FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2896475994> ?p ?o ?g. }

• W2896475994 abstract "Diagnosis of Alzheimer's disease in the clinic is challenging, particularly in the early stages. A simple, non-invasive, inexpensive clinical test could be of great value. Speculating that such a test based on a speech sample might be possible, we have developed a range of technologies needed to produce such a test: a speech acquisition protocol (describing a picture), digital recording the speech and initial quality control, automatic speech recognition (ASR) to produce a transcript (with punctuation), extraction of signal and linguistic features. In addition we applied machine learning technologies to extract multiple feature subsets for the diagnostic task, and combining them into a single ensemble diagnosis predictor. Commercial ASR systems tested proved inadequate for several reasons, so we developed an ASR technology aimed at this specific application. It includes inserting punctuation as required for some linguistic features. We acquired a modest number (72) of speech samples, and augmented these with additional (140) samples from the Dementia Bank. Each subject provided one sample (cross sectional design), and a clinically verified diagnosis of probably Alzheimer's disease or cognitive normalcy. A substantial battery (232) features were automatically extracted from the audio signal and from the transcripts, and combined with a psychological test (MMSE) and four demographic features. These samples were used for feature set discovery, in a 10-fold cross validation fashion. A set of 12 feature sets were found promising, all including the MMSE combined with 2 or more of 10 speech-based features. These 12 classifiers were combined into an ensemble classifier that performed better than the MMSE alone, and achieved an AUC value > 0.95. An examination of the errors showed that they tended to be Alzheimer's subjects with high (normal range) MMSE scores, but that more than half of such cases were correctly detected with the speech features. We believe these results provide initial evidence that a speech-based diagnostic for Alzheimer's is feasible. The technology is now ready to be applied to larger clinical sample sets. We seek clinical collaborators willing to cooperate in this research, ideally applied to longitudinal samples." @default.
• W2896475994 created "2018-10-26" @default.
• W2896475994 creator A5003159330 @default.
• W2896475994 creator A5073541706 @default.
• W2896475994 creator A5074804583 @default.
• W2896475994 date "2018-07-01" @default.
• W2896475994 modified "2023-10-16" @default.
• W2896475994 title "TD‐P‐020: TOWARD AN AUTOMATIC SPEECH‐BASED DIAGNOSTIC TEST FOR ALZHEIMER'S DISEASE" @default.
• W2896475994 doi "https://doi.org/10.1016/j.jalz.2018.06.2036" @default.
• W2896475994 hasPublicationYear "2018" @default.
• W2896475994 type Work @default.
• W2896475994 sameAs 2896475994 @default.
• W2896475994 citedByCount "0" @default.
• W2896475994 crossrefType "journal-article" @default.
• W2896475994 hasAuthorship W2896475994A5003159330 @default.
• W2896475994 hasAuthorship W2896475994A5073541706 @default.
• W2896475994 hasAuthorship W2896475994A5074804583 @default.
• W2896475994 hasConcept C138885662 @default.
• W2896475994 hasConcept C142724271 @default.
• W2896475994 hasConcept C154945302 @default.
• W2896475994 hasConcept C169903167 @default.
• W2896475994 hasConcept C177264268 @default.
• W2896475994 hasConcept C199360897 @default.
• W2896475994 hasConcept C2776401178 @default.
• W2896475994 hasConcept C2779134260 @default.
• W2896475994 hasConcept C2779483572 @default.
• W2896475994 hasConcept C28490314 @default.
• W2896475994 hasConcept C41008148 @default.
• W2896475994 hasConcept C41895202 @default.
• W2896475994 hasConcept C52622490 @default.
• W2896475994 hasConcept C540372491 @default.
• W2896475994 hasConcept C71924100 @default.
• W2896475994 hasConcept C95623464 @default.
• W2896475994 hasConceptScore W2896475994C138885662 @default.
• W2896475994 hasConceptScore W2896475994C142724271 @default.
• W2896475994 hasConceptScore W2896475994C154945302 @default.
• W2896475994 hasConceptScore W2896475994C169903167 @default.
• W2896475994 hasConceptScore W2896475994C177264268 @default.
• W2896475994 hasConceptScore W2896475994C199360897 @default.
• W2896475994 hasConceptScore W2896475994C2776401178 @default.
• W2896475994 hasConceptScore W2896475994C2779134260 @default.
• W2896475994 hasConceptScore W2896475994C2779483572 @default.
• W2896475994 hasConceptScore W2896475994C28490314 @default.
• W2896475994 hasConceptScore W2896475994C41008148 @default.
• W2896475994 hasConceptScore W2896475994C41895202 @default.
• W2896475994 hasConceptScore W2896475994C52622490 @default.
• W2896475994 hasConceptScore W2896475994C540372491 @default.
• W2896475994 hasConceptScore W2896475994C71924100 @default.
• W2896475994 hasConceptScore W2896475994C95623464 @default.
• W2896475994 hasIssue "7S_Part_3" @default.
• W2896475994 hasLocation W28964759941 @default.
• W2896475994 hasOpenAccess W2896475994 @default.
• W2896475994 hasPrimaryLocation W28964759941 @default.
• W2896475994 hasRelatedWork W1575762866 @default.
• W2896475994 hasRelatedWork W205648239 @default.
• W2896475994 hasRelatedWork W2348531541 @default.
• W2896475994 hasRelatedWork W2368779261 @default.
• W2896475994 hasRelatedWork W2474469336 @default.
• W2896475994 hasRelatedWork W2546942002 @default.
• W2896475994 hasRelatedWork W2553408784 @default.
• W2896475994 hasRelatedWork W2778699561 @default.
• W2896475994 hasRelatedWork W4281689716 @default.
• W2896475994 hasRelatedWork W4320802741 @default.
• W2896475994 hasVolume "14" @default.
• W2896475994 isParatext "false" @default.
• W2896475994 isRetracted "false" @default.
• W2896475994 magId "2896475994" @default.
• W2896475994 workType "article" @default.