Perseverometry: a novel performance marker in dementia

Patients with Alzheimer s disease are frequently noted by their relatives or carers to repeat themselves in conversation. In the early stages such behaviour might be regarded with no more than mild annoyance, but worsening of the tendency often motivates seeking a medical opinion, and leads eventually to the diagnosis of AD. With worsening of the disease, repetitions become more frequent, so if there were a means of measuring their occurence, this could be used as a diagnostic aid and a tool for monitoring the condition. Disease monitoring in particular is critical to the evaluation of novel drug treatments for Alzheimer s. Recording of patients conversations with their family members would be intrusive, and analysis of any such tape recorded record to look for repetitions would be a very laborious process. The solution we propose here is to make a continuous recording of energy fluctuations from the vocal apparatus. Such information would not be interpretable as language (thus safeguarding privacy) but could be rapidly and automatically processed to reveal segments that represented a previously repeated statement or phrase (solving the practical problems of interpretation). The aim of the present study is to determine the best methods for acquiring and analysing the data, and to establish the feasibility of such a recording method from AD patients as they go about their day to day lives.

Perseveration is an abnormal form of behaviour in which a speaker excessively repeats the same words, phrases or actions. Perseverative speech is a common presenting feature of Alzheimer s disease (AD) and becomes more prominent as the condition progresses. Recording the occurrence and severity of such behaviour would thus provide not only a measurable diagnostic marker of AD, but also a means of monitoring the patient s clinical course. The latter would be particularly important when evaluating the response to disease-modifying treatments. Continuous recording of an individual s speech in a real world setting, however, would not only violate the privacy of patients and their associates or carers, but would be difficult to interpret without manually segmenting the sources of recorded speech. A data source that was both devoid of linguistic meaning and automatically interpretable using signal processing technology can be achieved by recording acoustic energy rather than speech. The present study aims to develop and test a device and analytical method for capturing and interpreting this information in real world environments, allowing perseverative speech to be detected and quantified.