kg/min; 78.3 ± 8.1 kg; 183.1 ± 7.2 cm) performed three 50-min (10 min of 40%, 70%, 40%, 60%, 50% vVO2max) treadmill bouts in the heat (36 °C, 30% relative humidity). Thirty minutes of cooling followed each bout, using one of the three strategies per trial. Rectal temperature (Tcore), skin temperature (Tsk), heart rate (HR), heart rate recovery (HRR), rating of perceived exertion (RPE), thirst, thermal sensation (TS), and fatigue were obtained. Repeated-measures analysis of variance (condition x time) detected differences between interventions. RESULTS: Final Tcore was similar between trials (P > .05). Cooling rates were larger in FC and MC vs PC following bout one (P < .05). HRR was greatest in FC following bouts two (P = .013) and three (P < .001). Tsk, fluid consumption, and sweat rate were similar between all trials (P > .05). TS and fatigue during bout three were lower in MC, despite similar Tcore and HR. CONCLUSION: Utilizing FC and MC during intermittent work in the heat with PPE yields some thermoregulatory and cardiovascular benefit, but military health and safety personnel should explore new and novel strategies to mitigate risk and maximize performance under hot conditions while wearing PPE.

Development and validation of a mathematical model to simulate human cardiovascular and respiratory responses to battlefield trauma.

Mathematical models of human cardiovascular and respiratory systems provide a viable alternative to generate synthetic data to train artificial intelligence (AI) clinical decision-support systems and assess closed-loop control technologies, for military medical applications. However, existing models are either complex, standalone systems that lack the interface to other applications or fail to capture the essential features of the physiological responses to the major causes of battlefield trauma (i.e., hemorrhage and airway compromise). To address these limitations, we developed the cardio-respiratory (CR) model by expanding and integrating two previously published models of the cardiovascular and respiratory systems. We compared the vital signs predicted by the CR model with those from three models, using experimental data from 27 subjects in five studies, involving hemorrhage, fluid resuscitation, and respiratory perturbations. Overall, the CR model yielded relatively small root mean square errors (RMSEs) for mean arterial pressure (MAP; 20.88 mm Hg), end-tidal CO2 (ETCO2 ; 3.50 mm Hg), O2 saturation (SpO2 ; 3.40%), and arterial O2 pressure (PaO2 ; 10.06 mm Hg), but a relatively large RMSE for heart rate (HR; 70.23 beats/min). In addition, the RMSEs for the CR model were 3% to 10% smaller than the three other models for HR, 11% to 15% for ETCO2 , 0% to 33% for SpO2 , and 10% to 64% for PaO2 , while they were similar for MAP. In conclusion, the CR model balances simplicity and accuracy, while qualitatively and quantitatively capturing human physiological responses to battlefield trauma, supporting its use to train and assess emerging AI and control systems.

Enabling data-limited chemical bioactivity predictions through deep neural network transfer learning.

The main limitation in developing deep neural network (DNN) models to predict bioactivity properties of chemicals is the lack of sufficient assay data to train the network's classification layers. Focusing on feedforward DNNs that use atom- and bond-based structural fingerprints as input, we examined whether layers of a fully trained DNN based on large amounts of data to predict one property could be used to develop DNNs to predict other related or unrelated properties based on limited amounts of data. Hence, we assessed if and under what conditions the dense layers of a pre-trained DNN could be transferred and used for the development of another DNN associated with limited training data. We carried out a quantitative study employing more than 400 pairs of assay datasets, where we used fully trained layers from a large dataset to augment the training of a small dataset. We found that the higher the correlation r between two assay datasets, the more efficient the transfer learning is in reducing prediction errors associated with the smaller dataset DNN predictions. The reduction in mean squared prediction errors ranged from 10 to 20% for every 0.1 increase in r2 between the datasets, with the bulk of the error reductions associated with transfers of the first dense layer. Transfer of other dense layers did not result in additional benefits, suggesting that deeper, dense layers conveyed more specialized and assay-specific information. Importantly, depending on the dataset correlation, training sample size could be reduced by up to tenfold without any loss of prediction accuracy.

Individuals with and without military-related PTSD differ in subjective sleepiness and alertness but not objective sleepiness.

Posttraumatic stress disorder-related sleep disturbances may increase daytime sleepiness and compromise performance in individuals with posttraumatic stress disorder. We investigated nighttime sleep predictors of sleepiness in Veterans with and without posttraumatic stress disorder. Thirty-seven post-9/11 Veterans with posttraumatic stress disorder and 47 without posttraumatic stress disorder (Control) completed a 48-h lab stay. Nighttime quantitative EEG and sleep architecture parameters were collected with polysomnography. Data from daytime sleepiness batteries assessing subjective sleepiness (global vigor questionnaire), objective sleepiness (Multiple Sleep Latency Tests) and alertness (psychomotor vigilance task) were included in analyses. Independent samples t-tests and linear regressions were performed to identify group differences in sleepiness and nighttime sleep predictors of sleepiness in the overall sample and within each group. Participants with posttraumatic stress disorder had higher subjective sleepiness (t = 4.20; p < .001) and lower alertness (psychomotor vigilance task reaction time (t = -3.70; p < .001) and lapses: t = -2.13; p = .04) than the control group. Objective daytime sleepiness did not differ between groups (t = -0.79, p = .43). In the whole sample, higher rapid eye movement delta power predicted lower alertness quantified by psychomotor vigilance task reaction time (ß = 0.372, p = .013) and lapses (ß = 0.388, p = .013). More fragmented sleep predicted higher objective sleepiness in the posttraumatic stress disorder group (ß = -.467, p = .005) but no other nighttime sleep measures influenced the relationship between group and sleepiness. Objective measures of sleep and sleepiness were not associated with the increased subjective sleepiness and reduced alertness of the posttraumatic stress disorder group.

A 3-D virtual human thermoregulatory model to predict whole-body and organ-specific heat-stress responses.

OBJECTIVE: This study aimed at assessing the risks associated with human exposure to heat-stress conditions by predicting organ- and tissue-level heat-stress responses under different exertional activities, environmental conditions, and clothing. METHODS: In this study, we developed an anatomically detailed three-dimensional thermoregulatory finite element model of a 50th percentile U.S. male, to predict the spatiotemporal temperature distribution throughout the body. The model accounts for the major heat transfer and thermoregulatory mechanisms, and circadian-rhythm effects. We validated our model by comparing its temperature predictions of various organs (brain, liver, stomach, bladder, and esophagus), and muscles (vastus medialis and triceps brachii) under normal resting conditions (errors between 0.0 and 0.5 °C), and of rectum under different heat-stress conditions (errors between 0.1 and 0.3 °C), with experimental measurements from multiple studies. RESULTS: Our simulations showed that the rise in the rectal temperature was primarily driven by the activity level (~ 94%) and, to a much lesser extent, environmental conditions or clothing considered in our study. The peak temperature in the heart, liver, and kidney were consistently higher than in the rectum (by ~ 0.6 °C), and the entire heart and liver recorded higher temperatures than in the rectum, indicating that these organs may be more susceptible to heat injury. CONCLUSION: Our model can help assess the impact of exertional and environmental heat stressors at the organ level and, in the future, evaluate the efficacy of different whole-body or localized cooling strategies in preserving organ integrity.

Identification of Veterans With PTSD Based on EEG Features Collected During Sleep.

Background: Previously, we identified sleep-electroencephalography (EEG) spectral power and synchrony features that differed significantly at a population-average level between subjects with and without posttraumatic stress disorder (PTSD). Here, we aimed to examine the extent to which a combination of such features could objectively identify individual subjects with PTSD. Methods: We analyzed EEG data recorded from 78 combat-exposed Veteran men with (n = 31) and without (n = 47) PTSD during two consecutive nights of sleep. To obviate the need for manual assessment of sleep staging and facilitate extraction of features from the EEG data, for each subject, we computed 780 stage-independent, whole-night features from the 10 most commonly used EEG channels. We performed feature selection and trained a logistic regression model using a training set consisting of the first 47 consecutive subjects (18 with PTSD) of the study. Then, we evaluated the model on a testing set consisting of the remaining 31 subjects (13 with PTSD). Results: Feature selection yielded three uncorrelated features that were consistent across the two consecutive nights and discriminative of PTSD. One feature was from the spectral power in the delta band (2-4 Hz) and the other two were from phase synchronies in the alpha (10-12 Hz) and gamma (32-40 Hz) bands. When we combined these features into a logistic regression model to predict the subjects in the testing set, the trained model yielded areas under the receiver operating characteristic curve of at least 0.80. Importantly, the model yielded a testing-set sensitivity of 0.85 and a positive predictive value (PPV) of 0.31. Conclusions: We identified robust stage-independent, whole-night features from EEG signals and combined them into a logistic regression model to discriminate subjects with and without PTSD. On the testing set, the model yielded a high sensitivity and a PPV that was twice the prevalence rate of PTSD in the U.S. Veteran population. We conclude that, using EEG signals collected during sleep, such a model can potentially serve as a means to objectively identify U.S. Veteran men with PTSD.

Inter-channel phase differences during sleep spindles are altered in Veterans with PTSD.

Sleep disturbances are common complaints in patients with post-traumatic stress disorder (PTSD). To date, however, objective markers of PTSD during sleep remain elusive. Sleep spindles are distinctive bursts of brain oscillatory activity during non-rapid eye movement (NREM) sleep and have been implicated in sleep protection and sleep-dependent memory processes. In healthy sleep, spindles observed in electroencephalogram (EEG) data are highly synchronized across different regions of the scalp. Here, we aimed to investigate whether the spatiotemporal synchronization patterns between EEG channels during sleep spindles, as quantified by the phase-locking value (PLV) and the mean phase difference (MPD), are altered in PTSD. Using high-density (64-channel) EEG data recorded from 78 combat-exposed Veteran men (31 with PTSD and 47 without PTSD) during two consecutive nights of sleep, we examined group differences in the PLV and MPD for slow (10-13 Hz) and fast (13-16 Hz) spindles separately. To evaluate the reproducibility of our findings, we set apart the first 47 consecutive participants (18 with PTSD) for the initial discovery and reserved the remaining 31 participants (13 with PTSD) for replication analysis. In the discovery analysis, compared to the non-PTSD group, the PTSD group showed smaller MPDs during slow spindles between the frontal and centro-parietal channel pairs on both nights. We obtained reproducible results in the replication analysis in terms of statistical significance and effect size. The PLVs during slow or fast spindles did not significantly differ between groups. The reduced inter-channel phase difference during slow spindles in PTSD may reflect pathological changes in the underlying thalamocortical circuits. This novel finding, if independently validated, may prove useful in developing sleep-focused PTSD diagnostics and interventions.

Increased oscillatory frequency of sleep spindles in combat-exposed veteran men with post-traumatic stress disorder.

STUDY OBJECTIVES: Sleep disturbances are core symptoms of post-traumatic stress disorder (PTSD), but reliable sleep markers of PTSD have yet to be identified. Sleep spindles are important brain waves associated with sleep protection and sleep-dependent memory consolidation. The present study tested whether sleep spindles are altered in individuals with PTSD and whether the findings are reproducible across nights and subsamples of the study. METHODS: Seventy-eight combat-exposed veteran men with (n = 31) and without (n = 47) PTSD completed two consecutive nights of high-density EEG recordings in a laboratory. We identified slow (10-13 Hz) and fast (13-16 Hz) sleep spindles during N2 and N3 sleep stages and performed topographical analyses of spindle parameters (amplitude, duration, oscillatory frequency, and density) on both nights. To assess reproducibility, we used the first 47 consecutive participants (18 with PTSD) for initial discovery and the remaining 31 participants (13 with PTSD) for replication assessment. RESULTS: In the discovery analysis, compared to non-PTSD participants, PTSD participants exhibited (1) higher slow-spindle oscillatory frequency over the antero-frontal regions on both nights and (2) higher fast-spindle oscillatory frequency over the centro-parietal regions on the second night. The first finding was preserved in the replication analysis. We found no significant group differences in the amplitude, duration, or density of slow or fast spindles. CONCLUSIONS: The elevated spindle oscillatory frequency in PTSD may indicate a deficient sensory-gating mechanism responsible for preserving sleep continuity. Our findings, if independently validated, may assist in the development of sleep-focused PTSD diagnostics and interventions.

Alterations in sleep electroencephalography synchrony in combat-exposed veterans with post-traumatic stress disorder.

STUDY OBJECTIVES: We assessed whether the synchrony between brain regions, analyzed using electroencephalography (EEG) signals recorded during sleep, is altered in subjects with post-traumatic stress disorder (PTSD) and whether the results are reproducible across consecutive nights and subpopulations of the study. METHODS: A total of 78 combat-exposed veteran men with (n = 31) and without (n = 47) PTSD completed two consecutive laboratory nights of high-density EEG recordings. We computed a measure of synchrony for each EEG channel-pair across three sleep stages (rapid eye movement [REM] and non-REM stages 2 and 3) and six frequency bands. We examined the median synchrony in 9 region-of-interest (ROI) pairs consisting of 6 bilateral brain regions (left and right frontal, central, and parietal regions) for 10 frequency-band and sleep-stage combinations. To assess reproducibility, we used the first 47 consecutive subjects (18 with PTSD) for initial discovery and the remaining 31 subjects (13 with PTSD) for replication. RESULTS: In the discovery analysis, five alpha-band synchrony pairs during non-REM sleep were consistently larger in PTSD subjects compared with controls (effect sizes ranging from 0.52 to 1.44) across consecutive nights: two between the left-frontal and left-parietal ROIs, one between the left-central and left-parietal ROIs, and two across central and parietal bilateral ROIs. These trends were preserved in the replication set. CONCLUSION: PTSD subjects showed increased alpha-band synchrony during non-REM sleep in the left frontoparietal, left centro-parietal, and inter-parietal brain regions. Importantly, these trends were reproducible across consecutive nights and subpopulations. Thus, these alterations in alpha synchrony may be discriminatory of PTSD.

An attempt to identify reproducible high-density EEG markers of PTSD during sleep.

STUDY OBJECTIVES: We examined electroencephalogram (EEG) spectral power to study abnormalities in regional brain activity in post-traumatic stress disorder (PTSD) during sleep. We aimed to identify sleep EEG markers of PTSD that were reproducible across nights and subsamples of our study population. METHODS: Seventy-eight combat-exposed veteran men with (n = 31) and without (n = 47) PTSD completed two consecutive nights of high-density EEG recordings in a laboratory. We performed spectral-topographical EEG analyses on data from both nights. To assess reproducibility, we used the first 47 consecutive participants (18 with PTSD) for initial discovery and the remaining 31 participants (13 with PTSD) for replication. RESULTS: In the discovery analysis, compared with non-PTSD participants, PTSD participants exhibited (1) reduced delta power (1-4 Hz) in the centro-parietal regions during nonrapid eye movement (NREM) sleep and (2) elevated high-frequency power, most prominent in the gamma band (30-40 Hz), in the antero-frontal regions during both NREM and rapid eye movement (REM) sleep. These findings were consistent across the two study nights, with reproducible trends in the replication analysis. We found no significant group differences in theta power (4-8 Hz) during REM sleep and sigma power (12-15 Hz) during N2 sleep. CONCLUSIONS: The reduced centro-parietal NREM delta power, indicating reduced sleep depth, and the elevated antero-frontal NREM and REM gamma powers, indicating heightened central arousal, are potential objective sleep markers of PTSD. If independently validated, these putative EEG markers may offer new targets for the development of sleep-specific PTSD diagnostics and interventions.

An increase in sleep slow waves predicts better working memory performance in healthy individuals.

Sleep is imperative for brain health and well-being, and restorative sleep is associated with better cognitive functioning. Increasing evidence indicates that electrophysiological measures of sleep, especially slow wave activity (SWA), regulate the consolidation of motor and perceptual procedural memory. In contrast, the role of sleep EEG and SWA in modulating executive functions, including working memory (WM), has been far less characterized. Here, we investigated across-night changes in sleep EEG that may ameliorate WM performance. Participants (N = 25, M = 100%) underwent two consecutive nights with high-density EEG, along with N-back tasks, which were administered at three time points the day before and after the second night of sleep. Non-rapid eye movement sleep EEG power spectra, power topography, as well as several slow-wave parameters were computed and compared across nights. Improvers on the 1-back, but not non-improvers, showed a significant increase in SWA as well as in down slope and negative peak amplitude, in a fronto-parietal region, and these parameters increases predicted better WM performance. Overall, these findings show that slow-wave sleep has a beneficial effect on WM and that it can occur in the adult brain even after minimal training. This is especially relevant, when considering that WM and other executive function cognitive deficits are present in several neuropsychiatric disorders, and that slow-wave enhancing interventions can improve cognition, thus providing novel insights and treatment strategies for these patients.

Individualized estimation of human core body temperature using noninvasive measurements.

A rising core body temperature (Tc) during strenuous physical activity is a leading indicator of heat-injury risk. Hence, a system that can estimate Tc in real time and provide early warning of an impending temperature rise may enable proactive interventions to reduce the risk of heat injuries. However, real-time field assessment of Tc requires impractical invasive technologies. To address this problem, we developed a mathematical model that describes the relationships between Tc and noninvasive measurements of an individual's physical activity, heart rate, and skin temperature, and two environmental variables (ambient temperature and relative humidity). A Kalman filter adapts the model parameters to each individual and provides real-time personalized Tc estimates. Using data from three distinct studies, comprising 166 subjects who performed treadmill and cycle ergometer tasks under different experimental conditions, we assessed model performance via the root mean squared error (RMSE). The individualized model yielded an overall average RMSE of 0.33 (SD = 0.18)°C, allowing us to reach the same conclusions in each study as those obtained using the Tc measurements. Furthermore, for 22 unique subjects whose Tc exceeded 38.5°C, a potential lower Tc limit of clinical relevance, the average RMSE decreased to 0.25 (SD = 0.20)°C. Importantly, these results remained robust in the presence of simulated real-world operational conditions, yielding no more than 16% worse RMSEs when measurements were missing (40%) or laden with added noise. Hence, the individualized model provides a practical means to develop an early warning system for reducing heat-injury risk. NEW & NOTEWORTHY A model that uses an individual's noninvasive measurements and environmental variables can continually "learn" the individual's heat-stress response by automatically adapting the model parameters on the fly to provide real-time individualized core body temperature estimates. This individualized model can replace impractical invasive sensors, serving as a practical and effective surrogate for core temperature monitoring.

Effects of signal artefacts on electroencephalography spectral power during sleep: quantifying the effectiveness of automated artefact-rejection algorithms.

Electroencephalography (EEG) recordings during sleep are often contaminated by muscle and ocular artefacts, which can affect the results of spectral power analyses significantly. However, the extent to which these artefacts affect EEG spectral power across different sleep states has not been quantified explicitly. Consequently, the effectiveness of automated artefact-rejection algorithms in minimizing these effects has not been characterized fully. To address these issues, we analysed standard 10-channel EEG recordings from 20 subjects during one night of sleep. We compared their spectral power when the recordings were contaminated by artefacts and after we removed them by visual inspection or by using automated artefact-rejection algorithms. During both rapid eye movement (REM) and non-REM (NREM) sleep, muscle artefacts contaminated no more than 5% of the EEG data across all channels. However, they corrupted delta, beta and gamma power levels substantially by up to 126, 171 and 938%, respectively, relative to the power level computed from artefact-free data. Although ocular artefacts were infrequent during NREM sleep, they affected up to 16% of the frontal and temporal EEG channels during REM sleep, primarily corrupting delta power by up to 33%. For both REM and NREM sleep, the automated artefact-rejection algorithms matched power levels to within ~10% of the artefact-free power level for each EEG channel and frequency band. In summary, although muscle and ocular artefacts affect only a small fraction of EEG data, they affect EEG spectral power significantly. This suggests the importance of using artefact-rejection algorithms before analysing EEG data.

Real-time individualization of the unified model of performance.

Existing mathematical models for predicting neurobehavioural performance are not suited for mobile computing platforms because they cannot adapt model parameters automatically in real time to reflect individual differences in the effects of sleep loss. We used an extended Kalman filter to develop a computationally efficient algorithm that continually adapts the parameters of the recently developed Unified Model of Performance (UMP) to an individual. The algorithm accomplishes this in real time as new performance data for the individual become available. We assessed the algorithm's performance by simulating real-time model individualization for 18 subjects subjected to 64 h of total sleep deprivation (TSD) and 7 days of chronic sleep restriction (CSR) with 3 h of time in bed per night, using psychomotor vigilance task (PVT) data collected every 2 h during wakefulness. This UMP individualization process produced parameter estimates that progressively approached the solution produced by a post-hoc fitting of model parameters using all data. The minimum number of PVT measurements needed to individualize the model parameters depended upon the type of sleep-loss challenge, with ~30 required for TSD and ~70 for CSR. However, model individualization depended upon the overall duration of data collection, yielding increasingly accurate model parameters with greater number of days. Interestingly, reducing the PVT sampling frequency by a factor of two did not notably hamper model individualization. The proposed algorithm facilitates real-time learning of an individual's trait-like responses to sleep loss and enables the development of individualized performance prediction models for use in a mobile computing platform.

Bolus Estimation--Rethinking the Effect of Meal Fat Content.

BACKGROUND: Traditionally, insulin bolus calculations for managing postprandial glucose levels in individuals with type 1 diabetes rely solely on the carbohydrate content of a meal. However, recent studies have reported that other macronutrients in a meal can alter the insulin required for good postprandial control. Specifically, studies have shown that high-fat (HF) meals require more insulin than low-fat (LF) meals with identical carbohydrate content. Our objective was to assess the mechanisms underlying the higher insulin requirement observed in one of these studies. MATERIALS AND METHODS: We used a combination of previously validated metabolic models to fit data from a study comparing HF and LF dinners with identical carbohydrate content in seven subjects with type 1 diabetes. For each subject and dinner type, we estimated the model parameters representing the time of peak meal-glucose appearance (τ(m)), insulin sensitivity (S(I)), the net hepatic glucose balance, and the glucose effect at zero insulin in four time windows (dinner, early night, late night, and breakfast) and assessed the differences in model parameters via paired Wilcoxon signed-rank tests. RESULTS: During the HF meal, the τ(m) was significantly delayed (mean and standard error [SE]: 102 [14] min vs. 71 [4] min; P = 0.02), and S(I) was significantly lower (7.25 × 10(-4) [1.29 × 10(-4)] mL/µU/min vs. 8.72 × 10(-4) [1.08 × 10(-4)] mL/µU/min; P = 0.02). CONCLUSIONS: In addition to considering the putative delay in gastric emptying associated with HF meals, we suggest that clinicians reviewing patient records consider that the fat content of these meals may alter S(I).

Can a mathematical model predict an individual's trait-like response to both total and partial sleep loss?

Humans display a trait-like response to sleep loss. However, it is not known whether this trait-like response can be captured by a mathematical model from only one sleep-loss condition to facilitate neurobehavioural performance prediction of the same individual during a different sleep-loss condition. In this paper, we investigated the extent to which the recently developed unified mathematical model of performance (UMP) captured such trait-like features for different sleep-loss conditions. We used the UMP to develop two sets of individual-specific models for 15 healthy adults who underwent two different sleep-loss challenges (order counterbalanced; separated by 2-4 weeks): (i) 64 h of total sleep deprivation (TSD) and (ii) chronic sleep restriction (CSR) of 7 days of 3 h nightly time in bed. We then quantified the extent to which models developed using psychomotor vigilance task data under TSD predicted performance data under CSR, and vice versa. The results showed that the models customized to an individual under one sleep-loss condition accurately predicted performance of the same individual under the other condition, yielding, on average, up to 50% improvement over non-individualized, group-average model predictions. This finding supports the notion that the UMP captures an individual's trait-like response to different sleep-loss conditions.

Human Core Temperature Prediction for Heat-Injury Prevention.

Previously, our group developed autoregressive (AR) models to predict human core temperature and help prevent hyperthermia (temperature > 39°C). However, the models often yielded delayed predictions, limiting their application as a real-time warning system. To mitigate this problem, here we combined AR-model point estimates with statistically derived prediction intervals (PIs) and assessed the performance of three new alert algorithms [AR model plus PI, median filter of AR model plus PI decisions, and an adaptation of the sequential probability ratio test (SPRT)]. Using field-study data from 22 soldiers, including five subjects who experienced hyperthermia, we assessed the alert algorithms for AR-model prediction windows from 15-30 min. Cross-validation simulations showed that, as the prediction windows increased, improvements in the algorithms' effective prediction horizons were offset by deteriorating accuracy, with a 20-min window providing a reasonable compromise. Model plus PI and SPRT yielded the largest effective prediction horizons (≥18 min), but these were offset by other performance measures. If high sensitivity and a long effective prediction horizon are desired, model plus PI provides the best choice, assuming decision switches can be tolerated. In contrast, if a small number of decision switches are desired, SPRT provides the best compromise as an early warning system of impending heat illnesses.