Development and validation of bone-suppressed deep learning classification of COVID-19 presentation in chest radiographs.

Background: Coronavirus disease 2019 (COVID-19) is a pandemic disease. Fast and accurate diagnosis of COVID-19 from chest radiography may enable more efficient allocation of scarce medical resources and hence improved patient outcomes. Deep learning classification of chest radiographs may be a plausible step towards this. We hypothesize that bone suppression of chest radiographs may improve the performance of deep learning classification of COVID-19 phenomena in chest radiographs. Methods: Two bone suppression methods (Gusarev et al. and Rajaraman et al.) were implemented. The Gusarev and Rajaraman methods were trained on 217 pairs of normal and bone-suppressed chest radiographs from the X-ray Bone Shadow Suppression dataset (https://www.kaggle.com/hmchuong/xray-bone-shadow-supression). Two classifier methods with different network architectures were implemented. Binary classifier models were trained on the public RICORD-1c and RSNA Pneumonia Challenge datasets. An external test dataset was created retrospectively from a set of 320 COVID-19 positive patients from Queen Elizabeth Hospital (Hong Kong, China) and a set of 518 non-COVID-19 patients from Pamela Youde Nethersole Eastern Hospital (Hong Kong, China), and used to evaluate the effect of bone suppression on classifier performance. Classification performance, quantified by sensitivity, specificity, negative predictive value (NPV), accuracy and area under the receiver operating curve (AUC), for non-suppressed radiographs was compared to that for bone suppressed radiographs. Some of the pre-trained models used in this study are published at (https://github.com/danielnflam). Results: Bone suppression of external test data was found to significantly (P<0.05) improve AUC for one classifier architecture [from 0.698 (non-suppressed) to 0.732 (Rajaraman-suppressed)]. For the other classifier architecture, suppression did not significantly (P>0.05) improve or worsen classifier performance. Conclusions: Rajaraman suppression significantly improved classification performance in one classification architecture, and did not significantly worsen classifier performance in the other classifier architecture. This research could be extended to explore the impact of bone suppression on classification of different lung pathologies, and the effect of other image enhancement techniques on classifier performance.

Multi-Organ Omics-Based Prediction for Adaptive Radiation Therapy Eligibility in Nasopharyngeal Carcinoma Patients Undergoing Concurrent Chemoradiotherapy.

PURPOSE: To investigate the role of different multi-organ omics-based prediction models for pre-treatment prediction of Adaptive Radiotherapy (ART) eligibility in patients with nasopharyngeal carcinoma (NPC). METHODS AND MATERIALS: Pre-treatment contrast-enhanced computed tomographic and magnetic resonance images, radiotherapy dose and contour data of 135 NPC patients treated at Hong Kong Queen Elizabeth Hospital were retrospectively analyzed for extraction of multi-omics features, namely Radiomics (R), Morphology (M), Dosiomics (D), and Contouromics (C), from a total of eight organ structures. During model development, patient cohort was divided into a training set and a hold-out test set in a ratio of 7 to 3 via 20 iterations. Four single-omics models (R, M, D, C) and four multi-omics models (RD, RC, RM, RMDC) were developed on the training data using Ridge and Multi-Kernel Learning (MKL) algorithm, respectively, under 10-fold cross validation, and evaluated on hold-out test data using average area under the receiver-operator-characteristics curve (AUC). The best-performing single-omics model was first determined by comparing the AUC distribution across the 20 iterations among the four single-omics models using two-sided student t-test, which was then retrained using MKL algorithm for a fair comparison with the four multi-omics models. RESULTS: The R model significantly outperformed all other three single-omics models (all p-value<0.0001), achieving an average AUC of 0.942 (95%CI: 0.938-0.946) and 0.918 (95%CI: 0.903-0.933) in training and hold-out test set, respectively. When trained with MKL, the R model (R_MKL) yielded an increased AUC of 0.984 (95%CI: 0.981-0.988) and 0.927 (95%CI: 0.905-0.948) in training and hold-out test set respectively, while demonstrating no significant difference as compared to all studied multi-omics models in the hold-out test sets. Intriguingly, Radiomic features accounted for the majority of the final selected features, ranging from 64% to 94%, in all the studied multi-omics models. CONCLUSIONS: Among all the studied models, the Radiomic model was found to play a dominant role for ART eligibility in NPC patients, and Radiomic features accounted for the largest proportion of features in all the multi-omics models.

Development and initial validation of the Therapeutic Misunderstanding Scale for use with clinical trials research participants.

Therapeutic misconception is evident when clinical trials participants conflate research and treatment, erroneously believing that every aspect of the research is intended to be for their direct benefit. We developed the 20-item Therapeutic Misunderstanding Scale (TMU) based on responses from 464 community-dwelling adults 50+ years of age (Study 1). A three-factor solution based on Horng and Grady's (2003) three-facets definition was identified using both exploratory and confirmatory factor analyses (EFA and CFA; these analyses were performed on separate samples). CFA results point to a second-order solution where each of Horng and Grady's three facets contribute significantly to the measurement of a higher-order therapeutic misunderstanding latent construct. Internal consistency of TMU responses (full scale) as well as the therapeutic misconception, misestimation, and optimism subscales were calculated as α = 0.88, α = 0.83, α = 0.79, and α = 0.75, respectively. These results were subsequently supported with responses from former clinical trials participants (Study 2). This TMU provides applied researchers a brief measure for use in future studies as well as a screening instrument for clinicians to more fully assess informed consent for participation in clinical trials research.

Personality trait levels within older couples and between-spouse trait differences as predictors of marital satisfaction.

In this study of 125 older couples married for an average of 34 years, multilevel models were computed to simultaneously examine intra-couple personality trait averages and between-spouse trait similarity as predictors of marital satisfaction. Our findings suggest that higher intra-couple levels of extraversion predict marital satisfaction, both husbands and wives. In addition, between-spouse similarity in openness to experience appears associated with higher levels of marital satisfaction as reported by husbands; concomitantly, between-spouse similarity in agreeableness predicts wives' marital satisfaction. With respect to openness (husbands) and agreeableness (wives), it did not matter which spouse within couples reported higher or lower trait levels. The most notable finding to emerge from this study is that neuroticism is not associated with marital satisfaction, neither husbands nor wives. This result stands in contrast to previously reported findings--the vast majority of prior research conducted with dating and newlywed couples. Conflicting results may reflect the degree to which neuroticism determines divorce within the first years of married life, adaptation to the foibles of one's spouse over time, overreliance on younger samples in marriage and family research, or some combination of these alternate explanations.

Mental alternation test: administration mode, age, and practice effects.

Administration mode, age, education, and practice effects were examined for the Mental Alternation Test (MAT), a brief orally administered measure of executive function. Participants (N = 135) between the ages of 65 and 85 years completed the MAT twice in person, twice over the telephone, or once in person and once over the telephone. MAT scores did not differ across administration modes. Furthermore, the MAT detected normative decreases in executive ability in later life. The correlation between MAT performance and educational attainment was small. Finally, practice effects, which were influenced by age, were found between administration times. These findings provide important implications for both research and clinical applications of the MAT in older populations.

Telephone administration of the Mental Alternation Test: sensitivity to cognitive decline and practice effects across midlife and late life.

BACKGROUND: We evaluated the utility of the telephone-administered Mental Alternation Test (MAT, an oral variant of the Trail-Making Test) for remote assessment of cognitive functioning in older adults. We examined (1) the sensitivity of MAT scores to cognitive change across 4 age groups, (2) practice effects associated with repeat administration, and (3) the uniformity of practice effects across age groups. METHODS: Community-dwelling volunteers were recruited randomly and categorized as young-middle-aged (45-54 years; n = 51), middle-aged (55-64 years; n = 58), young-old (65-74 years; n = 43) or old-old (75-85 years; n = 43). The participants completed the MAT twice within 2 weeks. The data were analyzed using mixed ANOVA. RESULTS: We found an effect of age on MAT performance [F(3, 191) = 11.37; p < 0.001], with planned comparisons revealing significantly lower scores in the old-old (p < 0.05). The scores on the second MAT administration were significantly higher than on the first administration [F(1, 191) = 12.82; p < 0.001], but this practice effect did not differ across age groups. CONCLUSIONS: The MAT was sensitive to cognitive decline in older adulthood. Practice effects were measurable but uniform across the observed age cohorts. As a brief telephone-administered test, the MAT represents a promising measure of cognitive functioning in older adults that is feasible for use in large-scale epidemiological studies.

Partial measurement equivalence of French and English versions of the Canadian Study of Health and Aging neuropsychological battery.

Neuropsychological batteries are often translated for use across populations differing in preferred language. Yet, equivalence in construct measurement across groups cannot be assumed. To address this issue, we examined data from the Canadian Study of Health and Aging, a large study of older adults. We tested the hypothesis that the latent variables underlying the neuropsychological battery administered in French or English were the same (invariant). The best-fitting baseline model, established in the English-speaking Exploratory sample (n = 716), replicated well in the English-speaking Validation sample (n = 715), and the French-speaking sample (FS, n = 446). Across the English- and FSs, two of the factors, Long-term Retrieval and Visuospatial speed, displayed invariance, that is, reflected the same constructs measured in the same scales. In contrast, the Verbal Ability factor showed only partial invariance, reflecting differences in the relative difficulty of some tests of language functions. This empirical demonstration of partial measurement invariance lends support to the continued use of these translated measures in clinical and research contexts and illustrates a framework for detailed evaluation of the generality of models of cognition and psychopathology, across groups of any sort.