Description

Frailty is a common clinical syndrome especially in older adults that carries an increased risk for poor health outcomes including falls, incident disability, hospitalization, and mortality. The early detection of frailty is of importance in many patient populations to predict treatment outcomes, identify patient needs and coordinate efficient and meaningful care. An electronic assessment of the degree of distress in patients, who are unable to report, would be important to be able to routinely and objectively identify suffering in these patients. Digital voice analysis (DVA) gathers speech samples from individuals via different kinds of recording devices (smartphone, tablet, etc.) and examines a large variety of specific acoustic parameters such as for example frequency and voice quality features. This study is to analyse the potential to evaluate distress and frailty through digital voice analysis. On the contrary to the existing studies, it is intended to record audio and clinical evaluation data from the same subject multiple times during several weeks to be able to analyse temporal changes. This will allow to not only perform inter-subject but as well intra-subject comparisons of changes in audio features with changes of the patient's wellbeing over time. To make the patient speak as freely and relaxed as possible, the patient will describe different images. Different features will be extracted from the audios and potential candidates for a larger patient study will be identified, if data quantity permits using machine learning algorithms. Therefore this study evaluates if it is feasible to gather digital voice samples for voice analyses from cancer patients alongside conventional assessments for frailty (G8 questionnaire and distress (Distress Thermometer) to conduct first, preliminary analyses for identification of potential correlates between voice features and frailty or distress and between changes over time.