The potential of language tasks for identifying senile dementia

Automatic speech analysis for detecting cognitive decline of older adults

Background

Speech analysis has been expected to help as a screening tool for early detection of Alzheimer's disease (AD) and mild-cognitively impairment (MCI). Acoustic features and linguistic features are usually used in speech analysis. However, no studies have yet determined which type of features provides better screening effectiveness, especially in the large aging population of China.

Objective

Firstly, to compare the screening effectiveness of acoustic features, linguistic features, and their combination using the same dataset. Secondly, to develop Chinese automated diagnosis model using self-collected natural discourse data obtained from native Chinese speakers.

Methods

A total of 92 participants from communities in Shanghai, completed MoCA-B and a picture description task based on the Cookie Theft under the guidance of trained operators, and were divided into three groups including AD, MCI, and heathy control (HC) based on their MoCA-B score. Acoustic features (Pitches, Jitter, Shimmer, MFCCs, Formants) and linguistic features (part-of-speech, type-token ratio, information words, information units) are extracted. The machine algorithms used in this study included logistic regression, random forest (RF), support vector machines (SVM), Gaussian Naive Bayesian (GNB), and k-Nearest neighbor (kNN). The validation accuracies of the same ML model using acoustic features, linguistic features, and their combination were compared.

Results

The accuracy with linguistic features is generally higher than acoustic features in training. The highest accuracy to differentiate HC and AD is 80.77% achieved by SVM, based on all the features extracted from the speech data, while the highest accuracy to differentiate HC and AD or MCI is 80.43% achieved by RF, based only on linguistic features.

Conclusion

Our results suggest the utility and validity of linguistic features in the automated diagnosis of cognitive impairment, and validated the applicability of automated diagnosis for Chinese language data.

Alzheimer’s Dementia Recognition From Spontaneous Speech Using Disfluency and Interactional Features

Alzheimer’s disease (AD) is a progressive, neurodegenerative disorder mainly characterized by memory loss with deficits in other cognitive domains, including language, visuospatial abilities, and changes in behavior. Detecting diagnostic biomarkers that are noninvasive and cost-effective is of great value not only for clinical assessments and diagnostics but also for research purposes. Several previous studies have investigated AD diagnosis via the acoustic, lexical, syntactic, and semantic aspects of speech and language. Other studies include approaches from conversation analysis that look at more interactional aspects, showing that disfluencies such as fillers and repairs, and purely nonverbal features such as inter-speaker silence, can be key features of AD conversations. These kinds of features, if useful for diagnosis, may have many advantages: They are simple to extract and relatively language-, topic-, and task-independent. This study aims to quantify the role and contribution of these features of interaction structure in predicting whether a dialogue participant has AD. We used a subset of the Carolinas Conversation Collection dataset of patients with AD at moderate stage within the age range 60–89 and similar-aged non-AD patients with other health conditions. Our feature analysis comprised two sets: disfluency features, including indicators such as self-repairs and fillers, and interactional features, including overlaps, turn-taking behavior, and distributions of different types of silence both within patient speech and between patient and interviewer speech. Statistical analysis showed significant differences between AD and non-AD groups for several disfluency features (edit terms, verbatim repeats, and substitutions) and interactional features (lapses, gaps, attributable silences, turn switches per minute, standardized phonation time, and turn length). For the classification of AD patient conversations vs. non-AD patient conversations, we achieved 83% accuracy with disfluency features, 83% accuracy with interactional features, and an overall accuracy of 90% when combining both feature sets using support vector machine classifiers. The discriminative power of these features, perhaps combined with more conventional linguistic features, therefore shows potential for integration into noninvasive clinical assessments for AD at advanced stages.

Syntactic Complexity as a Linguistic Marker to Differentiate Mild Cognitive Impairment From Normal Aging

Purpose In this study, we sought to identify critical linguistic markers that can differentiate sentence processing of individuals with mild cognitive impairment (MCI) from the sentence processing of normal-aging populations by manipulating sentences' linguistic complexity. We investigated whether passive sentences, as linguistically complex structures, can serve as linguistic markers that can contribute to diagnoses that distinguish MCI from normal aging. Method In total, 52 participants, including 26 adults with amnestic MCI and 26 cognitively unimpaired adults, participated in the study. All participants were native speakers of Korean. We administered the two subsets of active and passive conditions using a sentence-picture paradigm with semantically reversible sentences to both groups. Results A mixed-effects model using PROC NLMIXED demonstrated that the MCI group exhibited differentially greater difficulty in processing passive than active sentences compared to the normal-aging group. A logistic regression fitted with the PROC LOGISTIC model identified the sum of the passive sentences, with age and education effects as the best models to distinguish individuals with MCI from the normal-aging group. Conclusion Sentence comprehension deficits emerged in the MCI stage when the syntactic complexity was increased. Furthermore, a passive structure was the best predictor for efficiently distinguishing the MCI group from the normal-aging group. These results are clinically and theoretically important, given that linguistic complexity can serve as a critical behavioral marker in the detection of early symptoms associated with linguistic-cognitive decline.

Sentence comprehension in patients with dementia of the Alzheimer's type

Sentence comprehension is diminished in patients with dementia of the Alzheimer’s type (DAT). However, the underlying reason for such deficits is still not entirely clear. The Syntactic Deficit Hypothesis attributes sentence comprehension deficits in DAT patients to the impairment in syntactic ability, whereas the Processing Resource Deficit Hypothesis proposes that sentence comprehension deficits are the result of working memory deficiency. This study investigated the deficits in sentence comprehension in Chinese-speaking DAT patients with different degrees of severity using sentence-picture matching tasks. The study revealed a significant effect of syntactic complexity in patients and healthy controls, but the effect was stronger in patients than in healthy controls. When working memory demand was minimized, the effect of syntactic complexity was only significant in patients with moderate DAT, but not in healthy controls or those with mild DAT. The findings suggest that in patients with mild DAT, working memory decline was the major source of sentence comprehension difficulty and in patients with moderate DAT, working memory decline and syntactic impairment jointly contributed to the impairments in sentence comprehension. The source of sentence comprehension deficits varied with degree of dementia severity.

An Interactional Profile to Assist the Differential Diagnosis of Neurodegenerative and Functional Memory Disorders

OBJECTIVE

Specialist services for dementia are seeing an increasing number of patients. We investigated whether interactional and linguistic features in the communication behavior of patients with memory problems could help distinguish between those with problems secondary to neurological disorders (ND) and those with functional memory disorder (FMD).

METHODS

In part 1 of this study, a diagnostic scoring aid (DSA) was developed encouraging linguists to provide quantitative ratings for 14 interactional features. An optimal cut-off differentiating ND and FMD was established by applying the DSA to 30 initial patient-doctor memory clinic encounters. In part 2, the DSA was tested prospectively in 10 additional cases analyzed independently by 2 conversation analysts blinded to medical information.

RESULTS

In part 1, the median score of the DSA was +5 in ND and -5 in FMD (P<0.001). The optimal numeric DSA cut-off (+1) identified patients with ND with a sensitivity of 86.7% and a specificity of 100%. In part 2, DSA scores of rater 1 correctly predicted 10/10 and those of rater 2 predicted 9/10 diagnoses.

CONCLUSIONS

This study indicates that interactional and linguistic features can help distinguish between patients developing dementia and those with FMD and could aid the stratification of patients with memory problems.

Development of a word accentuation test for predicting cognitive performance in Portuguese-speaking populations

The Word Accentuation Test (WAT) has been used to predict premorbid intelligence and cognitive performance in Spanish-speaking populations. It requires participants to read a list of words without the accent marks that indicate the stressed syllable. Label="OBJECTIVE">As Portuguese pronunciation is also strongly based on accent marks, our aim was to develop a Brazilian version of the WAT. METHODS An initial pool of 60 items was constructed and a final version of 40 items (named WAT-Br) was derived by item response theory. A sample of 206 older adults underwent the WAT-Br and a standardized neuropsychological battery. Independent ratings were performed by two observers in 58 random participants. RESULTS The items showed moderate to high discrimination (α between 0.93 and 25.04) and spanned a wide range of difficulty (β between -2.07 and 1.40). The WAT-Br was shown to have an excellent internal consistency (Kuder-Richardson Formula 20 = 0.95) and inter-rater reliability (intraclass correlation coefficient = 0.92). It accounted for 61% of the variance in global cognitive performance. CONCLUSION A version of the WAT for Portuguese-speaking populations was developed and proved to be a valuable tool for estimating cognitive performance.

Using the Rasch analysis for the psychometric validation of the Irregular Word Reading Test (TeLPI): A Portuguese test for the assessment of premorbid intelligence

OBJECTIVE

The present study aims to analyze the psychometric characteristics of the TeLPI (Irregular Words Reading Test), a Portuguese premorbid intelligence test, using the Rasch model for dichotomous items.

RESULTS

The results reveal an overall adequacy and a good fit of values regarding both items and persons. A high variability of cognitive performance level and a good quality of the measurements were also found. The TeLPI has proved to be a unidimensional measure with reduced DIF effects.

CONCLUSIONS

The present findings contribute to overcome an important gap in the psychometric validity of this instrument and provide good evidence of the overall psychometric validity of TeLPI results.

The Breakfast Club: Results of a study examining the effectiveness of a multi-modality group communication treatment

Twenty mid-stage Alzheimer's patients at the Jewish Home and Hospital in the Bronx participatedfor 12 weeks each in four groups offive in a five-day-a-week program of structured multi-modality group communication intervention called “The Breakfast Club.” Twenty matched patients participated in a standard conversation group and served as controls. The Breakfast Club attempted to incorporate all that was currently known about the residual communication strengths of Alzheimer's patients and about previous treatments shown to be effective with this population. Results showed that Breakfast Club participants improved significantly on measures of language performance, functional independence and use of social communication while control subjects did not. Breakfast Club members also showed significant increases in “interest and involvement” and the use of procedural memories over the 12-week period.

Speaking in Alzheimer's Disease, is That an Early Sign? Importance of Changes in Language Abilities in Alzheimer's Disease

It is known that Alzheimer’s disease (AD) influences the temporal characteristics of spontaneous speech. These phonetical changes are present even in mild AD. Based on this, the question arises whether an examination based on language analysis could help the early diagnosis of AD and if so, which language and speech characteristics can identify AD in its early stage. The purpose of this article is to summarize the relation between prodromal and manifest AD and language functions and language domains. Based on our research, we are inclined to claim that AD can be more sensitively detected with the help of a linguistic analysis than with other cognitive examinations. The temporal characteristics of spontaneous speech, such as speech tempo, number of pauses in speech, and their length are sensitive detectors of the early stage of the disease, which enables an early simple linguistic screening for AD. However, knowledge about the unique features of the language problems associated with different dementia variants still has to be improved and refined.

Verbal fluency, naming and verbal comprehension: three aspects of language as predictors of cognitive impairment

OBJECTIVES

To establish the possible relationship among three components of language (verbal fluency, naming and comprehension) and cognitive impairment as well as to determine the usefulness of language assessment tests to predict or monitor the development of cognitive impairment.

METHOD

A comparative, descriptive and cross-sectional study was performed on 82 subjects ≥ 65 years of age who were cognitively assessed with the Mini Mental State Examination and were divided into two groups: Group A comprised of subjects classified as levels 1, 2 and 3 on the Reisberg's Global Deterioration Scale (GDS) and group B comprised of subjects at levels 4 and 5 of the GDS. Language skills were assessed by the Verbal Fluency Test, Boston Naming Test and Token Test.

RESULTS

An inverse relationship between performance on language tests and cognitive impairment level was observed with a more pronounced effect observed on fluency and comprehension tests.

CONCLUSION

Language assessments, especially fluency and comprehension, were good indicators of cognitive impairment. The use of these assessments as predictors of the degree of cognitive impairment is discussed in-depth.

Word sequence production in cognitive decline: forward ever, backward never

Backward recall of automatic word sequences involves declarative and working memory abilities known to be impaired in the early stages of cognitive decline. Yet its utility in the diagnosis of mild cognitive impairment and mild dementia has not been studied in detail. We analysed word sequence production in 234 participants drawn from three categories: subjective cognitive impairment, mild cognitive impairment, and mild dementia in Alzheimer's disease. The names of the months were used as a diagnostic target for investigating forward versus backward sequence production. Forward production remained normal across categories. In contrast, backward speed was significantly decreased in mild cognitive impairment. In dementia both speed and accuracy were impaired. Backward production had significant diagnostic classificatory power. We conclude that word sequence production yields data relevant to the diagnosis of dementia with a minimum of time and expense.

Dysgraphia in Alzheimer's disease: a review for clinical and research purposes

PURPOSE

This article presents a critical review of literature on dysgraphia associated with Alzheimer's disease (AD). Research presented includes discussions of central and peripheral spelling impairments as well as the impact of general, nonlinguistic cognitive functions on dysgraphia associated with AD.

METHOD

The studies critically reviewed were from a variety of disciplines, with emphasis on seminal work, recent literature, and the first author's research.

CONCLUSIONS

Studies have shown that writing impairment is heterogeneous within the AD population; however, there are certain aspects of the writing process that are more vulnerable than others and may serve as diagnostic signs. Identifying patterns of writing impairment at different stages of AD may help to chart disease progression and assist in the development of appropriate interventions.

Communication abilities of individuals with late-stage Alzheimer disease

The communication abilities of 49 individuals in the late stage of Alzheimer disease were examined in relation to other markers of late-stage Alzheimer disease (e.g., incontinence and ambulatory ability). Two existing tools used to stage severity of dementia, the Global Deterioration Scale and the Functional Assessment Stages, have represented communication abilities in individuals with late-stage Alzheimer disease to be minimal to nonexistent. The individuals in this sample showed greater communication skills and verbal output than would be predicted by these scales.

Estimation of premorbid intelligence in Spanish people with the Word Accentuation Test and its application to the diagnosis of dementia

The Word Accentuation Test assesses the accentuation of 30 infrequent Spanish words written without the accentuation mark and is an easy-to-use tool for estimating premorbid intelligence of Spanish-speaking people. Its intraobserver (0.97) and interobserver (0.93) reliabilities and its correlation with the Wechsler Adult Intelligence Scale (.837) and Raven's Progressive Matrices (.655) are high, offering a good prediction of general intelligence. It is resistant to mental deterioration; 20 demented and 40 controls matched by sex, age, and education obtained similar scores. The discrepancies between current and predicted scores in Raven's scale can diagnose mild-moderate dementia with 0.79 accuracy (sensitivity, 0.78; specificity, 0.82).

Ideational and semantic contributions to narrative production in adults with dementia of the Alzheimer's type

The narrative production of adults with probable dementia of Alzheimer's type (DAT) was investigated to determine the extent to which cognitive or linguistic deficits might explain the breakdown of discourse production. The structure of narrative tasks was manipulated so that the relationships among task structure and language production corresponded to predictions about the suspected origin of deficit. Sixteen DAT and 16 normal control subjects were administered four narrative tasks that were controlled for the amount of content and picture format display. The DAT subjects exhibited reduced content in terms of propositions and lexical items, shorter sentence lengths with more sentence fragments and reference errors. The mode of picture display failed to influence the amount of target content, or the grammatical performance of subjects. However, when compared to the normal controls, the amount of information pictorially represented significantly influenced the content provided by the DAT subjects; they performed better when the message to be related contained relatively less information. These findings support the contributions of both semantic-lexical and ideational systems to narrative discourse to narrative discourse production.

Development of an "early" detection battery for dementia of the Alzheimer type

1. To develop a diagnostic battery sensitive to and specific for the early detection of Alzheimer disease (AD) dementia, the authors reviewed over 400 journal articles dealing with the diagnosis of A.D. or senile dementia and cognitive assessment in organic brain dysfunction and closed head injury. 2. We culled those studies that met our criteria for solid, reliable and statistically significant results and recommend the testing paradigms that most often produced good discrimination of mild AD dementia from normal senescence. 3. These include tests of language, verbal and non-verbal memory, perception, praxis, attention and reasoning. 4. The battery we assembled takes less than 1 hour to administer, requires no special equipment, and was designed as an early screen for use by psychologists, psychiatrists and other trained health care professionals; it is not intended for repeated administration, as in pharmacological or longitudinal studies.

Language and dementia: a review

In the literature on dementia the presence or absence of language deficits has come to occupy a pivotal position with respect to certain nosological and nosographical issues. These are reviewed critically, together with the results of experimental investigations which have sought to elucidate the nature of the language deficit(s) in patients with dementia, especially that of the Alzheimer type. Attention is drawn to important gaps in current knowledge and suggestions are made regarding the direction of future research.

Anomia in cerebral diseases

Naming performance was studied in demented patients, age-matched controls, left-hemisphere stroke patients with aphasia and right-hemisphere stroke patients. The experimental naming test compared four levels of perceptual difficulty and two language variables: word frequency and word length. Naming accuracy and error types were compared among subject groups. Perceptual difficulty influenced naming in the demented and right-hemisphere stroke patients, but not in aphasics and controls. Visual errors, likewise, characterized the former two groups. Semantically-related errors and circumlocutions characterized the naming of aphasic and demented patients, while phonemic errors were common only in aphasics. The results suggested differing patterns of anomia in different patient groups.

Controlled and automatic information processing in senile dementia: a review

The evidence on cognitive decline in senile dementia is reviewed with respect to the distinction between controlled and automatic processing. It is argued that controlled processing, which requires the attentional resources of the individual, declines in early senile dementia. However, automatic processing, which does not require attentional resources, is well maintained until late in the disorder. Implications of this view are drawn for the development of effective screening instruments which can be used for the early detection of dementia in the general population.

Dementia of the Alzheimer type. An inventory of diagnostic clinical features

An Inventory presenting characteristics useful in the clinical identification of dementia of the Alzheimer type (DAT) is presented. The Inventory includes aphasia, amnesia, abnormal cognition and visuospatial skills, inappropriate lack of concern, and normal motor functions. In a retrospective study of 50 consecutive dementia patients, the DAT Inventory correctly identified 100% of DAT subjects and 94% of non-DAT cases. Utilization of positive criteria can aid in the identification of DAT and can facilitate recognition of treatable illnesses masquerading as DAT.

Test profile of cholinergic dysfunction and of Alzheimer-type dementia

A characteristic profile of subtest scores from the Wechsler Adult Intelligence Scale (WAIS) similar to that seen in clinically tested dementia patients was found in 10 of 19 normal young adult subjects with a drug-induced cholinergic deficiency of mental functioning but in only 4 of 22 control subjects. The same subtest profile was then found in test data from two groups of consecutive dementia patients (61 and 77 patients, respectively) with research diagnoses of Alzheimer-type dementia (DAT, senile and presenile), multi-infarct, and other dementias. The profile identified 44% of testable patients with AD and was 96% specific to DAT (only two false positives). A Verbal-Performance IQ discrepancy of 15 or more points was associated with cholinergic dysfunction in the normal drug subjects, but this IQ-score discrepancy did not differentiate AD from multi-infarct dementia patients. It was concluded that the subtest profile could contribute to the differentiation of DAT from other dementias. The association of this profile with drug-induced cholinergic deficiency suggested that the cholinergic deficiency of DAT might be responsible for the intellectual changes seen in this disease.

Relationships between language skills as assessed by the Halstead-Reitan battery and the Luria-Nebraska language-related factor scales in a nonaphasic patient population

The study involved a comparison between language tests derived from the Halstead-Reitan and Luria-Nebraska neuropsychological test batteries. Language-related measures from each battery were entered into a factor analysis based on a sample of 150 nonaphasic neuropsychiatric patients. Five factors were extracted and were named Academic Achievement, Basic Skills, Verbal Intelligence, Auditory Discrimination, and Comprehension of Syntactical Relations. Measures from both the Halstead-Reitan and Luria-Nebraska loaded on most of these factors. A canonical correlation analysis revealed a strong relationship between the Halstead-Reitan and Luria-Nebraska measures. However, it was noted that interpretation of this relationship must be qualified by the association found between the measures used and educational achievement.

The naming disorder of dementia

Naming impairment is a common feature of the language disorder of dementia, yet agreement has not been reached on its mechanisms. In this study, the confrontation naming performance of twelve demented subjects was compared to that of age-matched controls. Naming deficits were studied in relation to overall language and cognitive dysfunction and analysed to assess the importance of both perceptual and linguistic factors. Naming dysfunction occurred even in mild dementia, in patients whose overall language function remained normal, and worsened in proportion to the degree both of language deficit and overall cognitive dysfunction. Perceptual difficulty and word frequency, but not word length, were important determinants of naming performance in demented patients.