-
Frontiers in Artificial Intelligence 2023In recent years, with the rapid development of deep learning technology, great progress has been made in computer vision, image recognition, pattern recognition, and... (Review)
Review
In recent years, with the rapid development of deep learning technology, great progress has been made in computer vision, image recognition, pattern recognition, and speech signal processing. However, due to the black-box nature of deep neural networks (DNNs), one cannot explain the parameters in the deep network and why it can perfectly perform the assigned tasks. The interpretability of neural networks has now become a research hotspot in the field of deep learning. It covers a wide range of topics in speech and text signal processing, image processing, differential equation solving, and other fields. There are subtle differences in the definition of interpretability in different fields. This paper divides interpretable neural network (INN) methods into the following two directions: model decomposition neural networks, and semantic INNs. The former mainly constructs an INN by converting the analytical model of a conventional method into different layers of neural networks and combining the interpretability of the conventional model-based method with the powerful learning capability of the neural network. This type of INNs is further classified into different subtypes depending on which type of models they are derived from, i.e., mathematical models, physical models, and other models. The second type is the interpretable network with visual semantic information for user understanding. Its basic idea is to use the visualization of the whole or partial network structure to assign semantic information to the network structure, which further includes convolutional layer output visualization, decision tree extraction, semantic graph, etc. This type of method mainly uses human visual logic to explain the structure of a black-box neural network. So it is a post-network-design method that tries to assign interpretability to a black-box network structure afterward, as opposed to the pre-network-design method of model-based INNs, which designs interpretable network structure beforehand. This paper reviews recent progress in these areas as well as various application scenarios of INNs and discusses existing problems and future development directions.
PubMed: 37899962
DOI: 10.3389/frai.2023.974295 -
Journal of the International... Oct 2023To compare longitudinal verbal fluency performance among Latinx Spanish speakers who develop Alzheimer's disease to those who do not develop dementia in absolute number...
OBJECTIVE
To compare longitudinal verbal fluency performance among Latinx Spanish speakers who develop Alzheimer's disease to those who do not develop dementia in absolute number of words produced on each task and their ratio to combine both scores.
METHOD
Participants included 833 Latinx Spanish-speaking older adults from a community-based prospective cohort in Manhattan. We performed growth curve modeling to investigate the trajectories of letter and semantic fluency, and their ratio (i.e., 'semantic index'), between individuals who developed Alzheimer's disease and those who did not (i.e., controls). The semantic index quantifies the proportion of words generated for semantic fluency in relation to the total verbal fluency performance.
RESULTS
Letter fluency performance did not decline in controls; we observed a linear decline in those who developed Alzheimer's disease. Semantic fluency declined in both groups and showed an increased rate of change over time in the incident Alzheimer's disease group; in comparison, the control group had a linear and slower decline. There were no group differences in the longitudinal trajectory (intercept and slope) of the semantic index.
CONCLUSION
A decline in letter fluency and a more rapid and accelerating decline over time in semantic fluency distinguished people who developed Alzheimer's disease from controls. Using the semantic index was not a superior marker of incident Alzheimer's disease compared to examining the two fluency scores individually. Results suggest the differential decline in verbal fluency tasks, when evaluated appropriately, may be useful for early identification of Alzheimer's disease in Latinx Spanish speakers, a historically understudied population.
Topics: Aged; Humans; Alzheimer Disease; Hispanic or Latino; Neuropsychological Tests; Prospective Studies; Semantics; Verbal Behavior; Speech Disorders
PubMed: 36637058
DOI: 10.1017/S1355617722000856 -
Journal of Dairy Science Oct 2023School lunch programs are mandated by the US Code of Federal Regulations to serve pasteurized milk that is skim or 1% fat and fortified with vitamins A and D. In recent...
School lunch programs are mandated by the US Code of Federal Regulations to serve pasteurized milk that is skim or 1% fat and fortified with vitamins A and D. In recent years, proposals have been made to alter nutritional requirements for school lunches and school lunch milk, including changes to the milk fat and flavor options available. The objective of this study was to evaluate parental understanding and perception of school lunch milk to better understand how changes to school lunch milk are perceived by parents. Four focus groups (n = 34) were conducted with parents of school-aged children (5-13 yr) who purchased milk as part of a lunch at school. Participants were asked about school lunch milk, including nutritional content, packaging, and flavoring. Focus groups included a build-your-own milk activity and discussion of children's milk products currently available on the market. Two subsequent online surveys were conducted with parents of school-aged children (survey 1, n = 216; survey 2, n = 133). Maximum difference scaling was used to evaluate what beverages parents would prefer their child to drink at school (survey 1) and which attributes of chocolate milk for children were most important to parents (survey 2). An adaptive choice-based conjoint activity (survey 1) included flavor, milk fat, heat treatment, label claims, and packaging type. Both surveys included questions to evaluate knowledge of milk nutrition and attitudes regarding milk and flavored milk. Agree/disagree questions were used in both surveys to assess parental opinions of school lunch milk. Survey 2 also included semantic differential (sliding scale) questions to assess parental opinions of chocolate milk and their acceptance of sugar alternatives in chocolate milk served in schools. Parents were familiar with the flavor options and packaging of school lunch milk, but expressed limited familiarity with school lunch milk fat content. Parents perceived milk to be healthy and a good source of vitamin D and calcium. From survey results, parents placed the highest importance on school lunch milk packaging, followed by milk fat percentage and flavoring over label claims and heat treatment. The ideal school lunch milk for parents was unflavored (white milk) or chocolate, 2% fat, and packaged in a cardboard gabletop carton. For school lunch chocolate milk, 3 distinct clusters of parents with differing opinions for children's chocolate milk were identified. Parents are largely unfamiliar with the specific attributes and nutritional profile of milk served in schools but believe that schools should offer milk to their children as part of breakfast and lunch. Parents in both surveys also displayed a preference for 2% fat milk over low-fat options, which provides actionable insight for both governmental bodies determining educational and nutritional policies for school meals and fluid milk producers seeking to optimize their products intended for schools.
Topics: Humans; Child; Animals; Milk; Lunch; Beverages; Parents; Perception
PubMed: 37210362
DOI: 10.3168/jds.2023-23326 -
Frontiers in Psychology 2023Alterations of verbalized thought occur frequently in psychotic disorders. We characterize linguistic findings in individuals with schizophrenia based on the current... (Review)
Review
INTRODUCTION
Alterations of verbalized thought occur frequently in psychotic disorders. We characterize linguistic findings in individuals with schizophrenia based on the current literature, including findings relevant for differential and early diagnosis.
METHODS
Review of literature published via PubMed search between January 2010 and May 2022.
RESULTS
A total of 143 articles were included. In persons with schizophrenia, language-related alterations can occur at all linguistic levels. Differentiating from findings in persons with affective disorders, typical symptoms in those with schizophrenia mainly include so-called "poverty of speech," reduced word and sentence production, impaired processing of complex syntax, pragmatic language deficits as well as reduced semantic verbal fluency. At the at-risk state, "poverty of content," pragmatic difficulties and reduced verbal fluency could be of predictive value.
DISCUSSION
The current results support multilevel alterations of the language system in persons with schizophrenia. Creative expressions of psychotic experiences are frequently found but are not in the focus of this review. Clinical examinations of linguistic alterations can support differential diagnostics and early detection. Computational methods (Natural Language Processing) may improve the precision of corresponding diagnostics. The relations between language-related and other symptoms can improve diagnostics.
PubMed: 38078276
DOI: 10.3389/fpsyg.2023.1287706 -
European Journal of Ageing Aug 2023The cognitive complaints encountered in late-life depression (LLD) make it difficult to distinguish from amnestic mild cognitive impairment (aMCI) and Alzheimer's... (Review)
Review
The role of semantic assessment in the differential diagnosis between late-life depression and Alzheimer's disease or amnestic mild cognitive impairment: systematic review and meta-analysis.
OBJECT
The cognitive complaints encountered in late-life depression (LLD) make it difficult to distinguish from amnestic mild cognitive impairment (aMCI) and Alzheimer's disease (AD) based on an analysis of neurocognitive disorders. The hypothesis of the early impairment of semantic memory in AD and aMCI is considered a potential differential cognitive clue, but the absence of this impairment has not yet been confirmed in LLD.
METHOD
Based on the PRISMA method, we systematically seek neuropsychological assessments of individuals with LLD, the present study included 31 studies representing 3291 controls and 2820 people with LLD. Wherever possible, studies that tested simultaneously groups with LLD, AD (or aMCI) were also included. The results of the group of neuropsychological tasks relying on semantic memory were analyzed in two groups of tasks with high- or low-executive demand. The mean average effect of LLD was calculated and compared to the incremental effect of aMCI or AD on the scores. Linear regressions including education, age, and severity and type of depression were run to seek their power of prediction for the mean average effects.
RESULTS
LLD has a medium effect on scores at semantic and phonemic fluency and naming and a small average effect on the low-executive demand tasks. Differences in education is a predictor of the effect of LLD on phonemic fluency and naming but not on semantic fluency or on low-executive demand tasks. Except for semantic fluency, aMCI did not demonstrate an incremental effect on the scores compared to LLD, while AD did, for all the tasks except phonemic fluency.
CONCLUSION
Assessment of semantic memory can be a discriminating clue for the distinction between depression and Alzheimer's disease but some methodological variables are highly influential to the scores, especially education. However, high-executive semantic tasks alone do not allow us to clearly distinguish LLD from AD or aMCI, as both pathologies seem to have a largely dialectical influential relationship, but low-executive semantic tasks appear as more sensible to this pathological distinction.
PubMed: 37563432
DOI: 10.1007/s10433-023-00780-z -
Assessment Dec 2023The ScreeLing is a screening instrument developed to assess post-stroke aphasia, via the linguistic levels Syntax, Phonology, and Semantics. It could also be a useful...
The ScreeLing is a screening instrument developed to assess post-stroke aphasia, via the linguistic levels Syntax, Phonology, and Semantics. It could also be a useful test for the clinical subtypes of frontotemporal dementia (FTD) and Alzheimer's dementia (AD), as specific and often selective disorders are expected. Its ability to differentiate between the clinical subtypes of FTD and AD is, however, still unknown. We investigated differences in ScreeLing total and subscores, linguistic-level disorders' relationship with disease severity, and classification abilities, in patients with behavioral variant FTD (bvFTD; = 46), patients with primary progressive aphasia (PPA; = 105) (semantic variant primary progressive aphasia [svPPA], non-fluent variant primary progressive aphasia [nfvPPA], and logopenic variant primary progressive aphasia [lvPPA], AD [ = 20] and controls [ = 35]). We examined group differences in ScreeLing total and subscores, and one-, two- or three-level linguistic disorders using one-way analyses of covariance (ANCOVAs) or Quade's rank ANCOVA. We used frequency analyses to obtain the occurrence of the linguistic-level disorders. We determined sensitivity and specificity by the area under the curve by receiver-operating characteristics analyses to investigate classification abilities. The total score was lower in patients (bvFTD: 63.8 ± 8.5, svPPA: 58.8 ± 11.3, nfvPPA: 63.5 ± 8.4, lvPPA: 61.7 ± 6.6, AD: 63.8 ± 5.5) than controls (71.3 ± 1.0) ( < .001). Syntax subscores were lower in svPPA (19.4 ± 4.6; < .001) and lvPPA (20.3 ± 3.2; = .002) than controls (23.8 ± 0.4). Phonology subscores were lower in lvPPA (19.8 ± 2.6) than bvFTD (21.7 ± 2.8) ( = .010). Semantics subscores were lowest in svPPA (17.8 ± 5.0; < .002). A selective phonological disorder was most prevalent in lvPPA (34.9%). The higher the disease severity, the more linguistic-level disorders. The optimal cutoff for the total score was 70, and 23 for all three subscores. Good classification abilities were found for the Semantics (svPPA vs. bvFTD), Phonology (lvPPA vs. svPPA), and Syntax (nfvPPA vs. lvPPA) subscores. This easy to administer test gives information about language processing with the potential to improve differential diagnosis in memory clinics and in the future potentially also clinical trial planning.
Topics: Humans; Alzheimer Disease; Semantics; Frontotemporal Dementia; Linguistics; Aphasia, Primary Progressive
PubMed: 36799220
DOI: 10.1177/10731911231154512 -
Frontiers in Human Neuroscience 2023Prior work has shown positive effects of High Definition transcranial direct current stimulation (HD-tDCS) over the dorsolateral prefrontal cortex (DLPFC) on semantic...
BACKGROUND
Prior work has shown positive effects of High Definition transcranial direct current stimulation (HD-tDCS) over the dorsolateral prefrontal cortex (DLPFC) on semantic memory performance and metamemory monitoring accuracy. However, HD-tDCS requires setup by a trained researcher, which is not always feasible. Few studies have used remotely supervised (rs) tDCS in healthy populations, and remote supervision has strong practical benefits.
OBJECTIVE/HYPOTHESIS
The goal of the current study was to test if previously shown effects of HD-tDCS over the left DLPFC on semantic memory performance and metamemory monitoring accuracy extended to conventional rs-tDCS, which is less focal than HD-tDCS, and to episodic memory and metamemory tasks.
MATERIALS AND METHODS
A total of 36 healthy participants completed 6 weeks of rs-tDCS sessions, with either active left or right anodal DLPFC stimulation, or sham. Participants completed semantic and episodic memory and metamemory tasks, which each lasted for three consecutive sessions, and session order was counterbalanced across participants.
RESULTS
Overall, there were no main effects of rs-tDCS on metamemory monitoring accuracy or memory performance for either the semantic or the episodic tasks. However, there were effects of rs-tDCS that depended on the order of completing the episodic and semantic task sessions. When participants completed the semantic task sessions after the episodic task sessions, semantic recognition was greater in the left anodal DLPFC condition. In a parallel effect, when participants completed the episodic task sessions after the semantic task sessions, episodic recognition was greater in the right anodal DLPFC condition.
CONCLUSION
Prior experience with tDCS is a factor for effects of rs-tDCS on cognition. Additionally, the current experiment provides evidence for the feasibility of fully remotely supervised tDCS in healthy participants.
PubMed: 37635805
DOI: 10.3389/fnhum.2023.1239126 -
NeuroImage Aug 2023Human neuroimaging studies have shown that the contents of episodic memories are represented in distributed patterns of neural activity. However, these studies have...
Human neuroimaging studies have shown that the contents of episodic memories are represented in distributed patterns of neural activity. However, these studies have mostly been limited to decoding simple, unidimensional properties of stimuli. Semantic encoding models, in contrast, offer a means for characterizing the rich, multidimensional information that comprises episodic memories. Here, we extensively sampled four human fMRI subjects to build semantic encoding models and then applied these models to reconstruct content from natural scene images as they were viewed and recalled from memory. First, we found that multidimensional semantic information was successfully reconstructed from activity patterns across visual and lateral parietal cortices, both when viewing scenes and when recalling them from memory. Second, whereas visual cortical reconstructions were much more accurate when images were viewed versus recalled from memory, lateral parietal reconstructions were comparably accurate across visual perception and memory. Third, by applying natural language processing methods to verbal recall data, we showed that fMRI-based reconstructions reliably matched subjects' verbal descriptions of their memories. In fact, reconstructions from ventral temporal cortex more closely matched subjects' own verbal recall than other subjects' verbal recall of the same images. Fourth, encoding models reliably transferred across subjects: memories were successfully reconstructed using encoding models trained on data from entirely independent subjects. Together, these findings provide evidence for successful reconstructions of multidimensional and idiosyncratic memory representations and highlight the differential sensitivity of visual cortical and lateral parietal regions to information derived from the external visual environment versus internally-generated memories.
Topics: Humans; Memory, Episodic; Brain Mapping; Mental Recall; Visual Perception; Parietal Lobe; Magnetic Resonance Imaging
PubMed: 37327954
DOI: 10.1016/j.neuroimage.2023.120222 -
Human Brain Mapping Feb 2024Wine tasting is a very complex process that integrates a combination of sensation, language, and memory. Taste and smell provide perceptual information that, together...
Wine tasting is a very complex process that integrates a combination of sensation, language, and memory. Taste and smell provide perceptual information that, together with the semantic narrative that converts flavor into words, seem to be processed differently between sommeliers and naïve wine consumers. We investigate whether sommeliers' wine experience shapes only chemosensory processing, as has been previously demonstrated, or if it also modulates the way in which the taste and olfactory circuits interact with the semantic network. Combining diffusion-weighted images and fMRI (activation and connectivity) we investigated whether brain response to tasting wine differs between sommeliers and nonexperts (1) in the sensory neural circuits representing flavor and/or (2) in the neural circuits for language and memory. We demonstrate that training in wine tasting shapes the microstructure of the left and right superior longitudinal fasciculus. Using mediation analysis, we showed that the experience modulates the relationship between fractional anisotropy and behavior: the higher the fractional anisotropy the higher the capacity to recognize wine complexity. In addition, we found functional differences between sommeliers and naïve consumers affecting the flavor sensory circuit, but also regions involved in semantic operations. The former reflects a capacity for differential sensory processing, while the latter reflects sommeliers' ability to attend to relevant sensory inputs and translate them into complex verbal descriptions. The enhanced synchronization between these apparently independent circuits suggests that sommeliers integrated these descriptions with previous semantic knowledge to optimize their capacity to distinguish between subtle differences in the qualitative character of the wine.
Topics: Humans; Semantics; Semantic Web; Smell; Taste Perception; Sensation; Taste
PubMed: 38339911
DOI: 10.1002/hbm.26564 -
Behavioral Sciences (Basel, Switzerland) Oct 2023Relationship satisfaction is at the core of a robust social life and is essential to mental health. The positive and negative semantic dimensions of the relationship...
Relationship satisfaction is at the core of a robust social life and is essential to mental health. The positive and negative semantic dimensions of the relationship satisfaction (PN-SMD) scale is considered in the field of relationship studies to be a reliable tool for assessing the quality of a person's interpersonal relationships. This study evaluated the psychometric properties of the PN-SMD scale by conducting multidimensional item response theory (MIRT) and differential item functioning (DIF) analyses, both of which are emerging assessment methods that focus on individual items. We recruited 511 Chinese undergraduate students for this study. Construct validity, internal consistency, and concurrent validity were assessed, and MIRT and DIF analyses were conducted. Five of the 14 items were found to have gender-based DIF traits, affecting the scale's construct validity. A revised nine-item scale (DIF items excluded) had a significantly better model fit and demonstrated comparable concurrent validity to the original scale. The implications of our results and future research directions are discussed.
PubMed: 37887475
DOI: 10.3390/bs13100825