Δelta is more than just an app for digitizing cognitive tests to bring them from paper to iPad. The analyses are based on years of research and the knowledge of scientists and experts in the fields of artificial intelligence, natural language processing, neurology and psychology. On this page you will find the research literature that forms the scientific basis of Δelta .
Fully automatic speech-based analysis of the semantic verbal fluency task.
Alexandra König; Nicklas Linz; Johannes Tröger; Maria Wolters; Jan Alexandersson; Philippe Robert:
Background: Semantic verbal fluency (SVF) tests are routinely used in screening for mild cognitive impairment (MCI). In this task, participants name as many items as possible of a semantic category under a time constraint. Clinicians measure task performance manually by summing the number of correct words and errors. More fine-grained variables add valuable information to clinical assessment, but are time-consuming. Therefore, the aim of this study is to investigate whether automatic analysis of the SVF could provide these as accurate as manual and thus, support qualitative screening of neurocognitive impairment.
Methods: SVF data were collected from 95 older people with MCI (n = 47), Alzheimer’s or related dementias (ADRD; n = 24), and healthy controls (HC; n = 24). All data were annotated manually and automatically with clusters and switches. The obtained metrics were validated using a classifier to distinguish HC, MCI, and ADRD.
Results: Automatically extracted clusters and switches were highly correlated (r = 0.9) with manually established values, and performed as well on the classification task separating HC from persons with ADRD (area under curve [AUC] = 0.939) and MCI (AUC = 0.758).
Conclusion: The results show that it is possible to automate fine-grained analyses of SVF data for the assessment of cognitive decline.
2018. Dementia and Geriatric Cognitive Disorders. (accepted for publication)
Telephone-based Dementia Screening I: Automated Semantic Verbal Fluency Assessment.
Johannes Tröger; Nicklas Linz; Alexandra König; Philippe Robert; Jan Alexandersson:
Dementia has a large economic impact on our society as cost-effective population-wide screening for early signs of dementia is still an unsolved medical supply resource problem. A solution should be fast, require a minimum of external material, and automatically indicate potential persons at risk of cognitive decline. Despite encouraging results, leveraging pervasive sensing technologies for automatic dementia screening, there are still two main issues: significant hardware costs or installation efforts and the challenge of effective pattern recognition. Conversely, automatic speech recognition (ASR) and speech analysis have reached sufficient maturity and allow for low-tech remote telephone-based screening scenarios. Therefore, we examine the technologic feasibility of automatically assessing a neuropsychological test—Semantic Verbal Fluency (SVF)–via a telephone-based solution. We investigate its suitability for inclusion into an automated dementia frontline screening and global risk assessment, based on concise telephone-sampled speech, ASR and machine learning classification. Results are encouraging showing an area under the curve (AUC) of 0.85. We observe a relatively low word error rate of 33% despite phone-quality speech samples and a mean age of 77 years of the participants. The automated classification pipeline performs equally well compared to the classifier trained on manual transcriptions of the same speech data. Our results indicate SVF as a prime candidate for inclusion into an automated telephone-screening system.
In: Proceedings of the 12th EAI International Conference on Pervasive Computing Technologies for Healthcare, PervasiveHealth ’18. ICST. In press 21-24 May 2018, New York, USA.
Language Modelling in the Clinical Semantic Verbal Fluency Task.
Nicklas Linz; Johannes Tröger; Hali Lindsay; Alexandra König; Philippe Robert; Jessica Peter; Jan Alexandersson:
Semantic Verbal Fluency (SVF) tests are common neuropsychological tasks, in which patients are asked to name as many words belonging to a semantic category as they can in 60 seconds. These tests are sensitive to even early forms of dementia caused by e.g. Alzheimer’s disease. Performance is usually measured as the total number of correct responses. Clinical research has shown that not only the raw count, but also production strategy is a relevant clinical marker. We employed language modelling (LM) as a natural technique to model production in this task. Comparing different LMs, we show that perplexity of a persons SVF production predicts dementia well (F1 = 0.83). Demented patients show significantly lower perplexity, thus are more predictable. Persons in advanced stages of de-mentia differ in predictability of word choice and production strategy-people in early stages only in predictability of production strategy.
In: proceedings of the International Conference on Language Resources and Evaluation (LREC) – Workshop on Resources and ProcessIng of linguistic, para-linguistic and extra-linguistic Data from people with various forms of cognitive/psychiatric impairments (RaPID-2), 8th of May 2018, Miyazaki, Japan.
Automated Analysis of Verbal Fluency Ability for Detection of Cognitive Impairment in Elderly People.
Alexandra König; Nicklas Linz; Johannes Tröger; Jan Alexandersson; Philippe Robert:
In: Proceedings of 26th European Congress of Psychiatry. European Congress of Psychiatry (EPA-18), 26th, March 3-6, Nice, France, Elsevier, 2018.
Fully Automated Speech-based Frontline Screening for Dementia.
Johannes Tröger; Nicklas Linz; Alexandra König; Jessica Peter; Jan Alexandersson; Philippe Robert:
In: Proceedings of 26th European Congress of Psychiatry. European Congress of Psychiatry (EPA-18), Elsevier, 2018.
Automated Speech-based Screening for Alzheimer’s Disease in a Care Service Scenario.
Johannes Tröger; Nicklas Linz; Jan Alexandersson; Alexandra König; Philippe Robert:
This paper describes a benchmark study for a lightweight and low-cost dementia screening tool. The tool is easy to administer, requires no additional experimentation material, and automatically evaluates and indicates potential subjects with dementia. The protocol foresees that subjects answer four distinct tasks, three of which are ordinary questions and one is a counting prompt. In our care use case, older people are assessed remotely via the tool, potentially even via telephone or within a daily care service routine. The assessment results are subsequently sent to professionals who initiate further steps. A machine learning classifier was trained on the French Dem@Care corpus. Solely utilizing vocal features, the classifier reaches 89% accuracy. Implications for the use case and further steps are discussed.
In: Proceedings of the 11th EAI International Conference on Pervasive Computing Technologies for Healthcare. International ICST Conference on Pervasive Computing Technologies for Healthcare (Pervasive Health-17), 11th EAI International Conference on Pervasive Computing Technologies for Healthcare, May 23-26, Barcelona, Spain, ACM Digital Library, 2017.
Using Neural Word Embeddings in the Analysis of the Clinical Semantic Verbal Fluency Task.
Nicklas Linz; Johannes Tröger; Jan Alexandersson; Alexandra König:
The Semantic Verbal Fluency Task is a common neuropsychological assessment for cognitive disorders: patients are prompted to name as many words from a semantic category as possible in a time interval; the count of correctly named concepts is assessed. Patients often organise their retrieval around semantically related clusters. The definition of clusters is usually based on handmade taxonomies and the patient’s performance is manually evaluated. In order to overcome limitations of such an approach, we propose a statistical method using distributional semantics. Based on transcribed speech samples from 100 French elderly, 53 diagnosed with Mild Cognitive Impairment and 47 healthy, we used distributional semantic models to cluster words in each sample and compare performance with a taxonomic baseline approach in a realistic classification task. The distributional models outperform the baseline. Comparing different linguistic corpora as basis for the models, our results indicate that models trained on larger corpora perform better.
In: 12th International Conference on Computational Semantics. International Conference on Computational Semantics (IWCS-17), 12th, September 20-22, Montpellier, France, 2017.
Predicting Dementia Screening and Staging Scores From Semantic Verbal Fluency Performance.
Nicklas Linz; Johannes Tröger; Jan Alexandersson; Maria Wolters; Alexandra König; Philippe Robert:
The standard dementia screening tool Mini Mental State Examination (MMSE) and the standard dementia staging tool Clinical Dementia Rating Scale (CDR) are prominent methods for answering questions whether a person might have dementia and about the dementia severity respectively. These methods are time consuming and require well-educated personnel to administer. Conversely, cognitive tests, such as the Semantic Verbal Fluency (SVF), demand little time. With this as a starting point, we investigate the relation between SVF results and MMSE/CDR-SOB scores. We use regression models to predict scores based on persons’ SVF performance. Over a set of 179 patients with different degree of dementia, we achieve a mean absolute error of of 2.2 for MMSE (range 0-30) and 1.7 for CDR-SOB (range 0-18). True and predicted scores agree with a Cohen’s κ of 0.76 for MMSE and 0.52 for CDR-SOB. We conclude that our approach has potential to serve as a cheap dementia screening, possibly even in non-clinical settings.
In: 2017 IEEE 17th International Conference on Data Mining Workshops (ICDMW). Workshop on Data Mining for Ageing, Rehabilitation and Independent Assisted Living (ARIAL-17), located at 17th, November 18-21, New Orleans, LA, USA, IEEE Computer Society, 2017
Early detection of cognitive disorders such as dementia on the basis of speech analysis – a cross-linguistic comparison of speech features.
Nicklas Linz; Johannes Tröger; Jan Alexandersson; Alexandra König:
The people best placed to spot early cognitive decline are carers, social workers, and family. But there is a clear lack of affordable, usable screening apps that people without medical training can use to validate these concerns and to provide actionable data for medical professionals. The study aims to validate a new tool for fully-automated, reliable, unobtrusive, self-managed screening for cognitive decline, in particular dementia, and other cognitive disorders based on automatic speech analysis.
In: Proceedings of the Alzheimer’s Association International Conference (AAIC). Alzheimer’s Association International Conference (AAIC-17), July 16-20, London, United Kingdom, Alzheimer’s Association, 2017.