Evaluating Toothbrushing Performance Using GMM-based Sound Recognition and Regression Analysis

カテゴリ: 国際会議

論文No: PS-02

グループ名: ACIS2015

発行日: 2015/10/15

著者名(英語): Joseph Korpela(Osaka University), Ryosuke Miyaji (Osaka University),Takuya Maekawa(Osaka University), Kazunori Nozaki (Osaka University),Hiroo Tamagawa(Osaka University)

キーワード: Toothbrushing, healthcare, smartphone,\naudio

要約(英語): This paper presents a method for evaluating toothbrushing performance using audio data collected from a smartphone. In our method, we recognize several classes of toothbrushing activities in audio data using an environmental sound recognition technique based on hidden Markov models. These recognition results are used to generate several independent variables, which are then used to train regression models for estimating evaluation scores for sessions of toothbrushing audio. The dependent variables used to train these regression models are derived from evaluation scores assigned to sessions of data by a dentist. Using these independent and dependent variables, the resulting regression models are able to estimate evaluation scores for toothbrushing audio that represent a dentist’s evaluation of toothbrushing performance. We evaluated our method on 94 sessions of toothbrushing audio, achieving 83.1% accuracy when comparing our estimated overall performance scores with those assigned by the dentist.

原稿種別: 英語

PDFファイルサイズ: 686 Kバイト