A novel speech analysis algorithm to detect cognitive impairment in a Spanish population.

Objective: Early detection of cognitive impairment in the elderly is crucial for diagnosis and appropriate care. Brief, cost-effective cognitive screening instruments are needed to help identify individuals who require further evaluation. This study presents preliminary data on a new screening technology using automated voice recording analysis software in a Spanish population. Method: Data were collected from 174 Spanish-speaking individuals clinically diagnosed as cognitively normal (CN, n = 87) or impaired (mild cognitive impairment [MCI], n = 63; all-cause dementia, n = 24). Participants were recorded performing four common language tasks (Animal fluency, alternating fluency [sports and fruits], phonemic "F" fluency, and Cookie Theft Description). Recordings were processed via text-transcription and digital-signal processing techniques to capture neuropsychological variables and audio characteristics. A training sample of 122 subjects with similar demographics across groups was used to develop an algorithm to detect cognitive impairment. Speech and task features were used to develop five independent machine learning (ML) models to compute scores between 0 and 1, and a final algorithm was constructed using repeated cross-validation. A socio-demographically balanced subset of 52 participants was used to test the algorithm. Analysis of covariance (ANCOVA), covarying for demographic characteristics, was used to predict logistically-transformed algorithm scores. Results: Mean logit algorithm scores were significantly different across groups in the testing sample (p < 0.01). Comparisons of CN with impaired (MCI + dementia) and MCI groups using the final algorithm resulted in an AUC of 0.93/0.90, with overall accuracy of 88.4%/87.5%, sensitivity of 87.5/83.3, and specificity of 89.2/89.2, respectively. Conclusion: Findings provide initial support for the utility of this automated speech analysis algorithm as a screening tool for cognitive impairment in Spanish speakers. Additional study is needed to validate this technology in larger and more diverse clinical populations.

Development and Validity of Norms for Cognitive Dispersion on the Uniform Data Set 3.0 Neuropsychological Battery.

OBJECTIVE: Cognitive dispersion indexes intraindividual variability in performance across a battery of neuropsychological tests. Measures of dispersion show promise as markers of cognitive dyscontrol and everyday functioning difficulties; however, they have limited practical applicability due to a lack of normative data. This study aimed to develop and evaluate normed scores for cognitive dispersion among older adults. METHOD: We analyzed data from 4,283 cognitively normal participants aged ≥50 years from the Uniform Data Set (UDS) 3.0. We describe methods for calculating intraindividual standard deviation (ISD) and coefficient of variation (CoV), as well as associated unadjusted scaled scores and demographically adjusted z-scores. We also examined the ability of ISD and CoV scores to differentiate between cognitively normal individuals (n = 4,283) and those with cognitive impairment due to Lewy body disease (n = 282). RESULTS: We generated normative tables to map raw ISD and CoV scores onto a normal distribution of scaled scores. Cognitive dispersion indices were associated with age, education, and race/ethnicity but not sex. Regression equations were used to develop a freely accessible Excel calculator for deriving demographically adjusted normed scores for ISD and CoV. All measures of dispersion demonstrated excellent diagnostic utility when evaluated by the area under the curve produced from receiver operating characteristic curves. CONCLUSIONS: Results of this study provide evidence for the clinical utility of sample-based and demographically adjusted normative standards for cognitive dispersion on the UDS 3.0. These standards can be used to guide interpretation of intraindividual variability among older adults in clinical and research settings.

A Critical Review of Neuropsychological Actuarial Criteria for Mild Cognitive Impairment.

BACKGROUND: The concept of mild cognitive impairment (MCI) has evolved since its original conception. So, too, have MCI diagnostic methods, all of which have varying degrees of success in identifying individuals at risk of conversion to dementia. The neuropsychological actuarial method is a straightforward diagnostic approach that has shown promise in large datasets in identifying individuals with MCI who are likely to have progressive courses. This method has been increasingly applied in various iterations and samples, raising questions of how best to apply this method and when caution should be used. OBJECTIVE: Our objective was to review the literature investigating use of the neuropsychological actuarial method to diagnose MCI to identify strengths and weaknesses of this approach, as well as highlight areas for further research. METHODS: Databases PubMed and PsychInfo were systematically searched for studies that compared the neuropsychological actuarial method to some other diagnostic method. RESULTS: We identified 13 articles and extracted relevant study characteristics and findings. Existing literature was reviewed and integrated, with focus on the neuropsychological actuarial method's performance relative to existing diagnostic methods/criteria as well as associations with longitudinal outcomes and biomarkers. Tables with pertinent methodological information and general findings are also provided. CONCLUSION: The neuropsychological actuarial method to diagnose MCI has shown utility some in large-scale homogenous databases compared to research criteria. However, its standing relative to consensus diagnostic methods is unclear, and emerging evidence suggests the neuropsychological actuarial method may be more prone to diagnostic errors in more demographically diverse populations.

The impact of conventional versus robust norming on cognitive characterization and clinical classification of MCI and dementia.

We examined the impact of conventional versus robust normative approaches on cognitive characterization and clinical classification of MCI versus dementia. The sample included participants from the National Alzheimer's Coordinating Center Uniform Data Set. Separate demographically adjusted z-scores for cognitive tests were derived from conventional (n = 4273) and robust (n = 602) normative groups. To assess the impact of deriving scores from a conventional versus robust normative group on cognitive characterization, we examined likelihood of having a low score on each neuropsychological test. Next, we created receiver operating characteristic (ROC) curves for the ability of normed scores derived from each normative group to differentiate between MCI (n = 3570) and dementia (n = 1564). We examined the impact of choice of normative group on classification accuracy by comparing sensitivity and specificity values and areas under the curves (AUC). Compared with using a conventional normative group, using a robust normative group resulted in a higher likelihood of low cognitive scores for individuals classified with MCI and dementia. Comparison of the classification accuracy for distinguishing MCI from dementia did not suggest a statistically significant advantage for either normative approach (Z = -0.29, p = .77; AUC = 0.86 for conventional and AUC = 0.86 for robust). In summary, these results indicate that using a robust normative group increases the likelihood of characterizing cognitive performance as low. However, there is not a clear advantage of using a robust over a conventional normative group when differentiating between MCI and dementia.

An Initial Empirical Operationalization of the Earliest Stages of the Alzheimer's Continuum.

PURPOSE: The Alzheimer's Continuum (AC) includes 2 preclinical stages defined by subjective cognitive complaints, transitional cognitive declines, and neurobehavioral symptoms. Operationalization of these stages is necessary for them to be applied in research. METHODS: Cognitively normal individuals with known amyloid biomarker status were selected from the National Alzheimer's Coordinating Center Uniform Data Set. Participants and their caregivers provided information on subjective cognitive complaints, neurobehavioral features, and objective cognitive functioning. PATIENTS: The sample included 101 amyloid positive (A+) and 447 amyloid negative (A-) individuals. RESULTS: Rates of subjective cognitive complaints (A+: 34.90%, A-: 29.90%) and neurobehavioral symptoms (A+: 22.40%, A-: 22.40%) did not significantly differ between A+/- individuals. However, the frequency of transitional cognitive decline was significantly higher among A+ (38.00%) than A- participants (24.90%). We explored various empirical definitions for defining the early stages of the AC among A+ participants. Rates of classification into AC stage 1 versus AC stage 2 varied depending on the number of symptoms required: 57.40% versus 42.60% (1 symptom), 28.70% versus 71.30% (2 symptoms), and 6.90% versus 93.10% (all 3 symptoms). CONCLUSION: The presence of 2 of the proposed symptom classes to separate AC stage 2 from stage 1 seems to provide a good empirical balance.

Development and Preliminary Validation of Standardized Regression-Based Change Scores as Measures of Transitional Cognitive Decline.

OBJECTIVE: An increasing focus in Alzheimer's disease and aging research is to identify transitional cognitive decline. One means of indexing change over time in serial cognitive evaluations is to calculate standardized regression-based (SRB) change indices. This paper includes the development and preliminary validation of SRB indices for the Uniform Data Set 3.0 Neuropsychological Battery, as well as base rate data to aid in their interpretation. METHOD: The sample included 1,341 cognitively intact older adults with serial assessments over 0.5-2 years in the National Alzheimer's Coordinating Center Database. SRB change scores were calculated in half of the sample and then validated in the other half of the sample. Base rates of SRB decline were evaluated at z-score cut-points, corresponding to two-tailed p-values of .20 (z = -1.282), .10 (z = -1.645), and .05 (z = -1.96). We examined convergent associations of SRB indices for each cognitive measure with each other as well as concurrent associations of SRB indices with clinical dementia rating sum of box scores (CDR-SB). RESULTS: SRB equations were able to significantly predict the selected cognitive variables. The base rate of at least one significant SRB decline across the entire battery ranged from 26.70% to 58.10%. SRB indices for cognitive measures demonstrated theoretically expected significant positive associations with each other. Additionally, CDR-SB impairment was associated with an increasing number of significantly declined test scores. CONCLUSIONS: This paper provides preliminary validation of SRB indices in a large sample, and we present a user-friendly tool for calculating SRB values.