Latent information in fluency lists predicts functional decline in persons at risk for Alzheimer disease

Applications of Artificial Intelligence in the Neuropsychological Assessment of Dementia: A Systematic Review

In the context of advancing healthcare, the diagnosis and treatment of cognitive disorders, particularly Mild Cognitive Impairment (MCI) and Alzheimer's Disease (AD), pose significant challenges. This review explores Artificial Intelligence (AI) and Machine Learning (ML) in neuropsychological assessment for the early detection and personalized treatment of MCI and AD. The review includes 37 articles that demonstrate that AI could be an useful instrument for optimizing diagnostic procedures, predicting cognitive decline, and outperforming traditional tests. Three main categories of applications are identified: (1) combining neuropsychological assessment with clinical data, (2) optimizing existing test batteries using ML techniques, and (3) employing virtual reality and games to overcome the limitations of traditional tests. Despite advancements, the review highlights a gap in developing tools that simplify the clinician's workflow and underscores the need for explainable AI in healthcare decision making. Future studies should bridge the gap between technical performance measures and practical clinical utility to yield accurate results and facilitate clinicians' roles. The successful integration of AI/ML in predicting dementia onset could reduce global healthcare costs and benefit aging societies.

Examining the diagnostic value of the mnemonic discrimination task for classification of cognitive status and amyloid-beta burden

Alzheimer's disease (AD) is the most common type of dementia, characterized by early memory impairments and gradual worsening of daily functions. AD-related pathology, such as amyloid-beta (Aβ) plaques, begins to accumulate many years before the onset of clinical symptoms. Predicting risk for AD via related pathology is critical as the preclinical stage could serve as a therapeutic time window, allowing for early management of the disease and reducing health and economic costs. Current methods for detecting AD pathology, however, are often expensive and invasive, limiting wide and easy access to a clinical setting. A non-invasive, cost-efficient platform, such as computerized cognitive tests, could be potentially useful to identify at-risk individuals as early as possible. In this study, we examined the diagnostic value of an episodic memory task, the mnemonic discrimination task (MDT), for predicting risk of cognitive impairment or Aβ burden. We constructed a random forest classification algorithm, utilizing MDT performance metrics and various neuropsychological test scores as input features, and assessed model performance using area under the curve (AUC). Models based on MDT performance metrics achieved classification results with an AUC of 0.83 for cognitive status and an AUC of 0.64 for Aβ status. Our findings suggest that mnemonic discrimination function may be a useful predictor of progression to prodromal AD or increased risk of Aβ load, which could be a cost-efficient, noninvasive cognitive testing solution for potentially wide-scale assessment of AD pathological and cognitive risk.

Characterizing phonemic fluency by transfer learning with deep language models

Though phonemic fluency tasks are traditionally indexed by the number of correct responses, the underlying disorder may shape the specific choice of words-both correct and erroneous. We report the first comprehensive qualitative analysis of incorrect and correct words generated on the phonemic ('S') fluency test, in a large sample of patients (n = 239) with focal, unilateral frontal or posterior lesions and healthy controls (n = 136). We conducted detailed qualitative analyses of the single words generated in the phonemic fluency task using categorical descriptions for different types of errors, low-frequency words and clustering/switching. We further analysed patients' and healthy controls' entire sequences of words by employing stochastic block modelling of Generative Pretrained Transformer 3-based deep language representations. We conducted predictive modelling to investigate whether deep language representations of word sequences improved the accuracy of detecting the presence of frontal lesions using the phonemic fluency test. Our qualitative analyses of the single words generated revealed several novel findings. For the different types of errors analysed, we found a non-lateralized frontal effect for profanities, left frontal effects for proper nouns and permutations and a left posterior effect for perseverations. For correct words, we found a left frontal effect for low-frequency words. Our novel large language model-based approach found five distinct communities whose varied word selection patterns reflected characteristic demographic and clinical features. Predictive modelling showed that a model based on Generative Pretrained Transformer 3-derived word sequence representations predicted the presence of frontal lesions with greater fidelity than models of native features. Our study reveals a characteristic pattern of phonemic fluency responses produced by patients with frontal lesions. These findings demonstrate the significant inferential and diagnostic value of characterizing qualitative features of phonemic fluency performance with large language models and stochastic block modelling.

Neuropsychological differential diagnosis of Alzheimer's disease and vascular dementia: a systematic review with meta-regressions

Introduction

Diagnostic classification systems and guidelines posit distinguishing patterns of impairment in Alzheimer's (AD) and vascular dementia (VaD). In our study, we aim to identify which diagnostic instruments distinguish them.

Methods

We searched PubMed and PsychInfo for empirical studies published until December 2020, which investigated differences in cognitive, behavioral, psychiatric, and functional measures in patients older than 64 years and reported information on VaD subtype, age, education, dementia severity, and proportion of women. We systematically reviewed these studies and conducted Bayesian hierarchical meta-regressions to quantify the evidence for differences using the Bayes factor (BF). The risk of bias was assessed using the Newcastle-Ottawa-Scale and funnel plots.

Results

We identified 122 studies with 17,850 AD and 5,247 VaD patients. Methodological limitations of the included studies are low comparability of patient groups and an untransparent patient selection process. In the digit span backward task, AD patients were nine times more probable (BF = 9.38) to outperform VaD patients (β g = 0.33, 95% ETI = 0.12, 0.52). In the phonemic fluency task, AD patients outperformed subcortical VaD (sVaD) patients (β g = 0.51, 95% ETI = 0.22, 0.77, BF = 42.36). VaD patients, in contrast, outperformed AD patients in verbal (β g = -0.61, 95% ETI = -0.97, -0.26, BF = 22.71) and visual (β g = -0.85, 95% ETI = -1.29, -0.32, BF = 13.67) delayed recall. We found the greatest difference in verbal memory, showing that sVaD patients outperform AD patients (β g = -0.64, 95% ETI = -0.88, -0.36, BF = 72.97). Finally, AD patients performed worse than sVaD patients in recognition memory tasks (β g = -0.76, 95% ETI = -1.26, -0.26, BF = 11.50).

Conclusion

Our findings show inferior performance of AD in episodic memory and superior performance in working memory. We found little support for other differences proposed by diagnostic systems and diagnostic guidelines. The utility of cognitive, behavioral, psychiatric, and functional measures in differential diagnosis is limited and should be complemented by other information. Finally, we identify research areas and avenues, which could significantly improve the diagnostic value of cognitive measures.

A comparison of techniques for deriving clustering and switching scores from verbal fluency word lists

Objective

To compare techniques for computing clustering and switching scores in terms of agreement, correlation, and empirical value as predictors of incident cognitive impairment (ICI).

Methods

We transcribed animal and letter F fluency recordings on 640 cases of ICI and matched controls from a national epidemiological study, amending each transcription with word timings. We then calculated clustering and switching scores, as well as scores indexing speed of responses, using techniques described in the literature. We evaluated agreement among the techniques with Cohen’s κ and calculated correlations among the scores. After fitting a base model with raw scores, repetitions, and intrusions, we fit a series of Bayesian logistic regression models adding either clustering and switching scores or speed scores, comparing the models in terms of several metrics. We partitioned the ICI cases into acute and progressive cases and repeated the regression analysis for each group.

Results

For animal fluency, we found that models with speed scores derived using the slope difference algorithm achieved the best values of the Watanabe–Akaike Information Criterion (WAIC), but with good net reclassification improvement (NRI) only for the progressive group (8.2%). For letter fluency, different models excelled for prediction of acute and progressive cases. For acute cases, NRI was best for speed scores derived from a network model (3.4%), while for progressive cases, the best model used clustering and switching scores derived from the same network model (5.1%). Combining variables from the best animal and letter F models led to marginal improvements in model fit and NRI only for the all-cases and acute-cases analyses.

Conclusion

Speed scores improve a base model for predicting progressive cognitive impairment from animal fluency. Letter fluency scores may provide complementary information.

Identifying neurocognitive disorder using vector representation of free conversation

In recent years, studies on the use of natural language processing (NLP) approaches to identify dementia have been reported. Most of these studies used picture description tasks or other similar tasks to encourage spontaneous speech, but the use of free conversation without requiring a task might be easier to perform in a clinical setting. Moreover, free conversation is unlikely to induce a learning effect. Therefore, the purpose of this study was to develop a machine learning model to discriminate subjects with and without dementia by extracting features from unstructured free conversation data using NLP. We recruited patients who visited a specialized outpatient clinic for dementia and healthy volunteers. Participants’ conversation was transcribed and the text data was decomposed from natural sentences into morphemes by performing a morphological analysis using NLP, and then converted into real-valued vectors that were used as features for machine learning. A total of 432 datasets were used, and the resulting machine learning model classified the data for dementia and non-dementia subjects with an accuracy of 0.900, sensitivity of 0.881, and a specificity of 0.916. Using sentence vector information, it was possible to develop a machine-learning algorithm capable of discriminating dementia from non-dementia subjects with a high accuracy based on free conversation.

Random Forest Model in the Diagnosis of Dementia Patients with Normal Mini-Mental State Examination Scores

Background: Mini-Mental State Examination (MMSE) is the most widely used tool in cognitive screening. Some individuals with normal MMSE scores have extensive cognitive impairment. Systematic neuropsychological assessment should be performed in these patients. This study aimed to optimize the systematic neuropsychological test battery (NTB) by machine learning and develop new classification models for distinguishing mild cognitive impairment (MCI) and dementia among individuals with MMSE ≥ 26. Methods: 375 participants with MMSE ≥ 26 were assigned a diagnosis of cognitively unimpaired (CU) (n = 67), MCI (n = 174), or dementia (n = 134). We compared the performance of five machine learning algorithms, including logistic regression, decision tree, SVM, XGBoost, and random forest (RF), in identifying MCI and dementia. Results: RF performed best in identifying MCI and dementia. Six neuropsychological subtests with high-importance features were selected to form a simplified NTB, and the test time was cut in half. The AUC of the RF model was 0.89 for distinguishing MCI from CU, and 0.84 for distinguishing dementia from nondementia. Conclusions: This simplified cognitive assessment model can be useful for the diagnosis of MCI and dementia in patients with normal MMSE. It not only optimizes the content of cognitive evaluation, but also improves diagnosis and reduces missed diagnosis.

Application Scenarios for Artificial Intelligence in Nursing Care: Rapid Review

Background

Artificial intelligence (AI) holds the promise of supporting nurses’ clinical decision-making in complex care situations or conducting tasks that are remote from direct patient interaction, such as documentation processes. There has been an increase in the research and development of AI applications for nursing care, but there is a persistent lack of an extensive overview covering the evidence base for promising application scenarios.

Objective

This study synthesizes literature on application scenarios for AI in nursing care settings as well as highlights adjacent aspects in the ethical, legal, and social discourse surrounding the application of AI in nursing care.

Methods

Following a rapid review design, PubMed, CINAHL, Association for Computing Machinery Digital Library, Institute of Electrical and Electronics Engineers Xplore, Digital Bibliography & Library Project, and Association for Information Systems Library, as well as the libraries of leading AI conferences, were searched in June 2020. Publications of original quantitative and qualitative research, systematic reviews, discussion papers, and essays on the ethical, legal, and social implications published in English were included. Eligible studies were analyzed on the basis of predetermined selection criteria.

Results

The titles and abstracts of 7016 publications and 704 full texts were screened, and 292 publications were included. Hospitals were the most prominent study setting, followed by independent living at home; fewer application scenarios were identified for nursing homes or home care. Most studies used machine learning algorithms, whereas expert or hybrid systems were entailed in less than every 10th publication. The application context of focusing on image and signal processing with tracking, monitoring, or the classification of activity and health followed by care coordination and communication, as well as fall detection, was the main purpose of AI applications. Few studies have reported the effects of AI applications on clinical or organizational outcomes, lacking particularly in data gathered outside laboratory conditions. In addition to technological requirements, the reporting and inclusion of certain requirements capture more overarching topics, such as data privacy, safety, and technology acceptance. Ethical, legal, and social implications reflect the discourse on technology use in health care but have mostly not been discussed in meaningful and potentially encompassing detail.

Conclusions

The results highlight the potential for the application of AI systems in different nursing care settings. Considering the lack of findings on the effectiveness and application of AI systems in real-world scenarios, future research should reflect on a more nursing care–specific perspective toward objectives, outcomes, and benefits. We identify that, crucially, an advancement in technological-societal discourse that surrounds the ethical and legal implications of AI applications in nursing care is a necessary next step. Further, we outline the need for greater participation among all of the stakeholders involved.

How well do people living with neurodegenerative diseases manage their finances? A meta-analysis and systematic review on the capacity to make financial decisions in people living with neurodegenerative diseases

Self and proxy reported questionnaires indicate that people living with a neurodegenerative disease (NDD) have more difficulties with financial decision-making (FDM) than healthy controls. Self-reports, however, rely on adequate insight into everyday functioning and might, therefore, be less reliable. The present study provides a comprehensive overview and meta-analysis of studies evaluating FDM in people living with an NDD. For this, the reliability of performance-based tests to consistently identify FDM difficulties in people living with an NDD compared to healthy controls is evaluated. Furthermore, the associations between FDM and disease severity, performances on standard measures of cognition and demographics are evaluated. All 47 included articles, consistently reported lower performances on performance-based FDM tests of people living with an NDD (including Alzheimer's disease, mild cognitive impairment, frontotemporal dementia, Parkinson's disease, multiple sclerosis or Huntington's disease) compared to healthy controls. The majority of studies, however, focused on Alzheimer's disease and mild cognitive impairment (k = 38). FDM performance appears to be related to cognitive decline, specifically in working memory, processing speed and numeracy.

Early Identification of Alzheimer's Disease in Mouse Models: Application of Deep Neural Network Algorithm to Cognitive Behavioral Parameters

Alzheimer's disease (AD) is a worldwide burden. Diagnosis is complicated by the fact that AD is asymptomatic at an early stage. Studies using AD-modeled animals offer important and useful insights. Here, we classified mice with a high risk of AD at a preclinical stage by using only their behaviors. Wild-type and knock-in AD-modeled (App NL-G-F/NL-G-F ) mice were raised, and their cognitive behaviors were assessed in an automated monitoring system. The classification utilized a machine learning method, i.e., a deep neural network, together with optimized stepwise feature selection and cross-validation. The AD risk could be identified on the basis of compulsive and learning behaviors (89.3% ± 9.8% accuracy) shown by AD-modeled mice in the early age (i.e., 8-12 months old) when the AD symptomatic cognitions were relatively underdeveloped. This finding reveals the advantage of machine learning in unveiling the importance of compulsive and learning behaviors for early AD diagnosis in mice.

[Diagnostics and Therapy 24/7? Artificial Intelligence as a Challenge and Opportunity in Psychiatry and Psychotherapy]

OBJECTIVE

The aim of the article is to enable a fundamental understanding of the potentials and requirements of Artificial Intelligence (AI) for psychiatrists in the present and for the development of future working environments. Psychiatrists will need to understand the function of AI-systems and personalized AI-assistants in therapy systems and as part of their patients' daily life.

METHOD

The article provides an overview of basic categories and fields of application of AI and machine learning in the diagnosis, prevention and therapy of mental disorders.

RESULTS

AI-applications will shape the prevention, diagnosis and treatment as well as the basic etiological understanding of mental disorders. Notably, the treatment of mental disorders is significantly influenced by commercial product development and assistance systems outside the medical system, as the corresponding developments can exploit large data pools with significantly lower restrictions.

CONCLUSION

Psychiatrists should now seize the opportunity to actively shape the implementation of AI-systems as otherwise key competences could be transferred to a primary field outside the medical system to the detriment of the patient and the therapist.

Assessing Intervention Effects in Sentence Processing: Object Relatives vs. Subject Control

Object relative clauses are harder to process than subject relative clauses. Under Grillo’s (2009) Generalized Minimality framework, complexity effects of object relatives are construed as intervention effects, which result from an interaction between locality constraints on movement (Relativized Minimality) and the sentence processing system. Specifically, intervention of the subject DP in the movement dependency is expected to generate a minimality violation whenever processing limitations render the moved object underspecified, resulting in compromised comprehension. In the present study, assuming Generalized Minimality, we compared the processing of object relatives with the processing of subject control in ditransitives, which, like object relatives, instantiates a syntactic dependency across an intervening DP. This comparison is justified by the current debate on whether Control should be analyzed as movement: if control involves movement of the controller DP, as proposed by Hornstein (1999), a parallel between the processing of object relatives and subject control in ditransitives may be anticipated on the basis of intervention. In addition, we explored whether general cognitive factors contribute to complexity effects ascribed to movement across a DP. Sixty-nine adult speakers of European Portuguese read sentences and answered comprehension probes in a self-paced reading task with moving-window display, comprising four experimental conditions: Subject Relatives; Object Relatives; Subject Control; Object Control. Furthermore, participants performed four supplementary tasks, serving as measures of resistance to interference, lexical knowledge, working memory capacity and lexical access ability. The results from the reading task showed that whereas object relatives were harder to process than subject relatives, subject control was not harder to process than object control, arguing against recent movement accounts of control. Furthermore, we found that whereas object relative complexity effects assessed by response times to comprehension probes interacted with Reading Span, object relative complexity effects assessed by comprehension accuracy and reading times did not interact with any of the supplementary tasks. We discuss these results in light of Generalized Minimality and the hypothesis of modularity in syntactic processing (Caplan and Waters, 1999).

Integrative Base Ontology for the Research Analysis of Alzheimer's Disease-Related Mild Cognitive Impairment

Early detection of mild cognitive impairment (MCI) has become a priority in Alzheimer's disease (AD) research, as it is a transitional phase between normal aging and dementia. However, information on MCI and AD is scattered across different formats and standards generated by different technologies, making it difficult to work with them manually. Ontologies have emerged as a solution to this problem due to their capacity for homogenization and consensus in the representation and reuse of data. In this context, an ontology that integrates the four main domains of neurodegenerative diseases, diagnostic tests, cognitive functions, and brain areas will be of great use in research. Here, we introduce the first approach to this ontology, the Neurocognitive Integrated Ontology (NIO), which integrates the knowledge regarding neuropsychological tests (NT), AD, cognitive functions, and brain areas. This ontology enables interoperability and facilitates access to data by integrating dispersed knowledge across different disciplines, rendering it useful for other research groups. To ensure the stability and reusability of NIO, the ontology was developed following the ontology-building life cycle, integrating and expanding terms from four different reference ontologies. The usefulness of this ontology was validated through use-case scenarios.

Which cognitive functions subserve clustering and switching in category fluency? Generalisations from an extended set of semantic categories using linear mixed-effects modelling

Clustering and switching are hypothesised to reflect the automatic and controlled components in category fluency, respectively, but how they are associated with cognitive functions has not been fully elucidated, due to several uncertainties. (1) The conventional scoring method that segregates responses by semantic categories could not optimally dissociate the automatic and controlled components. (2) The temporal structure of individual responses, as characterised by mean retrieval time (MRT) and mean switching time (MST), has seldom been analysed alongside the more well-studied variables, cluster size (CS) and number of switches (NS). (3) Most studies examined only one to a few semantic categories, raising concerns of generalisability. This study built upon a distance-based automatic clustering procedure, referred to as temporal-semantic distance procedure, to thoroughly characterise the category fluency performance. Linear mixed-effects (LME) modelling was applied to re-examine the differential associations of clustering and switching with cognitive functions with a sample of 80 university students. Our results revealed that although lexical retrieval speed (LRS) is clearly the determining factor for effective clustering and switching, matrix reasoning and processing speed also have significant roles to play, possibly in the processes of identifying and validating the semantic relationships. Interestingly, total fluency score was accurately predicted by the four clustering/switching indices alone; including the cognitive variables did not significantly improve the prediction. These findings underline the importance of the clustering and switching indices in explaining the category fluency performance and the cognitive demands in category fluency.

The Role of Word Properties in Performance on Fluency Tasks in People with Primary Progressive Aphasia

People with primary progressive aphasia (PPA) present language difficulties that require lengthy assessments and follow-ups. Despite individual differences, people with PPA are often classified into three variants that present some distinctive language difficulties. We analyzed the data of 6 fluency tasks (i.e., "F", "A", "S", "Fruits", "Animals", "Vegetables"). We used random forests to pinpoint relevant word properties and error types in the classification of the three PPA variants, conditional inference trees to indicate how relevant variables may interact with one another and ANOVAs to cross-validate the results. Results indicate that total word count helps distinguish healthy individuals (N = 10) from people with PPA (N = 29). Furthermore, mean familiarity differentiates people with svPPA (N = 8) from people with lvPPA (N = 10) and nfvPPA (N = 11). No other word property or error type was relevant in the classification. These results relate to previous literature, as familiarity effects have been reported in people with svPPA in naming and spontaneous speech. Also, they strengthen the relevance of using familiarity to identify a specific group of people with PPA. This paper enhances our understanding of what determines word retrieval in people with PPA, complementing and extending data from naming studies.

Brief Cognitive Tests for Distinguishing Clinical Alzheimer-Type Dementia From Mild Cognitive Impairment or Normal Cognition in Older Adults With Suspected Cognitive Impairment

BACKGROUND

The accuracy and harms of brief cognitive tests for identifying clinical Alzheimer-type dementia (CATD) are uncertain.

PURPOSE

To summarize evidence on accuracy and harms of brief cognitive tests for CATD in older adults with suspected cognitive impairment.

DATA SOURCES

Electronic bibliographic databases (from inception to November 2019) and systematic review bibliographies.

STUDY SELECTION

English-language, controlled observational studies in older adults that evaluated the accuracy of brief cognitive tests (standalone tests; memory, executive function, and language tests; and brief multidomain batteries) for distinguishing CATD from mild cognitive impairment (MCI) or normal cognition as defined by established diagnostic criteria. Studies with low or medium risk of bias (ROB) were analyzed.

DATA EXTRACTION

Two reviewers rated ROB. One reviewer extracted data; the other verified extraction accuracy.

DATA SYNTHESIS

Fifty-seven studies met analysis criteria. Many brief, single cognitive tests were highly sensitive and specific for distinguishing CATD from normal cognition. These included standalone tests (clock-drawing test, median sensitivity 0.79 and specificity 0.88 [8 studies]; Mini-Mental State Examination, 0.88 and 0.94 [7 studies]; Montreal Cognitive Assessment, 0.94 and 0.94 [2 studies]; and Brief Alzheimer Screen, 0.92 and 0.97 [1 study]), memory tests (list delayed recall, 0.89 and 0.94 [5 studies]), and language tests (category fluency, 0.92 and 0.89 [9 studies]). Accuracy was lower in distinguishing mild CATD from normal cognition and distinguishing CATD from MCI. No studies reported on testing harms.

LIMITATIONS

Studies were small. Few test metrics were evaluated by multiple studies. Few studies directly compared different tests, scores, cut points, or test combinations.

CONCLUSION

Many brief, single cognitive tests accurately distinguish CATD from normal cognition in older adults but are less accurate in distinguishing mild CATD from normal cognition or CATD from MCI. No studies reported on testing harms.

PRIMARY FUNDING SOURCE

Agency for Healthcare Research and Quality. (PROSPERO: CRD42018117897).

Artificial intelligence and neuropsychological measures: The case of Alzheimer's disease

One of the current challenges in the field of Alzheimer's disease (AD) is to identify patients with mild cognitive impairment (MCI) that will convert to AD. Artificial intelligence, in particular machine learning (ML), has established as one of more powerful approach to extract reliable predictors and to automatically classify different AD phenotypes. It is time to accelerate the translation of this knowledge in clinical practice, mainly by using low-cost features originating from the neuropsychological assessment. We performed a meta-analysis to assess the contribution of ML and neuropsychological measures for the automated classification of MCI patients and the prediction of their conversion to AD. The pooled sensitivity and specificity of patients' classifications was obtained by means of a quantitative bivariate random-effect meta-analytic approach. Although a high heterogeneity was observed, the results of meta-analysis show that ML applied to neuropsychological measures can lead to a successful automatic classification, being more specific as screening rather than prognosis tool. Relevant categories of neuropsychological tests can be extracted by ML that maximize the classification accuracy.

A Novel Ensemble-Based Machine Learning Algorithm to Predict the Conversion From Mild Cognitive Impairment to Alzheimer's Disease Using Socio-Demographic Characteristics, Clinical Information, and Neuropsychological Measures

Background: Despite the increasing availability in brain health related data, clinically translatable methods to predict the conversion from Mild Cognitive Impairment (MCI) to Alzheimer's disease (AD) are still lacking. Although MCI typically precedes AD, only a fraction of 20-40% of MCI individuals will progress to dementia within 3 years following the initial diagnosis. As currently available and emerging therapies likely have the greatest impact when provided at the earliest disease stage, the prompt identification of subjects at high risk for conversion to AD is of great importance in the fight against this disease. In this work, we propose a highly predictive machine learning algorithm, based only on non-invasively and easily in-the-clinic collectable predictors, to identify MCI subjects at risk for conversion to AD. Methods: The algorithm was developed using the open dataset from the Alzheimer's Disease Neuroimaging Initiative (ADNI), employing a sample of 550 MCI subjects whose diagnostic follow-up is available for at least 3 years after the baseline assessment. A restricted set of information regarding sociodemographic and clinical characteristics, neuropsychological test scores was used as predictors and several different supervised machine learning algorithms were developed and ensembled in final algorithm. A site-independent stratified train/test split protocol was used to provide an estimate of the generalized performance of the algorithm. Results: The final algorithm demonstrated an AUROC of 0.88, sensitivity of 77.7%, and a specificity of 79.9% on excluded test data. The specificity of the algorithm was 40.2% for 100% sensitivity. Conclusions: The algorithm we developed achieved sound and high prognostic performance to predict AD conversion using easily clinically derived information that makes the algorithm easy to be translated into practice. This indicates beneficial application to improve recruitment in clinical trials and to more selectively prescribe new and newly emerging early interventions to high AD risk patients.

A clinically-translatable machine learning algorithm for the prediction of Alzheimer's disease conversion: further evidence of its accuracy via a transfer learning approach

BACKGROUND

In a previous study, we developed a highly performant and clinically-translatable machine learning algorithm for a prediction of three-year conversion to Alzheimer's disease (AD) in subjects with Mild Cognitive Impairment (MCI) and Pre-mild Cognitive Impairment. Further tests are necessary to demonstrate its accuracy when applied to subjects not used in the original training process. In this study, we aimed to provide preliminary evidence of this via a transfer learning approach.

METHODS

We initially employed the same baseline information (i.e. clinical and neuropsychological test scores, cardiovascular risk indexes, and a visual rating scale for brain atrophy) and the same machine learning technique (support vector machine with radial-basis function kernel) used in our previous study to retrain the algorithm to discriminate between participants with AD (n = 75) and normal cognition (n = 197). Then, the algorithm was applied to perform the original task of predicting the three-year conversion to AD in the sample of 61 MCI subjects that we used in the previous study.

RESULTS

Even after the retraining, the algorithm demonstrated a significant predictive performance in the MCI sample (AUC = 0.821, 95% CI bootstrap = 0.705-0.912, best balanced accuracy = 0.779, sensitivity = 0.852, specificity = 0.706).

CONCLUSIONS

These results provide a first indirect evidence that our original algorithm can also perform relevant generalized predictions when applied to new MCI individuals. This motivates future efforts to bring the algorithm to sufficient levels of optimization and trustworthiness that will allow its application in both clinical and research settings.

Automatic Scoring of Semantic Fluency

In neuropsychological assessment, semantic fluency is a widely accepted measure of executive function and access to semantic memory. While fluency scores are typically reported as the number of unique words produced, several alternative manual scoring methods have been proposed that provide additional insights into performance, such as clusters of semantically related items. Many automatic scoring methods yield metrics that are difficult to relate to the theories behind manual scoring methods, and most require manually-curated linguistic ontologies or large corpus infrastructure. In this paper, we propose a novel automatic scoring method based on Wikipedia, Backlink-VSM, which is easily adaptable to any of the 61 languages with more than 100k Wikipedia entries, can account for cultural differences in semantic relatedness, and covers a wide range of item categories. Our Backlink-VSM method combines relational knowledge as represented by links between Wikipedia entries (Backlink model) with a semantic proximity metric derived from distributional representations (vector space model; VSM). Backlink-VSM yields measures that approximate manual clustering and switching analyses, providing a straightforward link to the substantial literature that uses these metrics. We illustrate our approach with examples from two languages (English and Korean), and two commonly used categories of items (animals and fruits). For both Korean and English, we show that the measures generated by our automatic scoring procedure correlate well with manual annotations. We also successfully replicate findings that older adults produce significantly fewer switches compared to younger adults. Furthermore, our automatic scoring procedure outperforms the manual scoring method and a WordNet-based model in separating younger and older participants measured by binary classification accuracy for both English and Korean datasets. Our method also generalizes to a different category (fruit), demonstrating its adaptability.

Talk2Me: Automated linguistic data collection for personal assessment

Language is one the earliest capacities affected by cognitive change. To monitor that change longitudinally, we have developed a web portal for remote linguistic data acquisition, called Talk2Me, consisting of a variety of tasks. In order to facilitate research in different aspects of language, we provide baselines including the relations between different scoring functions within and across tasks. These data can be used to augment studies that require a normative model; for example, we provide baseline classification results in identifying dementia. These data are released publicly along with a comprehensive open-source package for extracting approximately two thousand lexico-syntactic, acoustic, and semantic features. This package can be applied arbitrarily to studies that include linguistic data. To our knowledge, this is the most comprehensive publicly available software for extracting linguistic features. The software includes scoring functions for different tasks.

Language in Amnestic Mild Cognitive Impairment and Dementia of Alzheimer’s Type: Quantitatively or Qualitatively Different?

Background/Aims: The purpose of this study was to explore language differences between individuals diagnosed with amnestic mild cognitive impairment multiple domain (aMCIm) and those with probable Alzheimer’s disease, with a goal of (i) characterizing the language profile of aMCIm and (ii) determining whether the profiles of dementia of Alzheimer’s type (DAT) and aMCIm individuals are on a continuum of one diagnostic entity or represent two distinct ­cognitive disorders. Methods: Language data from 28 patients with consensus diagnosis of aMCIm and DAT derived from a retrospective chart review were compared to that of healthy controls. Results: A non-parametric statistic established that there was no significant difference between groups in age, years of education or duration of symptoms and that expressive language was found to be relatively intact in both patient groups. In contrast, both groups exhibited significant impairments on receptive language tests and on linguistically complex tasks that rely on episodic memory and executive functions. Individuals with aMCIm and DAT present with configurations of language features that are largely in parallel to each other and reflect predominantly quantitative differences. Conclusion: Language tests provide an important contribution to the diagnostic process in their capacity to identify language impairments at an early stage. Understanding the nature of language decline is critically important to the intervention process as this information would inform cognitive intervention approaches aimed at promoting quality of life in people living with MCI and dementia.

Evaluating Mild Cognitive Impairment in Essential Tremor: How Many and Which Neuropsychological Tests?

OBJECTIVES

Essential tremor (ET) confers an increased risk for developing both amnestic and non-amnestic mild cognitive impairment (MCI). Yet, the optimal measures for detecting mild cognitive changes in individuals with this movement disorder have not been established. We sought to identify the cognitive domains and specific motor-free neuropsychological tests that are most sensitive to mild deficits in cognition as defined by a Clinical Dementia Rating (CDR) of 0.5, which is generally associated with a clinical diagnosis of MCI.

METHODS

A total of 196 ET subjects enrolled in a prospective, longitudinal, clinical-pathological study underwent an extensive motor-free neuropsychological test battery and were assigned a CDR score. Logistic regression analyses were performed to identify the neuropsychological tests which best identified individuals with CDR of 0.5 (mild deficits in cognition) versus 0 (normal cognition).

RESULTS

In regression models, we identified five tests in the domains of Memory and Executive Function which best discriminated subjects with CDR of 0.5 versus 0 (86.9% model classification accuracy). These tests were the California Verbal Learning Test II Total Recall, Logical Memory II, Verbal-Paired Associates I, Category Switching Fluency, and Color-Word Inhibition.

CONCLUSIONS

Mild cognitive difficulty among ET subjects is best predicted by combined performance on five measures of memory and executive function. These results inform the nature of cognitive dysfunction in ET and the creation of a brief cognitive battery to assess patients with ET for cognitively driven dysfunction in life that could indicate the presence of MCI. (JINS, 2018, 24, 1084-1098).

Diagnostic and prognostic role of semantic processing in preclinical Alzheimer's disease

Relatively spared during most of the timeline of normal aging, semantic memory shows a subtle yet measurable decline even during the pre-clinical stage of Alzheimer's disease. This decline is thought to reflect early neurofibrillary changes and impairment is detectable using tests of language relying on lexical-semantic abilities. A promising approach is the characterization of semantic parameters such as typicality and age of acquisition of words, and propositional density from verbal output. Seminal research like the Nun Study or the analysis of the linguistic decline of famous writers and politicians later diagnosed with Alzheimer's disease supports the early diagnostic value of semantic processing and semantic memory. Moreover, measures of these skills may play an important role for the prognosis of patients with mild cognitive impairment.

A Clinically-Translatable Machine Learning Algorithm for the Prediction of Alzheimer's Disease Conversion in Individuals with Mild and Premild Cognitive Impairment

BACKGROUND

Available therapies for Alzheimer's disease (AD) can only alleviate and delay the advance of symptoms, with the greatest impact eventually achieved when provided at an early stage. Thus, early identification of which subjects at high risk, e.g., with MCI, will later develop AD is of key importance. Currently available machine learning algorithms achieve only limited predictive accuracy or they are based on expensive and hard-to-collect information.

OBJECTIVE

The current study aims to develop an algorithm for a 3-year prediction of conversion to AD in MCI and PreMCI subjects based only on non-invasively and effectively collectable predictors.

METHODS

A dataset of 123 MCI/PreMCI subjects was used to train different machine learning techniques. Baseline information regarding sociodemographic characteristics, clinical and neuropsychological test scores, cardiovascular risk indexes, and a visual rating scale for brain atrophy was used to extract 36 predictors. Leave-pair-out-cross-validation was employed as validation strategy and a recursive feature elimination procedure was applied to identify a relevant subset of predictors.

RESULTS

16 predictors were selected from all domains excluding sociodemographic information. The best model resulted a support vector machine with radial-basis function kernel (whole sample: AUC = 0.962, best balanced accuracy = 0.913; MCI sub-group alone: AUC = 0.914, best balanced accuracy = 0.874).

CONCLUSIONS

Our algorithm shows very high cross-validated performances that outperform the vast majority of the currently available algorithms, and all those which use only non-invasive and effectively assessable predictors. Further testing and optimization in independent samples will warrant its application in both clinical practice and clinical trials.

Genome-wide association study of language performance in Alzheimer's disease

Language impairment is common in prodromal stages of Alzheimer's disease (AD) and progresses over time. However, the genetic architecture underlying language performance is poorly understood. To identify novel genetic variants associated with language performance, we analyzed brain MRI and performed a genome-wide association study (GWAS) using a composite measure of language performance from the Alzheimer's Disease Neuroimaging Initiative (ADNI; n=1560). The language composite score was associated with brain atrophy on MRI in language and semantic areas. GWAS identified GLI3 (GLI family zinc finger 3) as significantly associated with language performance (p<5×10^-8). Enrichment of GWAS association was identified in pathways related to nervous system development and glutamate receptor function and trafficking. Our results, which warrant further investigation in independent and larger cohorts, implicate GLI3, a developmental transcription factor involved in patterning brain structures, as a putative gene associated with language dysfunction in AD.

Advancing Alzheimer's research: A review of big data promises

OBJECTIVE

To review the current state of science using big data to advance Alzheimer's disease (AD) research and practice. In particular, we analyzed the types of research foci addressed, corresponding methods employed and study findings reported using big data in AD.

METHOD

Systematic review was conducted for articles published in PubMed from January 1, 2010 through December 31, 2015. Keywords with AD and big data analytics were used for literature retrieval. Articles were reviewed and included if they met the eligibility criteria.

RESULTS

Thirty-eight articles were included in this review. They can be categorized into seven research foci: diagnosing AD or mild cognitive impairment (MCI) (n=10), predicting MCI to AD conversion (n=13), stratifying risks for AD (n=5), mining the literature for knowledge discovery (n=4), predicting AD progression (n=2), describing clinical care for persons with AD (n=3), and understanding the relationship between cognition and AD (n=3). The most commonly used datasets are AD Neuroimaging Initiative (ADNI) (n=16), electronic health records (EHR) (n=11), MEDLINE (n=3), and other research datasets (n=8). Logistic regression (n=9) and support vector machine (n=8) are the most used methods for data analysis.

CONCLUSION

Big data are increasingly used to address AD-related research questions. While existing research datasets are frequently used, other datasets such as EHR data provide a unique, yet under-utilized opportunity for advancing AD research.

Predicting mild cognitive impairment from spontaneous spoken utterances

INTRODUCTION

Trials in Alzheimer's disease are increasingly focusing on prevention in asymptomatic individuals. We hypothesized that indicators of mild cognitive impairment (MCI) may be present in the content of spoken language in older adults and be useful in distinguishing those with MCI from those who are cognitively intact. To test this hypothesis, we performed linguistic analyses of spoken words in participants with MCI and those with intact cognition participating in a clinical trial.

METHODS

Data came from a randomized controlled behavioral clinical trial to examine the effect of unstructured conversation on cognitive function among older adults with either normal cognition or MCI (ClinicalTrials.gov: NCT01571427). Unstructured conversations (but with standardized preselected topics across subjects) were recorded between interviewers and interviewees during the intervention sessions of the trial from 14 MCI and 27 cognitively intact participants. From the transcription of interviewees recordings, we grouped spoken words using Linguistic Inquiry and Word Count (LIWC), a structured table of words, which categorizes 2500 words into 68 different word subcategories such as positive and negative words, fillers, and physical states. The number of words in each LIWC word subcategory constructed a vector of 68 dimensions representing the linguistic features of each subject. We used support vector machine and random forest classifiers to distinguish MCI from cognitively intact participants.

RESULTS

MCI participants were distinguished from those with intact cognition using linguistic features obtained by LIWC with 84% classification accuracy which is well above chance 60%.

DISCUSSION

Linguistic analyses of spoken language may be a powerful tool in distinguishing MCI subjects from those with intact cognition. Further studies to assess whether spoken language derived measures could detect changes in cognitive functions in clinical trials are warrented.

Novel verbal fluency scores and structural brain imaging for prediction of cognitive outcome in mild cognitive impairment

INTRODUCTION

The objective of this study was to assess the utility of novel verbal fluency scores for predicting conversion from mild cognitive impairment (MCI) to clinical Alzheimer's disease (AD).

METHOD

Verbal fluency lists (animals, vegetables, F, A, and S) from 107 MCI patients and 51 cognitively normal controls were transcribed into electronic text files and automatically scored with traditional raw scores and five types of novel scores computed using methods from machine learning and natural language processing. Additional scores were derived from structural MRI scans: region of interest measures of hippocampal and ventricular volumes and gray matter scores derived from performing ICA on measures of cortical thickness. Over 4 years of follow-up, 24 MCI patients converted to AD. Using conversion as the outcome variable, ensemble classifiers were constructed by training classifiers on the individual groups of scores and then entering predictions from the primary classifiers into regularized logistic regression models. Receiver operating characteristic curves were plotted, and the area under the curve (AUC) was measured for classifiers trained with five groups of available variables.

RESULTS

Classifiers trained with novel scores outperformed those trained with raw scores (AUC 0.872 vs 0.735; P < .05 by DeLong test). Addition of structural brain measurements did not improve performance based on novel scores alone.

CONCLUSION

The brevity and cost profile of verbal fluency tasks recommends their use for clinical decision making. The word lists generated are a rich source of information for predicting outcomes in MCI. Further work is needed to assess the utility of verbal fluency for early AD.

The impacts of a GO-game (Chinese chess) intervention on Alzheimer disease in a Northeast Chinese population

A GO game can enhance mental health, but its effects on Alzheimer Disease (AD) remains unknown. To address the issue, 147 AD patients were randomly assigned into control (without GO-game intervention), Short-time GO-Game Intervention (SGGI, 1 h daily) and Long-time GO-game Intervention (LGGI, 2 h daily) groups. After 6-month follow-up, the game reduced the mean score of Montgomery-Asberg Depression Rating Scales (MADRS) of 4.72 (95% CI, 0.69 to 9.12) and Hospital Anxiety and Depression Scale (HADS) of 1.75 (95% CI, 0.17–3.68), and increased the mean score of Global Assessment of Functioning (GAF) of 4.95 (95% CI, −1.37–9.18) and RAND-36 of 4.61 (95% CI, −2.75–11.32) (P < 0.05 via controls). A GO-game intervention improved 9 of 11 items of KICA-dep (Kimberley Indigenous Cognitive Assessment of Depression). Meanwhile, serum levels of brain derived neurotrophic factor (BDNF) were higher in SGGI and LGGI groups (24.02 ± 7.16 and 28.88 ± 4.12 ng/ml respectively, P = 0.051) than those in controls (17.28 ± 7.75 ng/ml) (P < 0.001). The serum levels of BDNF showed a negative relation with MADRS and a positive relation with RAND-36 (P < 0.01). A GO-game intervention ameliorates AD manifestations by up-regulating BDNF levels.

Prediction of remission in obsessive compulsive disorder using a novel machine learning strategy

The study objective was to apply machine learning methodologies to identify predictors of remission in a longitudinal sample of 296 adults with a primary diagnosis of obsessive compulsive disorder (OCD). Random Forests is an ensemble machine learning algorithm that has been successfully applied to large-scale data analysis across vast biomedical disciplines, though rarely in psychiatric research or for application to longitudinal data. When provided with 795 raw and composite scores primarily from baseline measures, Random Forest regression prediction explained 50.8% (5000-run average, 95% bootstrap confidence interval [CI]: 50.3-51.3%) of the variance in proportion of time spent remitted. Machine performance improved when only the most predictive 24 items were used in a reduced analysis. Consistently high-ranked predictors of longitudinal remission included Yale-Brown Obsessive Compulsive Scale (Y-BOCS) items, NEO items and subscale scores, Y-BOCS symptom checklist cleaning/washing compulsion score, and several self-report items from social adjustment scales. Random Forest classification was able to distinguish participants according to binary remission outcomes with an error rate of 24.6% (95% bootstrap CI: 22.9-26.2%). Our results suggest that clinically-useful prediction of remission may not require an extensive battery of measures. Rather, a small set of assessment items may efficiently distinguish high- and lower-risk patients and inform clinical decision-making.