Our research
at a glance

With a focus on human language and cognition.

Linguistic Biomarkers and AI

Linguistic biomarkers are particular traits or characteristics in language and speech that can be quantified and utilized to identify and monitor the progression of neurological or psychiatric disorders affecting language. These biomarkers encompass a broad spectrum of features, including alterations in vocabulary, syntax, semantic content, speech rate, prosody, and other phonological, morphological, and discourse properties due to impairment. For instance, an individual with Alzheimer's disease may display a reduction in vocabulary, increased utilization of pronouns instead of proper nouns, and challenges in constructing complex sentences. These specific language alterations can act as linguistic biomarkers for Alzheimer's disease. The detection and analysis of these linguistic biomarkers are pivotal in the early diagnosis, monitoring of disease progression, and evaluation of the efficacy of interventions.

Open Brain AI facilitates access to linguistic biomarkers for language-impacting conditions. It provides a set of tools, invaluable for clinicians and researchers in eliciting linguistic biomarkers for various language-affecting conditions. Open Brain AI supports researchers and clinicians in assessing the influence of several factors, including prosodic and segmental measures (e.g., fundamental frequency, and vowel formants), pause duration, the count of pauses and filled pauses (e.g., um and hm), and phonological measures like sound deletions, insertions, transpositions, variations in speech sound quality (e.g., vowels and consonants), intonation, and voice quality impairments.

Open Brain AI employs advanced Natural Language Processing (NLP) methods and machine learning to extract novel grammatical measures and assess existing ones from the automated transcriptions of speech productions and writings. Moreover, Open Brain AI analyzes texts for lexical richness measures and assesses the significance of related measures such as utterance length, phonemes-to-word ratio, content words, function words, part of speech ratio, and morphological and semantic information about gender, person, number, and time. It can also pinpoint the syntactic role of textual constituents using automated syntactic analysis with parsers for English, Norwegian, Swedish, Greek, and other languages.

Additionally, Open Brain AI extracts semantic relationships from texts and assists researchers in scoring semantic performance, utilizing semantic distance metrics, known as word and textual embeddings, which characterize the relationships of words provided in speech productions.

Furthermore, Open Brain AI identifies the discourse properties that differentiate the speech of individuals with speech and language impairment from healthy controls by extracting information about cohesion and coherence. It can employ idea density as a cognitive decline indicator and use topic classification machine learning models to extract discourse macrostructure insights. Sentiment analysis can also be employed to interpret stance, such as the positive and negative emotions associated with pathology.

In conclusion, Open Brain AI is a comprehensive set of tools that empowers clinicians and researchers to better comprehend and analyze the speech of individuals with language impairments and healthy controls by evaluating various linguistic, syntactic, semantic, and discourse properties.