Development and Validation of a Model to Identify Critical Brain Injuries Using Natural Language Processing of Text Computed Tomography Reports.

Importance: Clinical text reports from head computed tomography (CT) represent rich, incompletely utilized information regarding acute brain injuries and neurologic outcomes. CT reports are unstructured; thus, extracting information at scale requires automated natural language processing (NLP). However, designing new NLP algorithms for each individual injury category is an unwieldy proposition. An NLP tool that summarizes all injuries in head CT reports would facilitate exploration of large data sets for clinical significance of neuroradiological findings. Objective: To automatically extract acute brain pathological data and their features from head CT reports. Design, Setting, and Participants: This diagnostic study developed a 2-part named entity recognition (NER) NLP model to extract and summarize data on acute brain injuries from head CT reports. The model, termed BrainNERD, extracts and summarizes detailed brain injury information for research applications. Model development included building and comparing 2 NER models using a custom dictionary of terms, including lesion type, location, size, and age, then designing a rule-based decoder using NER outputs to evaluate for the presence or absence of injury subtypes. BrainNERD was evaluated against independent test data sets of manually classified reports, including 2 external validation sets. The model was trained on head CT reports from 1152 patients generated by neuroradiologists at the Yale Acute Brain Injury Biorepository. External validation was conducted using reports from 2 outside institutions. Analyses were conducted from May 2020 to December 2021. Main Outcomes and Measures: Performance of the BrainNERD model was evaluated using precision, recall, and F1 scores based on manually labeled independent test data sets. Results: A total of 1152 patients (mean [SD] age, 67.6 [16.1] years; 586 [52%] men), were included in the training set. NER training using transformer architecture and bidirectional encoder representations from transformers was significantly faster than spaCy. For all metrics, the 10-fold cross-validation performance was 93% to 99%. The final test performance metrics for the NER test data set were 98.82% (95% CI, 98.37%-98.93%) for precision, 98.81% (95% CI, 98.46%-99.06%) for recall, and 98.81% (95% CI, 98.40%-98.94%) for the F score. The expert review comparison metrics were 99.06% (95% CI, 97.89%-99.13%) for precision, 98.10% (95% CI, 97.93%-98.77%) for recall, and 98.57% (95% CI, 97.78%-99.10%) for the F score. The decoder test set metrics were 96.06% (95% CI, 95.01%-97.16%) for precision, 96.42% (95% CI, 94.50%-97.87%) for recall, and 96.18% (95% CI, 95.151%-97.16%) for the F score. Performance in external institution report validation including 1053 head CR reports was greater than 96%. Conclusions and Relevance: These findings suggest that the BrainNERD model accurately extracted acute brain injury terms and their properties from head CT text reports. This freely available new tool could advance clinical research by integrating information in easily gathered head CT reports to expand knowledge of acute brain injury radiographic phenotypes.

Treatment Fidelity in a Randomized Controlled Trial of Cognitive Behavioral Therapy for Insomnia in Heart Failure.

BACKGROUND: Cognitive behavioral therapy for insomnia (CBT-I) is an efficacious treatment for people with chronic insomnia, including those with heart failure (HF). Treatment fidelity evaluation is needed to ensure study validity and reliability. OBJECTIVE: The aim of this study was to apply the National Institutes of Health Behavioral Change Consortium framework to ensure adequate treatment fidelity in a randomized controlled trial of CBT-I for people with stable HF. METHODS: We describe strategies to ensure treatment fidelity in study design, provider training, and treatment delivery. We measured treatment receipt (observation and self-report) and enactment of CBT-I strategies (self-report and actigraphy). We used the Dysfunctional Beliefs and Attitudes About Sleep Scale and the Sleep Disturbance Questionnaire to indicate sleep-related beliefs and cognitions. We computed descriptive statistics for demographic characteristics, treatment receipt, and enactment variables. We compared baseline and post-intervention dichotomous sleep behaviors using the two-sided Wilcoxon rank-sum test. We calculated the root mean square of successive difference in time of sleep onset and midpoint (actigraphy) to indicate day-by-day variability in bed and wake times. RESULTS: Most participants completed the CBT-I intervention and follow-up assessments and attended each group or make-up session. Most correctly computed their sleep efficiency and completed their homework. Most participants used the sleep tracker and reported using strategies to improve their sleep schedules. There was a significant decrease in actigraph-measured light intensity during the 30 minutes and 1 hour before bedtime between baseline and post-intervention and no statistically significant changes in light intensity after waking or in nap frequency. Most of those who woke during the night got out of bed, consistent with recommendations. There were significant improvements on all of the Dysfunctional Beliefs and Attitudes About Sleep items and total score and all but one item on the Sleep Disturbance Questionnaire. DISCUSSION: Preserved treatment fidelity may explain the large and sustained effects in people with HF found in our trial. Further research is needed to evaluate CBT-I's effectiveness and implementation strategies among people with HF in real-world clinical settings.

Anxiety Symptoms After Orthopedic Injury: A Systematic Review.

BACKGROUND: Millions of Americans are injured each year, with 75% suffering orthopedic injuries. Those with orthopedic injury often experience postinjury symptoms, such as anxiety, that negatively influence long-term outcomes. The purpose of this systematic review was to evaluate (1) the prevalence of anxiety symptoms chronologically in relation to time of orthopedic injury, (2) the associations of anxiety symptoms with other symptoms in individuals with orthopedic injury, and (3) the associations between anxiety symptoms and functional performance and mental health outcomes in individuals with orthopedic injury. METHODS: The databases CINAHL, Ovid MEDLINE, Ovid PsycInfo, Ovid Embase, ProQuest, and ClinicalTrials.gov were searched on June 25, 2020. Studies were selected for review if (1) participants were 16 years or older, (2) the prevalence of anxiety symptoms was listed, (3) anxiety symptoms were assessed using a validated instrument, (4) the study was published in English, and (5) more than 50% of the participants suffered orthopedic injury to the appendicular skeleton or pelvic ring. RESULTS: Anxiety symptoms were present in 11.5%-55.9% of individuals up to 10 years after orthopedic injury and often co-occurred with those of depression, pain, and posttraumatic stress disorder to negatively influence functional performance and mental health outcomes. CONCLUSIONS: Anxiety symptoms post-orthopedic injury are highly prevalent and persistent. Anxiety symptoms are associated with other distressing symptoms and influence long-term outcomes. These findings highlight the need to screen for and treat anxiety symptoms following orthopedic injury and may help in designing future self-management interventions.

Dopamine D2/3 receptor availability in cocaine use disorder individuals with obesity as measured by [11C]PHNO PET.

BACKGROUND: Positron emission tomography (PET) work with the dopamine D3 receptor (D3R) preferring ligand [11C]PHNO in obese individuals has demonstrated higher binding and positive correlations with body mass index (BMI) in otherwise healthy individuals. These findings implicated brain reward areas including the substantia nigra/ventral tegmental area (SN/VTA) and pallidum. In cocaine use disorder (CUD), similar SN/VTA binding profiles have been found compared to healthy control subjects. This study investigates whether BMI-[11C]PHNO relationships are similar in individuals with CUD. METHODS: Non-obese CUD subjects (N = 12) were compared to age-matched obese CUD subjects (N = 14). All subjects underwent [11C]PHNO acquisition using a High Resolution Research Tomograph PET scanner. Parametric images were computed using the simplified reference tissue model with cerebellum as the reference region. [11C]PHNO measures of receptor availability were calculated and expressed as non-displaceable binding potential (BPND). RESULTS: In between-group analyses, D2/3R availability in non-obese and obese CUD groups was not significantly different overall. BMI was inversely correlated withBPND in the SN/VTA (r = -0.45, p = 0.02 uncorrected) in all subjects. CONCLUSION: These data suggest that obesity in CUD was not associated with significant differences in D2/3R availability. This in contrast to previous findings in non-CUD individuals that found increased availability of D3Rs in the SN/VTA associated with obesity. These findings could potentially reflect dysregulation of D3R in CUD, impacting how affected individuals respond to natural stimuli such as food.

Evaluating Recruitment Strategies for a Randomized Clinical Trial with Heart Failure Patients.

Recruiting participants with chronic medical conditions is time-consuming and expensive. Electronic medical record databases and patient portals may enable outreach to larger numbers of patients in comparison to face-to-face methods. We aimed to describe the yields, benefits, and limitations of recruitment strategies used for a randomized clinical trial of cognitive behavioral therapy for insomnia among patients with chronic stable heart failure (NCT02660385). We used multiple recruitment strategies including clinic-based recruitment, letters to patients identified from electronic databases, the patient portal, brochures and posters placed in clinics, presentations to heart failure support groups, and online advertising. We screened 10,291 medical records, enrolled 231 participants, and 195 participants completed baseline data collection. We enrolled 92 (23%) participants using clinic-based recruitment, 24 and 29 (6% and 10%) using letters to patients from two electronic databases, and 42 (55%) via the patient portal. Multiple recruitment strategies and flexibility are needed to achieve recruitment goals.

In vivo 5-HT6 and 5-HT2A receptor availability in antipsychotic treated schizophrenia patients vs. unmedicated healthy humans measured with [11C]GSK215083 PET.

While 5-HT6 receptor is a potential therapeutic target for cognitive impairment in schizophrenia (SCZ), in vivo 5-HT6 receptor availability following antipsychotic treatment has not been examined to-date. We examined the availability of 5-HT6 and 5-HT2A receptors following treatment with olanzapine, risperidone, aripiprazole and quetiapine in male patients with SCZ vs unmedicated age-matched healthy male controls (HC) using positron emission tomography (PET) imaging with [11C]GSK215083. [11C]GSK215083 has been shown to have selectivity for 5-HT6 in the striatum and 5-HT2A in the cortex. Patients with SCZ (n = 9) were scanned with [11C]GSK215083 on HR+ PET scanner at presumed steady-state trough and peak serum levels following 7 days of confirmed inpatient antipsychotic treatment. Time-activity curves in regions-of-interest were fitted with multilinear analysis-1 (MA1). Regional nondisplaceable binding potential (BPND) values were calculated using cerebellum as the reference region and corrected for partial volume effects. Compared to HCs (n = 9), olanzapine was associated with significantly lower BPND (range: 53%-95%) in ventral striatum, putamen, caudate and frontal cortex at both trough and peak scans. Risperidone was associated with significantly lower BPND in frontal cortex at both trough and peak scans. The study provides preliminary evidence that treatment with different second-generation antipsychotics results in differing profiles of 5-HT2A and 5-HT6 availability.

Effects of age, BMI and sex on the glial cell marker TSPO - a multicentre [11C]PBR28 HRRT PET study.

PURPOSE: The purpose of this study was to investigate the effects of ageing, sex and body mass index (BMI) on translocator protein (TSPO) availability in healthy subjects using positron emission tomography (PET) and the radioligand [11C]PBR28. METHODS: [11C]PBR28 data from 140 healthy volunteers (72 males and 68 females; N = 78 with HAB and N = 62 MAB genotype; age range 19-80 years; BMI range 17.6-36.9) were acquired with High Resolution Research Tomograph at three centres: Karolinska Institutet (N = 53), Turku PET centre (N = 62) and Yale University PET Center (N = 25). The total volume of distribution (VT) was estimated in global grey matter, frontal, temporal, occipital and parietal cortices, hippocampus and thalamus using multilinear analysis 1. The effects of age, BMI and sex on TSPO availability were investigated using linear mixed effects model, with TSPO genotype and PET centre specified as random intercepts. RESULTS: There were significant positive correlations between age and VT in the frontal and temporal cortex. BMI showed a significant negative correlation with VT in all regions. Additionally, significant differences between males and females were observed in all regions, with females showing higher VT. A subgroup analysis revealed a positive correlation between VT and age in all regions in male subjects, whereas age showed no effect on TSPO levels in female subjects. CONCLUSION: These findings provide evidence that individual biological properties may contribute significantly to the high variation shown in TSPO binding estimates, and suggest that age, BMI and sex can be confounding factors in clinical studies.

Age-Related Change in 5-HT6 Receptor Availability in Healthy Male Volunteers Measured with 11C-GSK215083 PET.

Serotonin receptor 6 (5-hydroxytrypamine-6, or 5-HT6) is a potential therapeutic target given its distribution in brain regions that are important in depression, anxiety, and cognition. This study sought to investigate the effects of age on 5-HT6 receptor availability using 11C-GSK215083, a PET ligand with affinity for 5-HT6 in the striatum and 5-HT2A in the cortex. Methods: Twenty-eight healthy male volunteers (age range, 23-52 y) were scanned with 11C-GSK215083 PET. Time-activity curves in regions of interest were fitted using a multilinear analysis method. Nondisplaceable binding potential (BPND) was calculated using the cerebellum as the reference region and corrected for partial-volume effects. Results: In 5-HT6-rich areas, regional 11C-GSK215083 showed a negative correlation between BPND and age in the caudate (r = -0.41, P = 0.03) (14% change per decade) and putamen (r = -0.30, P = 0.04) (11% change per decade) but not in the ventral striatum or pallidum. A negative correlation with age was also seen in cortical regions (r = -0.41, P = 0.03) (7% change per decade), consistent with the literature on 5-HT2A availability. Conclusion: To our knowledge, this was the first in vivo study on humans to examine the effect of age on 5-HT6 receptor availability. The study demonstrated a significant age-related decline in 5-HT6 availability (BPND) in the caudate and putamen.

Regional and source-based patterns of [11C]-(+)-PHNO binding potential reveal concurrent alterations in dopamine D2 and D3 receptor availability in cocaine-use disorder.

Dopamine type 2 and type 3 receptors (D2R/D3R) appear critical to addictive disorders. Cocaine-use disorder (CUD) is associated with lower D2R availability and greater D3R availability in regions primarily expressing D2R or D3R concentrations, respectively. However, these CUD-related alterations in D2R and D3R have not been concurrently detected using available dopaminergic radioligands. Furthermore, receptor availability in regions of mixed D2R/D3R concentration in CUD remains unclear. The current study aimed to extend investigations of CUD-related alterations in D2R and D3R availability using regional and source-based analyses of [11C]-(+)-PHNO positron emission tomography (PET) of 26 individuals with CUD and 26 matched healthy comparison (HC) participants. Regional analysis detected greater binding potential (BPND) in CUD in the midbrain, consistent with prior [11C]-(+)-PHNO research, and lower BPND in CUD in the dorsal striatum, consistent with research using non-selective D2R/D3R radiotracers. Exploratory independent component analysis (ICA) identified three sources of BPND (striatopallidal, pallidonigral, and mesoaccumbens sources) that represent systems of brain regions displaying coherent variation in receptor availability. The striatopallidal source was associated with estimates of regional D2R-related proportions of BPND (calculated using independent reports of [11C]-(+)-PHNO receptor binding fractions), was lower in intensity in CUD and negatively associated with years of cocaine use. By comparison, the pallidonigral source was associated with estimates of regional D3R distribution, was greater in intensity in CUD and positively associated with years of cocaine use. The current study extends previous D2R/D3R research in CUD, demonstrating both lower BPND in the D2R-rich dorsal striatum and greater BPND in the D3R-rich midbrain using a single radiotracer. In addition, exploratory ICA identified sources of [11C]-(+)-PHNO BPND that were correlated with regional estimates of D2R-related and D3R-related proportions of BPND, were consistent with regional differences in CUD, and suggest receptor alterations in CUD may also be present in regions of mixed D2R/D3R concentration.

Elevated Dopamine D2/3 Receptor Availability in Obese Individuals: A PET Imaging Study with [11C](+)PHNO.

Most prior work with positron emission tomography (PET) dopamine subtype 2/3 receptor (D2/3R) non-selective antagonist tracers suggests that obese (OB) individuals exhibit lower D2/3Rs when compared with normal weight (NW) individuals. A D3-preferring D2/3R agonist tracer, [11C](+)PHNO, has demonstrated that body mass index (BMI) was positively associated with D2/3R availability within striatal reward regions. To date, OB individuals have not been studied with [11C](+)PHNO. We assessed D2/3R availability in striatal and extrastriatal reward regions in 14 OB and 14 age- and gender-matched NW individuals with [11C](+)PHNO PET utilizing a high-resolution research tomograph. Additionally, in regions where group D2/3R differences were observed, secondary analyses of 42 individuals that constituted an overweight cohort was done to study the linear association between BMI and D2/3R availability in those respective regions. A group-by-brain region interaction effect (F7, 182=2.08, p=0.047) was observed. Post hoc analyses revealed that OB individuals exhibited higher tracer binding in D3-rich regions: the substantia nigra/ventral tegmental area (SN/VTA) (+20%; p=0.02), ventral striatum (VST) (+14%; p<0.01), and pallidum (+11%; p=0.02). BMI was also positively associated with D2/3R availability in the SN/VTA (r=0.34, p=0.03), VST (r=0.36, p=0.02), and pallidum (r=0.30, p=0.05) across all subjects. These data suggest that individuals who are obese have higher D2/3R availability in brain reward regions densely populated with D3Rs, potentially identifying a novel pharmacologic target for the treatment of obesity.

Age-related changes in binding of the D2/3 receptor radioligand [(11)C](+)PHNO in healthy volunteers.

OBJECTIVE: Previous imaging studies with positron emission tomography (PET) have reliably demonstrated an age-associated decline in the dopamine system. Most of these studies have focused on the densities of dopamine receptor subtypes D2/3R (D2R family) in the striatum using antagonist radiotracers that are largely nonselective for D2R vs. D3R subtypes. Therefore, less is known about any possible age effects in D3-rich extrastriatal areas such as the substantia nigra/ventral tegmental area (SN/VTA) and hypothalamus. This study sought to investigate whether the receptor availability measured with [(11)C](+)PHNO, a D3R-preferring agonist radiotracer, also declines with age. METHODS: Forty-two healthy control subjects (9 females, 33 males; age range 19-55 years) were scanned with [(11)C](+)PHNO using a High Resolution Research Tomograph (HRRT). Parametric images were computed using the simplified reference tissue model (SRTM2) with cerebellum as the reference region. Binding potentials (BPND) were calculated for the amygdala, caudate, hypothalamus, pallidum, putamen, SN/VTA, thalamus, and ventral striatum and then confirmed at the voxel level with whole-brain parametric images. RESULTS: Regional [(11)C](+)PHNO BPND displayed a negative correlation between receptor availability and age in the caudate (r=-0.56, corrected p=0.0008) and putamen (r=-0.45, corrected p=0.02) in healthy subjects (respectively 8% and 5% lower per decade). No significant correlations with age were found between age and other regions (including the hypothalamus and SN/VTA). Secondary whole-brain voxel-wise analysis confirmed these ROI findings of negative associations and further identified a positive correlation in midbrain (SN/VTA) regions. CONCLUSION: In accordance with previous studies, the striatum (an area rich in D2R) is associated with age-related declines of the dopamine system. We did not initially find evidence of changes with age in the SN/VTA and hypothalamus, areas previously found to have a predominantly D3R signal as measured with [(11)C](+)PHNO. A secondary analysis did find a significant positive correlation in midbrain (SN/VTA) regions, indicating that there may be differential effects of aging, whereby D2R receptor availability decreases with age while D3R availability stays unchanged or is increased.

Domain Adaptation Methods for Improving Lab-to-field Generalization of Cocaine Detection using Wearable ECG.

Mobile health research on illicit drug use detection typically involves a two-stage study design where data to learn detectors is first collected in lab-based trials, followed by a deployment to subjects in a free-living environment to assess detector performance. While recent work has demonstrated the feasibility of wearable sensors for illicit drug use detection in the lab setting, several key problems can limit lab-to-field generalization performance. For example, lab-based data collection often has low ecological validity, the ground-truth event labels collected in the lab may not be available at the same level of temporal granularity in the field, and there can be significant variability between subjects. In this paper, we present domain adaptation methods for assessing and mitigating potential sources of performance loss in lab-to-field generalization and apply them to the problem of cocaine use detection from wearable electrocardiogram sensor data.

A preliminary study of dopamine D2/3 receptor availability and social status in healthy and cocaine dependent humans imaged with [(11)C](+)PHNO.

BACKGROUND: Previous work in healthy non-human primates and humans has shown that social status correlates positively with dopamine 2/3 receptor (D2/3R) availability imaged with antagonist radioligands and positron emission tomography (PET). Further work in non-human primates suggests that this relationship is disrupted by chronic cocaine administration. This exploratory study examined the relationship between social status and D2/3R availability in healthy (HH) and cocaine dependent (CD) humans using the D3-preferring, agonist radioligand, [(11)C](+)PHNO. METHODS: Sixteen HH and sixteen CD individuals completed the Barratt Simplified Measure of Social Status (BSMSS) and underwent [(11)C](+)PHNO scanning to measure regional brain D2/3R binding potentials (BPND). Correlations between BPND and BSMSS scores were then assessed within each group. RESULTS: Within HH and CD groups, inverse associations between BSMSS score and BPND were observed in the substantia nigra/ventral tegmental area (SN/VTA) and the ventral striatum, and for the CD group alone, the amygdala. After adjusting for body mass index and age, negative correlations remained significant in the SN/VTA for HH and in the amygdala for CD subjects. CONCLUSION: These preliminary data utilizing a dopamine agonist tracer demonstrate, for the first time, an inverse association between social status and D2/3R availability in the D3R rich extrastriatal regions of HH and CD humans.

Sex differences in the association between dietary restraint, insulin resistance and obesity.

BACKGROUND & AIMS: Restrained food consumption may alter metabolic function and contribute to eventual weight gain; however, sex differences in these relationships have not been assessed. The objective of this study was to examine the relationship between restrained eating and insulin resistance and the influence of body mass index and sex on this relationship in a large community sample of both men and women. We hypothesized that restrained eating would be related to insulin resistance and this relationship would be influenced by sex and body mass index. METHODS: In this cross-sectional, observational study, we studied 487 individuals from the community (men N = 222, women N = 265), who ranged from lean (body mass index 18.5-24.9 kg/m(2), N = 173), overweight (body mass index 25-29.9 kg/m(2), N = 159) to obese (body mass index >30 kg/m(2), N = 155) weight categories. We assessed restrained eating using the Dutch Eating Behavior Questionnaire and obtained fasting morning plasma insulin and glucose on all subjects. RESULTS: In men, but not in women, restrained eating was related to homeostatic model assessment of insulin resistance (HOMA-IR) (p < 0.0001). Furthermore, HOMA-IR was significantly higher in men who were high- versus low-restrained eaters (p = 0.0006). CONCLUSIONS: This study is the first to report sex differences with regard to the relationship between restrained eating and insulin resistance. Our results suggest that high restrained eating is associated with insulin resistance in men but not in women.

Dopamine D3 receptor alterations in cocaine-dependent humans imaged with [¹¹C](+)PHNO.

BACKGROUND: Evidence from animal models and postmortem human studies points to the importance of the dopamine D3 receptor (D3R) in cocaine dependence (CD). The objective of this pilot study was to use the D3R-preferring radioligand [(11)C](+)PHNO to compare receptor availability in groups with and without CD. METHODS: Ten medically healthy, non-treatment seeking CD subjects (mean age 41 ± 8) in early abstinence were compared to 10 healthy control (HC) subjects (mean age 41 ± 6) with no history of cocaine or illicit substance abuse. Binding potential (BPND), a measure of available receptors, was determined with parametric images, computed using the simplified reference tissue model (SRTM2) with the cerebellum as the reference region. RESULTS: BPND in CD subjects was higher in D3R-rich areas including the substantia nigra ((SN) 29%; P=0.03), hypothalamus (28%; P=0.02) and amygdala (35%; P=0.03). No between-group differences were observed in the striatum or pallidum. BPND values in the SN (r=+0.83; P=0.008) and pallidum (r=+0.67; P=0.03) correlated with years of cocaine use. CONCLUSIONS: Between-group differences suggest an important role for dopaminergic transmission in the SN, hypothalamus and amygdala in CD. Such findings also highlight the potential relevance of D3R as a medication development target in CD.

Conditional Random Fields for Morphological Analysis of Wireless ECG Signals.

Thanks to advances in mobile sensing technologies, it has recently become practical to deploy wireless electrocardiograph sensors for continuous recording of ECG signals. This capability has diverse applications in the study of human health and behavior, but to realize its full potential, new computational tools are required to effectively deal with the uncertainty that results from the noisy and highly non-stationary signals collected using these devices. In this work, we present a novel approach to the problem of extracting the morphological structure of ECG signals based on the use of dynamically structured conditional random field (CRF) models. We apply this framework to the problem of extracting morphological structure from wireless ECG sensor data collected in a lab-based study of habituated cocaine users. Our results show that the proposed CRF-based approach significantly out-performs independent prediction models using the same features, as well as a widely cited open source toolkit.