What is it about?
Speech is an amazing resource for clinicians looking for markers of cognitive change, because it is an everyday tool that is strongly associated with other cognitive functions. However, analyzing speech can be challenging because of the high number of features and the relation between them. We conducted a systematic review to identify the speech features that can best detect subtle cognitive impairment in function of the type of speech task used in the studies. We also wanted to see how linguistic features were described in clinical papers, and we found out that the studies with the best results concerning linguistic markers of cognitive impairment are studies in which the linguistic variables are described in details, and in which the speech task performed by patients reflects a type of discourse in line with the goal of the study.
Why is it important?
Patients with mild neurocognitive disorders represent a growing population of all ages whose cognitive changes remain underdiagnosed. Current neuropsychological tests often fail to detect subtle cognitive change, creating a gap between the difficulties reported by the patients and their test scores. The absence of clear pathological results leads to underdiagnose individuals with mild neurocognitive disorders, left with an invisible disability. Analyzing speech represents a new approach to detect subtle cognitive change. However, biases might arise in results if a same speech feature is unwillingly measured differently or if the discourse task is incompatible with the study objectives. Relying on clear linguistic definitions for features and informed speech tasks will improve the robustness and reproducibility of speech methods and results.