PROBLEM TO BE SOLVED: To provide an estimation method of a stomatognathic motion state which can connect a myogenic potential, even if comprehensively detected, with the stomatognathic motion state by one-to-one relation, allowing non-invasive, simple, and highly accurate estimation of the stomatognathic motion state.SOLUTION: The estimation method of a stomatognathic motion state includes: a biosignal detection means 1, which detects a myogenic potential signal, from the back area of a chin H reflecting a muscular movement which correlates with the stomatognathic motion state of a tongue T as a specific motion site and a neural network 10, which estimate and output the contact position and/or the contact strength of a tongue tip Ta against the internal surface J of an upper chin on the basis of the myogenic potential signal detected by the biosignal detection means 1. The neural network 10 has a function to carry out a learning process, which learns until the estimated state falls in an error within a predetermined range previously on the basis of a teachers signal about the stomatognathic motion, and an execution estimating process, which outputs the state, as an estimated execution state, which is estimated from another myogenic potential signal newly detected after the learning process is over.COPYRIGHT: (C)2013,JPO&INPIT【課題】 総合的な筋電位の検出であっても、これと顎口腔運動の状態とを一対一で関連付けることができるようにし、非侵襲的で簡便にしかも高精度で顎口腔運動の状態を推定することができるようにする。【解決手段】 特定の運動部位としての舌Tの顎口腔運動に関連する筋肉の動きが反映する顎裏エリアHから筋電位信号を検出する生体信号検出手段1と、生体信号検出手段1が検出した筋電位信号に基づいて舌先Taの上顎内面Jに対する接触位置及び/または接触力を推定して出力するニューラルネットワーク10とを備え、ニューラルネットワーク10は、推定に係る状態の誤差が所定範囲内になるまで、予め、顎口腔運動に係る教師信号に基づいて学習する学習プロセスと、学習プロセス終了後に新たに検出された筋電位信号について推定した状態を実行推定状態として出力する実行推定プロセスとを行う機能を備えた。【選択図】 図1
