JP2004206401A - Activity pattern anomaly monitoring system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an activity pattern anomaly monitoring system capable of quickly and positively finding anomaly of a person. <P>SOLUTION: A data processing part 26 derives an activity pattern symbol string on the basis of living activity data sent by a housing system 10 via a network 14, and it derives a usual activity pattern group on the basis of the derived activity pattern symbol string. An activity pattern symbol string of an observation date is derived on the basis of living activity data of the observation date sent by the housing system 10 via the network 14, similarity is measured by comparing the derived activity pattern symbol string of the observation date with the derived usual activity pattern group, determination of whether or not an activity pattern of the observation date is usual, and its determination result is sent and outputted to terminals 1-3 via a network 16. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

即ち、類似度Ｒ_５は、数式２により計測される。
【００３４】
【数２】

同様にして、類似度Ｒ_１〜Ｒ_４を計測すると、Ｒ_１＝０、Ｒ_２＝０、Ｒ_３＝０、Ｒ_４＝０となる。
【００３５】
データ処理部２６は、全ての日常行動パターンの組み合わせにおいて比較が行なわれたと判定された場合には（ステップＳ２６）、観測日における行動パターンが日常的であるか否かの判定を行なう（ステップＳ２７）。即ち、ステップＳ２５において計測された日常行動パターンの組み合わせそれぞれに対する類似度の中で最も高い類似度の値を取り出す。そして、その取り出された類似度の値が閾値を超えているか否かの判定を行なう。類似度の値が閾値を超えていると判定された場合には、観測日における行動パターンが日常行動パターンの組み合わせの何れか１つの日常行動パターンの組み合わせとの類似性があると判断され、日常的であると判定する。即ち、上述の例によれば、ステップＳ２６において計測された類似度Ｒ_ｉは、Ｒ_１＝０、Ｒ_２＝０、Ｒ_３＝０、Ｒ_４＝０、Ｒ_５＝０．７５となり、最も高い類似度の値は、０．７５である。ここで、例えば、閾値を０．６と設定した場合に、最も高い類似度の値０．７５が閾値０．６を超えているので、観測日における行動パターンは日常的であると判定される。
【００３６】
一方、データ処理部２６は、何れの日常行動パターンの組み合わせに対する類似度の値も閾値を超えていないと判定された場合には、観測日における行動パターンが日常行動パターンの組み合わせの何れにも類似性がないと判断され、日常的でないと判定する。次に、ステップＳ２７における判定結果を通信制御部３２及びネットワーク１６を介して端末１〜３に送信する。そして、端末１〜３は、判定結果を表示等により出力する（ステップＳ２８）。即ち、ステップＳ２７において日常的であると判定された場合には、その旨の出力を行なう。一方、ステップＳ２７において日常的でないと判定された場合には、被験者に異変が生じていると判断し、異変である旨の出力を行なう。
【００３７】
この第１の実施の形態にかかる行動パターン異変監視システムによれば、所定時間中に例えば便所等の特定の場所に滞在した場合には、その特定の場所を滞在場所とし、所定時間中に特定の場所に滞在していない場合には、所定時間中に最も長く滞在した場所を滞在場所として、行動パターン記号列を導出する。従って、細かい変化に対応した行動パターン記号列を導出することができるため、例えば便所等の異変の起こりやすい場所への移動を含む行動パターンを確実に把握することができる。また、導出された行動パターン記号列に基づいて日常行動パターン群を導出し、その導出された日常行動パターン群に基づいて多様な日常行動パターンを持っている人若しくは日常行動パターンを限定できない人等にも対応することができる高精度な日常行動パターンの組み合わせを導出することができる。従って、細かい変化に対応した観測日における行動パターン記号列と高精度な日常行動パターンの組み合わせとの類似度を計測することができるため、早期かつ確実に行動パターンの変化及び異変を発見することができる。
【００３８】
次に、図４及び図５を参照して、この発明の第２の実施の形態にかかる行動パターン異変監視システムについて説明する。
【００３９】
この第２の実施の形態にかかる行動パターン異変監視システムは、第１の実施の形態のかかる行動パターン異変監視システムが過去数か月分の全ての生活行動データに基づいて日常行動パターンを導出するのに対し、過去数か月分の所定の曜日の生活行動データに基づいて日常行動パターン群を導出する。従って、日常行動パターンが曜日毎に規則性のある人に使用することが望ましいシステムである。
【００４０】
なお、この第２の実施の形態にかかる行動パターン異変監視システムは、図１に示す第１の実施の形態にかかる行動パターン異変監視システムと同一の構成である。この第２の実施の形態にかかる行動パターン異変監視システムにおいては、第１の実施の形態にかかる行動パターン異変監視システムにおいて、全ての生活行動データに基づく日常行動パターン群が日常行動パターン記憶部３０に記憶されるのに代えて、所定の曜日における日常行動パターン群が日常行動パターン記憶部３０に記憶される。その他の点においては、第１の実施の形態にかかる行動パターン異変監視システムと同一である。
【００４１】
まず、住居システム１０において収集された生活行動データがネットワーク１４を介してセンタ内管理サーバ１２に送信され記憶される。なお、住居システム１０における動作は、第１の実施の形態のかかる住居システム１０における動作と同一であるため、詳細な説明を省略する。
【００４２】
次に、図４に示すフローチャートを参照して、行動パターン異変監視システムにおける所定の曜日の日常行動パターン導出のための処理について説明する。データ処理部２６は、生活行動データ記憶部２８により記憶されている過去数か月分の所定の曜日の被験者の移動場所及び時間を含む生活行動データを読み込む（ステップＳ３０）。次にステップＳ３０において読み込まれた過去数か月分の所定の曜日のそれぞれの日の行動パターン記号列を導出する（ステップＳ３１）。ステップＳ３１において導出された過去数か月分の所定の曜日のそれぞれの日の行動パターン記号列を行動パターン記号列記億部２９に記憶させる（ステップＳ３２）。なお、行動パターン記号列の導出は、第１の実施の形態にかかる行動パターン記号列の導出と同一のため、詳細な説明を省略する。
【００４３】
次に、データ処理部２６は、ステップＳ３２において記憶された過去数か月分の所定の曜日の行動パターン記号列に基づいて、所定の曜日の日常行動パターンを導出する（ステップＳ３３）。即ち、類似性の判定を行なうことができる関数（第１の関数）を用いて、ステップＳ３２において記憶された過去数か月分の所定の曜日のそれぞれの日の行動パターン記号列の類似度を計測する。その計測された類似度に基づいて、何種類かの所定の曜日の日常行動パターン（日常行動パターン群）を導出する。次に、ステップＳ３２において記憶された過去数か月分の所定の曜日の行動パターン記号列とステップＳ３３において導出された所定の曜日の日常行動パターン群に基づいて、過去数か月分の所定の曜日のそれぞれの日に対応した所定の曜日の日常行動パターンの組み合わせを導出する（ステップＳ３４）。
【００４４】
例えば、ステップＳ３３において１０種類の月曜日の日常行動パターンＡ，Ｂ，Ｃ，Ｄ，Ｅ，Ｆ，Ｇ，Ｈ，Ｉ，Ｊが導出されたとする。次に、過去数か月中の月曜日のそれぞれの日の行動パターンとステップＳ１３において導出された月曜日の日常行動パターンＡ〜Ｊ、若しくは月曜日の日常行動パターンＡ〜Ｊの組み合わせとを比較して、類似している月曜日の日常行動パターンの組み合わせを導出する。即ち、過去数ヶ月の中のある月曜日の行動パターンが月曜日の日常行動パターンＡ及びＤの両方に似ている場合には、類似している月曜日の日常行動パターンの組み合わせとして月曜日の日常行動パターンの組み合わせ{Ａ，Ｄ}が導出され、過去数ヶ月の中のある月曜日の行動パターンは月曜日の日常行動パターンの組み合わせ{Ａ，Ｄ}に分類される。一方、過去数ヶ月の中の別のある月曜日の行動パターンに類似している月曜日の日常行動パターンの組み合わせとして月曜日の日常行動パターンの組み合わせ{Ａ}が導出された場合に、過去数ヶ月の中の別のある月曜日の行動パターンは月曜日の日常行動パターンの組み合わせ{Ａ}に分類される。ここで、月曜日の日常行動パターンの組み合わせ{Ａ}は、月曜日の日常行動パターンＡと同一であるが、月曜日の日常行動パターンの組み合わせの１つであるとみなす。例えば、以下のように月曜日の日常行動パターンの組み合わせを導出する。
１回目の月曜日 {Ａ}
２回目の月曜日 {Ａ，Ｄ}
３回目の月曜日 {Ｂ}
４回目の月曜日 {Ｃ}
５回目の月曜日 {Ｄ，Ａ}
６回目の月曜日 {Ａ，Ｂ，Ｃ，Ｄ}
７回目の月曜日 {Ａ，Ｄ}
８回目の月曜日 {Ｃ}
上述のステップＳ３４における導出結果は、過去数ヶ月の中における８回目の月曜までの日常行動パターンの組み合わせの一例である。上述のステップＳ３４における導出結果に基づいて、月曜日の日常行動パターンの組み合わせを以下のように抽出することができる。
{Ａ}、{Ｂ}、{Ｃ}、{Ａ，Ｄ}、{Ａ，Ｂ，Ｃ，Ｄ}
なお、この実施の形態においては、ステップＳ３２において記憶された過去数か月分の所定の曜日の行動パターンに基づいて、所定の曜日の日常行動パターン群を導出し（ステップＳ３３）、ステップＳ３２において記憶された過去数か月分の所定の曜日の行動パターン及びステップＳ３３において導出された所定の曜日の日常行動パターン群に基づいて、所定の曜日の日常行動パターンの組み合わせを導出する（ステップＳ３４）。しかしながら、ステップＳ３２において記憶された過去数か月分の所定の曜日のそれぞれの日の行動パターンに基づいて、所定の曜日の日常行動パターンを導出し、所定の曜日の日常行動パターンの組み合わせを導出してもよい。即ち、ステップＳ３３における所定の曜日の日常行動パターン群の導出とステップＳ３４における所定の曜日の日常行動パターンの組み合わせの導出を同時に行ってもよい。同時に行う場合には、過去数ヶ月の中のある所定の曜日の行動パターンと所定の曜日の日常行動パターンとの比較を行う過程において所定の曜日の日常行動パターン及び所定の曜日の日常行動パターンの組み合わせが増加するので、上述の例と同一の条件下においては、以下のような所定の曜日の日常行動パターンの組み合わせとなる。
１回目の月曜日 {Ａ}
２回目の月曜日 {Ｂ}
３回目の月曜日 {Ｃ}
４回目の月曜日 {Ｄ}
５回目の月曜日 {Ｂ}
６回目の月曜日 {Ｂ，Ｃ，Ｄ}
７回目の月曜日 {Ｂ}
８回目の月曜日 {Ｄ}
次に、ステップＳ３３において導出された所定の曜日の日常行動パターン群及びステップＳ３４において導出された所定の曜日の日常行動パターンの組み合わせを日常行動パターン記憶部３０に記憶させる（ステップＳ３５）。
【００４５】
次に、図５に示すフローチャートを参照して、行動パターン異変監視システムにおける観測日の行動パターンに異変があるか否かの判定を行なうための処理について説明する。データ処理部２６は、ステップＳ４０〜ステップＳ４２において、図３におけるステップＳ２０〜ステップＳ２２と同様の処理を行い、観測日における行動パターン記号列を導出し（ステップＳ４０、Ｓ４１）、その導出された行動パターン記号列を行動パターン記号列２９に記憶させる（ステップＳ４２）。
【００４６】
次に、データ処理部２６は、日常行動パターン記憶部３０に記憶されている観測日の曜日の日常行動パターン群及び観測日の曜日の日常行動パターンの組み合わせを読み込む（ステップＳ４３）。ステップＳ４２において記憶された観測日における行動パターン記号列に基づいて、観測日における行動パターンの組み合わせを導出する（ステップＳ４４）。ステップＳ４４において導出された観測日における行動パターンの組み合わせとステップＳ４３で読み込まれた観測日における曜日の日常行動パターンの組み合わせの何れか１つとを、類似性の判定を行うことができる関数（第２の関数）を用いて比較し、類似度を計測する（ステップＳ４５）。
【００４７】
例えば、観測日における行動パターンの組み合わせとして{Ａ，Ｂ，Ｃ}が導出されたとする。次に、例えば、ステップＳ４３において読み込まれた観測日における曜日の日常行動パターンの組み合わせを{Ａ}、{Ｂ}、{Ｃ}、{Ａ，Ｄ}、{Ａ，Ｂ，Ｃ，Ｄ}とする（現実のデータでは、これよりはるかに多い数となる。数十以上となる。）と、この導出された観測日における行動パターンと読み込まれた観測日における曜日の日常行動パターンの組み合わせとの類似度を順次計測する。
【００４８】
例えば、ここで、ｉ番目の観測日の曜日の日常行動パターンの組み合わせをＭ_ｉとすると、この例においては、ｉは１〜５となる。従って、ｉ番目の観測日の曜日の日常行動パターンの組み合わせＭ_ｉ及び観測日における行動パターンの組み合わせは以下のようになる。
Ｍ_１＝{Ａ}
Ｍ_２＝{Ｂ}
Ｍ_３＝{Ｃ}
Ｍ_４＝{Ａ，Ｄ}
Ｍ_５＝{Ａ，Ｂ，Ｃ，Ｄ}
また、観測日における行動パターンの組み合わせをＮとすると、Ｎ＝{Ａ，Ｂ，Ｃ}となる。観測日における行動パターンの組み合わせＮとｉ番目の観測日の曜日の日常行動パターンの組み合わせＭ_ｉの類似度をＲ_ｉとすると、Ｒ_ｉは、第２の関数である数式１により計測される。即ち、類似度Ｒ_５は、数式２により計測される。同様にして、類似度Ｒ_１〜Ｒ_４を計測すると、Ｒ_１＝０、Ｒ_２＝０、Ｒ_３＝０、Ｒ_４＝０となる。
【００４９】
データ処理部２６は、全ての観測日の曜日の日常行動パターンの組み合わせにおいて比較が行なわれたと判定された場合には（ステップＳ４６）、観測日の行動パターンが日常的であるか否かの判定を行なう（ステップＳ４７）。即ち、ステップＳ４５において計測された類似度の中で最も高い類似度を取得し、取得された類似度が閾値を超えているか否かの判定を行なう。類似度が閾値を超えていると判定された場合には、観測日の行動パターンが日常的であると判定する。即ち、上述の例によれば、ステップＳ４６において計測された類似度Ｒ_ｉは、Ｒ_１＝０、Ｒ_２＝０、Ｒ_３＝０、Ｒ_４＝０、Ｒ_５＝０．７５となり、最も高い類似度の値は、０．７５である。ここで、例えば、閾値を０．６と設定した場合に、最も高い類似度の値０．７５が閾値０．６を超えているので、観測日における行動パターンは日常的であると判定される。
【００５０】
一方、データ処理部２６は、取得された最も高い類似度が閾値を超えていないと判定された場合には、観測日の行動パターンが日常的でないと判定する。次に、端末１〜３は、ステップＳ４７における判定結果を表示部により出力する（ステップＳ４８）。即ち、ステップＳ４７において日常的であると判定された場合には、その旨の出力を行なう。一方、ステップＳ４７において日常的でないと判定された場合には、被験者に異変が生じていると判断し、異変である旨の出力を行なう。
【００５１】
この第２に実施の形態にかかる行動パターン異変監視システムによれば、所定の曜日の日常行動パターン群を導出して被験者の行動パターンの変化及び異変を把握する。従って、日常行動パターンが曜日毎に規則性のある人、例えば、毎週所定の曜日に外出する予定がある人等の高精度な日常行動パターンの組み合わせを導出することができる。従って、日常行動パターンが曜日毎に規則性のある人の行動パターンの変化及び異変を早期かつ確実に発見することができる。
【００５２】
次に、図６及び図７を参照して、この発明の第３の実施の形態にかかる行動パターン異変監視システムについて説明する。
【００５３】
この第３の実施の形態にかかる行動パターン異変監視システムは、第１の実施の形態のかかる行動パターン異変監視システムが過去数か月分の全ての生活行動データに基づいて日常行動パターンを導出するのに対し、過去数か月分の特定の時間帯の生活行動データに基づいて日常行動パターン群を導出する。従って、日常行動パターンが特定の時間帯毎に規則性のある人に適しているシステムである。
【００５４】
なお、この第３の実施の形態にかかる行動パターン異変監視システムは、図１に示す第１の実施の形態にかかる行動パターン異変監視システムと同一の構成である。この第３の実施の形態にかかる行動パターン異変監視システムにおいては、第１の実施の形態にかかる行動パターン異変監視システムにおいて、全ての生活行動データに基づく日常行動パターン群が日常行動パターン記憶部３０に記憶されるのに代えて、特定の時間帯における日常行動パターン群が日常行動パターン記憶部３０に記憶される。その他の点においては、第１の実施の形態にかかる行動パターン異変監視システムと同一である。
【００５５】
まず、住居システム１０において収集された生活行動データがネットワーク１４を介してセンタ内管理サーバ１２に送信され記憶される。なお、住居システム１０における動作は、第１の実施の形態のかかる住居システム１０における動作と同一であるため、詳細な説明を省略する。
【００５６】
次に、図６に示すフローチャートを参照して、行動パターン異変監視システムにおける特定の時間帯の日常行動パターン導出のための処理について説明する。データ処理部２６は、生活行動データ記憶部２８により記憶されている過去数か月分の特定の時間帯の被験者の移動場所及び時間を含む生活行動データを読み込む（ステップＳ５０）。次にステップＳ５０において読み込まれた過去数か月分の特定の時間帯のそれぞれの日の行動パターン記号列を導出する（ステップＳ５１）。ステップＳ５１において導出された過去数か月分の特定の時間帯のそれぞれの日の行動パターン記号列を行動パターン記号列記億部２９に記憶させる（ステップＳ５２）。なお、行動パターン記号列の導出は、第１の実施の形態にかかる行動パターン記号列の導出と同一のため、詳細な説明を省略する。
【００５７】
次に、データ処理部２６は、ステップＳ５２において記憶された過去数か月分の特定の時間帯の行動パターン記号列に基づいて、特定の時間帯の日常行動パターンを導出する（ステップＳ５３）。即ち、類似性の判定を行なうことができる関数（第１の関数）を用いて、ステップＳ５２において記憶された過去数か月分の特定の時間帯のそれぞれの日の行動パターン記号列の類似度を計測する。その計測された類似度に基づいて、特定の時間帯における日常行動パターン群を導出する。次に、ステップＳ５２において記憶された過去数か月分の特定の時間帯の行動パターン記号列とステップＳ５３において導出された特定の時間帯の日常行動パターン群に基づいて、過去数か月分のそれぞれの特定の時間帯に対応した日常行動パターンの組み合わせを導出する（ステップＳ５４）。
【００５８】
例えば、ステップＳ５３において１０種類の午前中における日常行動パターンＡ，Ｂ，Ｃ，Ｄ，Ｅ，Ｆ，Ｇ，Ｈ，Ｉ，Ｊが導出されたとする。次に、過去数ヶ月の中のそれぞれの日の午前中における行動パターンとステップＳ５３において導出された午前中における日常行動パターンＡ〜Ｊ、若しくは午前中における日常行動パターンＡ〜Ｊの組み合わせとを比較して、類似している午前中の日常行動パターンの組み合わせを導出する。即ち、過去数ヶ月の中のある日の午前中における行動パターンが午前中における日常行動パターンＡ及びＤの両方に似ている場合には、類似している午前中における日常行動パターンの組み合わせとして午前中における日常行動パターンの組み合わせ{Ａ，Ｄ}が導出され、過去数ヶ月の中のある日の午前中における行動パターンは午前中における日常行動パターンの組み合わせ{Ａ，Ｄ}に分類される。一方、過去数ヶ月の中の別のある日の午前中における行動パターンに類似している午前中の日常行動パターンの組み合わせとして午前中における日常行動パターンの組み合わせ{Ａ}が導出された場合に、過去数ヶ月の中の別の日の午前中における行動パターンは午前中の日常行動パターンの組み合わせ{Ａ}に分類される。ここで、午前中の日常行動パターンの組み合わせ{Ａ}は、午前中の日常行動パターンＡと同一であるが、午前中の日常行動パターンの組み合わせの１つであるとみなす。例えば、以下のように午前中における日常行動パターンの組み合わせを導出する。
１日目の午前中 {Ａ}
２日目の午前中 {Ａ，Ｄ}
３日目の午前中 {Ｂ}
４日目の午前中 {Ｃ}
５日目の午前中 {Ｄ，Ａ}
６日目の午前中 {Ａ，Ｂ，Ｃ，Ｄ}
７日目の午前中 {Ａ，Ｄ}
８日目の午前中 {Ｃ}
上述のステップＳ５４における導出結果は、過去数ヶ月の中における８日目の午前中までの日常行動パターンの組み合わせの一例である。上述のステップＳ５４における導出結果に基づいて、午前中における日常行動パターンの組み合わせを以下のように抽出することができる。
{Ａ}、{Ｂ}、{Ｃ}、{Ａ，Ｄ}、{Ａ，Ｂ，Ｃ，Ｄ}
なお、この実施の形態においては、ステップＳ５２において記憶された過去数か月分の特定の時間帯の行動パターンに基づいて、日常行動パターン群を導出し（ステップＳ５３）、ステップＳ５２において記憶された過去数か月分の特定の時間帯の行動パターン及びステップＳ５３において導出された特定の時間帯の日常行動パターン群に基づいて、特定の時間帯の日常行動パターンの組み合わせを導出する（ステップＳ５４）。しかしながら、ステップＳ５２において記憶された過去数か月分のそれぞれの日の特定の時間帯の行動パターンに基づいて、特定の時間帯の日常行動パターンを導出し、特定の時間帯の日常行動パターンの組み合わせを導出してもよい。即ち、ステップＳ５３における特定の時間帯の日常行動パターン群の導出とステップＳ５４における特定の時間帯の日常行動パターンの組み合わせの導出を同時に行ってもよい。同時に行う場合には、過去数ヶ月の中のある日の特定の時間帯の行動パターンと特定の時間帯の日常行動パターンとの比較を行う過程において特定の時間帯の日常行動パターン及び特定の時間帯の日常行動パターンの組み合わせが増加するので、上述の例と同一の条件下においては、以下のような特定の時間帯（午前中）の日常行動パターンの組み合わせとなる。
１日目の午前中 {Ａ}
２日目の午前中 {Ｂ}
３日目の午前中 {Ｃ}
４日目の午前中 {Ｄ}
５日目の午前中 {Ｂ}
６日目の午前中 {Ｂ，Ｃ，Ｄ}
７日目の午前中 {Ｂ}
８日目の午前中 {Ｄ}
次に、ステップＳ５３において導出された特定の時間帯の日常行動パターン群及びステップＳ５４において導出された特定の時間帯の日常行動パターンの組み合わせを日常行動パターン記憶部３０に記憶させる（ステップＳ５５）。
【００５９】
次に、図７に示すフローチャートを参照して、行動パターン異変監視システムにおける観測日の行動パターンに異変があるか否かの判定を行なうための処理について説明する。データ処理部２６は、ステップＳ６０〜ステップＳ６２において、図３におけるステップＳ２０〜ステップＳ２２と同様の処理を行い、観測日における行動パターン記号列を導出し（ステップＳ６０、Ｓ６１）、その導出された行動パターン記号列を行動パターン記号列２９に記憶させる（ステップＳ６２）。
【００６０】
次に、データ処理部２６は、日常行動パターン記憶部３０に記憶されている特定の時間帯の日常行動パターン群及び特定の時間帯の日常行動パターンの組み合わせを読み込む（ステップＳ６３）。ステップＳ６２において記憶された観測日における行動パターン記号列に基づいて、観測日における特定の時間帯の行動パターンの組み合わせを導出する（ステップＳ６４）。ステップＳ６４において導出された観測日における特定の時間帯の行動パターンの組み合わせとステップＳ６３で読み込まれた特定の時間帯の日常行動パターンの組み合わせの何れか１つとを、類似性の判定を行うことができる関数（第２の関数）を用いて比較し、類似度を計測する（ステップＳ６５）。
【００６１】
例えば、観測日における特定の時間帯の行動パターンの組み合わせとして{Ａ，Ｂ，Ｃ}が導出されたとする。次に、例えば、ステップＳ６３において読み込まれた特定の時間帯の日常行動パターンの組み合わせを{Ａ}、{Ｂ}、{Ｃ}、{Ａ，Ｄ}、{Ａ，Ｂ，Ｃ，Ｄ}とする（現実のデータでは、これよりはるかに多い数となる。数十以上となる。）と、この導出された観測日における特定の時間帯の行動パターンと読み込まれた特定の時間帯の日常行動パターンの組み合わせとの類似度を順次計測する。
【００６２】
例えば、ここで、ｉ番目の特定の時間帯の日常行動パターンの組み合わせをＭ_ｉとすると、この例においては、ｉは１〜５となる。従って、ｉ番目の特定の時間帯の日常行動パターンの組み合わせＭ_ｉ及び観測日における特定の時間帯の行動パターンの組み合わせは以下のようになる。
Ｍ_１＝{Ａ}
Ｍ_２＝{Ｂ}
Ｍ_３＝{Ｃ}
Ｍ_４＝{Ａ，Ｄ}
Ｍ_５＝{Ａ，Ｂ，Ｃ，Ｄ}
また、観測日における特定の時間帯の行動パターンの組み合わせをＮとすると、Ｎ＝{Ａ，Ｂ，Ｃ}となる。観測日における特定の時間帯の行動パターンの組み合わせＮとｉ番目の特定の時間帯の日常行動パターンの組み合わせＭ_ｉの類似度をＲ_ｉとすると、Ｒ_ｉは、第２の関数である数式１により計測される。即ち、類似度Ｒ_５は、数式２により計測される。同様にして、類似度Ｒ_１〜Ｒ_４を計測すると、Ｒ_１＝０、Ｒ_２＝０、Ｒ_３＝０、Ｒ_４＝０となる。
【００６３】
次に、データ処理部２６は、全ての特定の時間帯の日常行動パターンの組み合わせにおいて比較が行なわれたと判定された場合には（ステップＳ６６）、観測日における特定の時間帯の行動パターンが日常的であるか否かの判定を行なう（ステップＳ６７）。即ち、ステップＳ６５において計測された類似度の中で最も高い類似度を取得し、取得された類似度が閾値を超えているか否かの判定を行なう。取得された類似度が閾値を超えていると判定された場合には、観測日における特定の時間帯の行動パターンが日常的であると判定する。即ち、上述の例によれば、ステップＳ２６において計測された類似度Ｒ_ｉは、Ｒ_１＝０、Ｒ_２＝０、Ｒ_３＝０、Ｒ_４＝０、Ｒ_５＝０．７５となり、最も高い類似度の値は、０．７５である。ここで、例えば、閾値を０．６と設定した場合に、最も高い類似度の値０．７５が閾値０．６を超えているので、観測日における特定の時間帯の行動パターンは日常的であると判定される。
【００６４】
一方、データ処理部２６は、取得された最も高い類似度が閾値を超えていないと判定された場合には、観測日における特定の時間帯の行動パターンが日常的でないと判定する。次に、端末１〜３は、ステップＳ６７における判定結果を表示部により出力する（ステップＳ６８）。即ち、ステップＳ６７において日常的であると判定された場合には、その旨の出力を行なう。一方、ステップＳ６７において日常的でないと判定された場合には、被験者に異変が生じていると判断し、異変である旨の出力を行なう。
【００６５】
この第３に実施の形態にかかる行動パターン異変監視システムによれば、例えば午前、午後等の特定の時間帯の日常行動パターン群を導出して被験者の行動パターンの変化及び異変を把握する。従って、日常行動パターンが特定の時間帯毎に規則性のある人、例えば、午後に外出する日及び外出しない日がある人等の高精度な日常行動パターンを導出することができる。従って、日常行動パターンが特定の時間帯に規則性のある人の行動パターンの変化及び異変を早期かつ確実に発見することができる。
【００６６】
なお、上述の実施の形態において用いる第１の関数は、統計処理によるもの、マルコフモデル、若しくはニュートラルネットワークであってもよい。
【００６７】
また、上述の実施の形態においては、類似度を判定することができる関数を用いて日常行動パターン群を導出し、その導出された日常行動パターン群に基づいて導出される日常行動パターン組み合わせとの類似度を計測することにより、観測日の行動パターンが日常的であるか否かの判定を行なっている。しかしながら、生活環境の変化等に伴い日常行動パターンが以前と比較して相違する人等に使用する場合に、日常的であった行動パターンが非日常的となる若しくは非日常的であった行動パターンが日常的となる事態が生じるため、日常行動パターン群を補正できるようにしてもよい。即ち、日常行動パターンでないと判定された行動パターンの行動パターン記号列を含む過去数か月分のそれぞれの日の行動パターン記号列に基づいて日常行動パターン群を再導出することができるようにしてもよい。また、日常行動パターンであると判定された行動パターンを日常行動パターン群から除外することができるようにしてもよい。
【００６８】
【発明の効果】
この発明によれば、所定時間中に例えば便所等の特定の場所に滞在した場合には、その特定の場所を滞在場所とすることができ、所定時間中に特定の場所に滞在していない場合には、所定時間中に最も長く滞在した場所を滞在場所とすることができるため、細かい変化に対応した行動パターン記号列を導出することができる。また、その導出された行動パターン記号列に基づいて日常行動パターン群を導出し、その導出された日常行動パターン群に基づいて、多様な日常行動パターンを持っている人若しくは日常行動パターンを限定できない人等にも対応することができる高精度な日常行動パターンの組み合わせを導出することができる。従って、細かい変化に対応した観測日の行動パターン記号列と高精度な日常行動パターンの組み合わせとの類似度を計測することができるため、早期かつ確実に行動パターンの変化及び異変を発見することができる。
【００６９】
また、日常行動パターンが曜日毎に規則性のある人、例えば、毎週特定の曜日に規則正しく外出する用件がある人等に使用する場合に、曜日別に行動パターンを分析した曜日毎の日常行動パターン群を導出し、その導出された日常行動パターン群に基づいて日常行動パターンの組み合わせを導出することができる。また、日常行動パターンが例えば午前、午後等の特定の時間帯に規則性のある人等に使用する場合に、時間帯別に行動パターンを分析した時間帯毎の日常行動パターン群を導出し、その導出された日常行動パターン郡に基づいて日常行動パターンの組み合わせを導出することができる。従って、日常行動パターンが曜日毎に規則性を有する人、時間帯により規則性を有する人であっても、早期かつ確実に行動パターンの変化及び異変を発見することができる。
【００７０】
また、生活環境の変化に伴い日常行動パターンが以前と比較して相違する人等に使用する場合に、日常的でないと判定された行動パターンの行動パターン記号列を含む行動パターン記号列に基づいて日常行動パターン群を導出することができ、日常的であると判定された行動パターンを日常行動パターン群から除外することができるため、行動パターンが以前と比較して相違している人であっても、早期かつ確実に行動パターンの変化及び異変を発見することができる。
【図面の簡単な説明】
【図１】この発明の第１の実施の形態にかかる行動パターン異変監視システムのブロック構成図である。
【図２】この発明の第１の実施の形態にかかる行動パターン異変監視システムにおける日常行動パターン導出のための処理を示すフローチャートである。
【図３】この発明の第１の実施の形態にかかる行動パターン異変監視システムにおける観測日の行動パターンに異変があるか否かの判定を行なうための処理を示すフローチャートである。
【図４】この発明の第２の実施の形態にかかる行動パターン異変監視システムにおける所定の曜日の日常行動パターン導出のための処理を行なうフローチャートである。
【図５】この発明の第２の実施の形態にかかる行動パターン異変監視システムにおける観測日の行動パターンに異変があるか否かの判定を行なうための処理を示すフローチャートである。
【図６】この発明の第３の実施の形態にかかる行動パターン異変監視システムにおける特定の時間帯の日常行動パターン導出のための処理を行なうフローチャートである。
【図７】この発明の第３の実施の形態にかかる行動パターン異変監視システムにおける観測日の行動パターンに異変があるか否かの判定を行なうための処理を示すフローチャートである。
【符号の説明】
１０…住居システム、１２…センタ内管理サーバ、１４、１６…ネットワーク、１８…住居内管理サーバ、２０…データ処理部、２２…生活行動データ記憶部、２４…通信制御部、２６…データ処理部、２８…生活行動データ記憶部、２９…行動パターン部記号列記憶部、３０…日常行動パターン記憶部、３２…通信制御部[0001]
TECHNICAL FIELD OF THE INVENTION
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a behavior pattern change monitoring system for early detecting a change in a behavior pattern based on data of a place and a time of stay of a person in daily life.
[0002]
[Prior art]
As the number of elderly people living alone increases, early detection of mental and physical changes due to such senile depression, suicide, chronic disease, sudden illness, etc., of such elderly people, and such elderly people maintain a safe and healthy life There is a need for a system for this.
[0003]
Conventionally, in order to detect an abnormal situation of a solitary elderly person at an early stage, a system that installs a person detection sensor in the elderly person's residence and outputs an abnormal signal when staying in a place where abnormalities easily occur for a long time has been developed. (For example, see Patent Document 1). In addition, even if the elderly living alone are not in a place where accidents are likely to occur for a long period of time, a thermal sensor is installed at the required place in the residence of the elderly living alone to detect an abnormal situation as early as possible. Based on the data of the place of stay and the time of stay of the elderly detected by the above, calculate an index that is the daily life behavior of this elderly, and compare the calculated index with the index of the current behavior, There is also a system for determining whether or not there is (for example, see Patent Document 2).
[0004]
[Patent Document 1]
JP-A-4-22334
[Patent Document 2]
JP-A-2000-90366
[0005]
[Problems to be solved by the invention]
However, in the device described in Patent Literature 1, since the staying time is determined only in a place where the accident is likely to occur, and the occurrence is determined, the extraordinary behavior in a place other than the place where the occurrence tends to occur cannot be sensed. .
[0006]
Further, in the system described in Patent Literature 2, the pattern of daily activities used for comparison of similarity is limited to one. It is not possible to discover small changes and abnormalities that cannot be limited.
[0007]
An object of the present invention is to provide a behavior pattern change monitoring system that can quickly and reliably detect a change in a person.
[0008]
[Means for Solving the Problems]
2. The behavior pattern abnormality monitoring system according to claim 1, wherein the living activity data storage means stores living activity data including a moving place and a time of a person detected based on a reaction of a human sensor for a plurality of days, and said living activity data. With respect to living activity data for a plurality of days stored by the data storage means, an action pattern symbol string deriving means for deriving an action pattern symbol string for each day, and an action pattern symbol derived by the action pattern symbol string deriving means. A daily action pattern deriving means for deriving a daily action pattern group based on the sequence; and a daily action pattern combination deriving means for deriving a combination of daily action patterns based on the daily action pattern group derived by the daily action pattern deriving means. , The action pattern symbol of the observation day derived by the action pattern symbol string deriving means And a comparing means for comparing the combination of the daily action patterns derived by the daily action pattern combination deriving means, and whether or not the action pattern on the observation day is a daily action pattern based on the comparison result by the comparing means. It is characterized by comprising a determining means for performing a determination, and an output means for outputting a determination result by the determining means.
[0009]
The behavior pattern abnormal monitoring system according to claim 2 determines whether the behavior pattern symbol string deriving means stays in a specific place during a predetermined time, and determines that the behavior pattern symbol string deriving means stays in the specific place. In this case, the specific place is set as a stay place during a predetermined time, and if it is determined that the user has not stayed at the specific place, the place where the user has stayed the longest during the predetermined time is set as a stay place. And
[0010]
According to the behavior pattern change monitoring system according to the first and second aspects, when the user stays at a specific place such as a toilet during a predetermined time, the specific place is set as a stay place, and during the predetermined time, If the user does not stay at a specific place, the place where the user has stayed the longest during the predetermined time is set as the stay place, so that an action pattern symbol string corresponding to a small change can be derived. Further, a daily activity pattern group including a plurality of daily activity patterns is derived based on the derived activity pattern symbol string, and further, a combination of daily activity patterns is derived based on the derived daily activity pattern group. Therefore, it is possible to quickly and surely detect a change and an unusual change in a behavior pattern of a person having various daily behavior patterns or a person who cannot limit the daily behavior patterns.
[0011]
The behavior pattern change monitoring system according to claim 3, wherein the daily activity pattern deriving unit is configured to determine a daily activity pattern of a predetermined day of the week based on an activity pattern symbol string of a predetermined day of the week derived by the activity pattern symbol string deriving unit. It is characterized in that a group is derived.
[0012]
According to the behavior pattern change monitoring system of the third aspect, the behavior pattern is analyzed for each day of the week and a daily behavior pattern group for each day of the week is derived. Therefore, when a person with a regular daily activity pattern is used for each day of the week, for example, a person who regularly goes out for each day of the week or a person who has a specific schedule for each day of the week, such a person is used. Changes and abnormalities in behavior patterns can be detected early and reliably.
[0013]
The behavior pattern change monitoring system according to claim 4, wherein the daily activity pattern deriving unit is configured to perform daily activities in a specific time zone based on the activity pattern symbol sequence in a specific time zone derived by the activity pattern symbol sequence deriving unit. It is characterized in that an action pattern group is derived.
[0014]
According to the behavior pattern change monitoring system of the fourth aspect, the behavior pattern is analyzed for each time zone, and a daily activity pattern group for each time zone is derived. Therefore, when a daily activity pattern is used for a person having regularity in a specific time zone such as, for example, in the morning or afternoon, it is possible to quickly and surely find a change and an unusual change in the behavior pattern of such a person. .
[0015]
Further, the behavior pattern abnormal monitoring system according to claim 5, wherein the daily activity pattern deriving unit performs daily routine based on an activity pattern symbol sequence including an activity pattern symbol sequence of an activity pattern determined to be not a daily activity pattern by the determination unit. It is characterized in that an action pattern group is derived.
[0016]
Further, the behavior pattern change monitoring system according to claim 6 is characterized in that at least one of the behavior patterns derived as daily behavior patterns by the daily behavior pattern deriving means is excluded.
[0017]
According to the behavior pattern change monitoring system according to the fifth and sixth aspects, it is possible to correct the daily behavior pattern group. Therefore, when used for a person or the like who has a different daily activity pattern compared to the past due to a change in the living environment, it is possible to derive a highly accurate daily activity pattern group, and In addition, it is possible to quickly and surely detect a change and an abnormality in the behavior pattern.
[0018]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, a first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the behavior pattern change monitoring system according to the first embodiment of the present invention. As shown in this figure, the behavior pattern change monitoring system includes a residence system 10 installed in the subject's residence, an in-center management server that determines whether the life behavior pattern of the subject on the observation day is daily or not. 12. Terminals 1 to 3 for displaying the determination results determined by the in-center management server 12 are provided. The residential system 10 and the in-center management server 12 are connected via a network 14, and the in-center management server 12 and terminals 1 to 3 are connected via a network 16. The dwelling system 10 includes motion sensors 1 to 5 for detecting whether or not the subject is staying at the place. Here, the human sensor 1 is installed in the entrance, the human sensor 2 is installed in the kitchen, the human sensor 3 is installed in the living room, the human sensor 4 is installed in the bedroom, and the human sensor 5 is installed in the toilet.
[0019]
Furthermore, the dwelling system 10 includes a dwelling management server 18 that stores living activity data including the moving place and time of the subject and controls data communication. The in-house management server 18 includes a data processing unit 20, a living activity data storage unit 22, and a communication control unit 24. The data processing unit 20 is connected to a living activity data storage unit 22 that stores living activity data including the moving place and time of the subject transmitted from the human sensors 1 to 5. The data processing unit 20 is connected to a communication control unit 24 that controls communication with the in-center management server 12.
[0020]
The in-center management server 12 includes a data processing unit 26 that derives a daily activity pattern based on the life activity pattern of the subject and determines whether the life activity pattern on the observation day is daily. The data processing unit 26 includes a living activity data storage unit 28 in which living activity data transmitted by the dwelling system 10 is stored, an activity pattern symbol string storage unit 29 in which an activity pattern symbol string is stored, and a daily activity pattern. Are connected, and a communication control unit 32 that performs communication control between the house system 10 and the terminals 1 to 3 is connected. The terminals 1 to 3 can be installed in the residence of the subject's family, the residence of the subject's doctor, or the residence of the caregiver of the subject.
[0021]
Next, with reference to the flowcharts shown in FIG. 2 and FIG. 3, an abnormal change monitoring process in the behavior pattern abnormal change monitoring system according to the first embodiment will be described.
[0022]
First, living activity data collected in the dwelling system 10 before the processing shown in the flowcharts of FIGS. 2 and 3 is transmitted to the in-center management server 12 via the network 14 and stored. That is, the data processing unit 20 provided in the in-house management server 18 determines whether a signal is received from any one of the human sensors 1 to 5. When a signal is received from any one of the motion sensors 1 to 5, it is determined that the subject is staying in the dwelling, and the signal transmitted from the motion sensors 1 to 5 is determined. Based on this, it is determined whether or not the subject has stayed at the same place for 30 seconds or more. If it is determined that the subject has stayed at the same place for 30 seconds or more, the subject is considered to have stayed at that place. Then, the place where the user was staying immediately before moving to the place, and the time of moving from the place where he was staying immediately before moving to the place to the current place are stored in the living activity data storage unit 22 as living activity data. Let it. On the other hand, if it is determined that the subject has not stayed at the same place for 30 seconds or more, or that the subject has not gone out for 30 seconds or more, it is determined that the subject has not stayed at that place. That is, it is assumed that the user stays at the place where he was staying immediately before moving to the place.
[0023]
On the other hand, when the data processing unit 20 does not receive a signal from any of the human sensors 1 to 5, it does not receive a signal from any of the human sensors 1 to 5. It is determined whether or not the place where the user was staying immediately before the state was the entrance. If it is determined that the place where you were staying just before the state where no signal is received from any of the motion sensors is the entrance, the state where no signal is received from any of the motion sensors continues for 30 seconds or more. It is determined whether or not it is performed. If it is determined that the state in which no signal is received from any of the motion sensors has continued for 30 seconds or more, it is determined that the subject is out. Then, the living activity data storage unit 22 stores the place where the user stayed just before going out and the time when he / she went out of the place where the user stayed just before going out as living activity data. If it is determined that the state in which no signal is received from any of the human sensors does not continue for 30 seconds or more, it is determined that the subject is not out of the office. That is, it is considered that the user stays at the place where he / she was staying immediately before receiving a signal from any of the motion sensors. In this way, the living activity data including the moving place and time of the subject is stored in the living activity data storage unit 22, and the living activity data for one day is acquired.
[0024]
The data processing unit 20 stores the daily activity data for one day stored in the daily activity data storage unit 22 once a day at a predetermined time via the communication control unit 24 and the network 14 through the in-center management server. 12 is transmitted. The data processing unit 26 provided in the in-center management server 12 receives the one-day living activity data transmitted from the in-house management server 18 via the network 14 and stores the data in the living activity data storage unit 28. .
[0025]
Next, a process for deriving a daily behavior pattern in the behavior pattern abnormal monitoring system will be described with reference to a flowchart shown in FIG. The data processing unit 26 provided in the in-center management server 22 reads the living activity data including the moving place and time of the subject for the past months stored in the living activity data storage unit 28 (step S10). .
[0026]
Next, an action pattern symbol string for each day of the past months is derived based on the life activity data of the past months read in step S10 (step S11). First, a day is divided into 10-minute intervals, that is, 144 time strings are created. Next, during the first 10 minutes of the day, it is determined whether or not the user has stayed at a specific place such as a toilet. If it is determined that the user has stayed at a specific place, the specific place is set as the place to stay for the first 10 minutes. On the other hand, when it is determined that the user has not stayed at the specific place, the place where the user has stayed for the first ten minutes for the longest time is set as the stay place. Next, in the same way as the above-described stay place deriving method, the stay place for the next 10 minutes is derived, and the operation is repeated until the stay place for the last 10 minutes of the day is derived. Then, 144 symbol strings indicating staying places every 10 minutes in a day are defined as action pattern symbol strings. In the same manner as described above, the action pattern symbol strings for the past several months are derived, and the derived action pattern symbol strings for the respective past months are stored in the action pattern symbol storage unit 29. (Step S12).
[0027]
Next, the data processing unit 26 derives a daily action pattern based on the action pattern symbol strings for the past several months stored in step S12 (step S13). That is, the similarity of the action pattern symbol string of each day for the past several months stored in step S12 is measured using a function (first function) capable of determining similarity. Based on the measured similarity, several types of daily action patterns (daily action pattern groups) are derived. Next, based on the action pattern symbol string for the past several months stored in step S12 and the daily action pattern group derived in step S13, the daily action pattern corresponding to each day for the past several months is obtained. A combination is derived (step S14).
[0028]
For example, it is assumed that ten types of daily action patterns A, B, C, D, E, F, G, H, I, and J are derived in step S13. Next, by comparing the activity pattern of each day in the past months with the daily activity patterns A to J or the combination of the daily activity patterns A to J derived in step S13, similar daily activities are calculated. A combination of action patterns is derived. That is, when the behavior pattern of a certain day in the past several months is similar to both the daily behavior patterns A and D, the combination of the daily behavior patterns {A, D} is regarded as a combination of similar daily behavior patterns. Is derived, and the behavior pattern of a certain day in the past several months is classified into a combination of daily behavior patterns {A, D}. On the other hand, when the combination {A} of daily activity patterns is derived as a combination of daily activity patterns similar to the activity pattern of another day in the past months, another day in the past months Is classified into a combination {A} of everyday behavior patterns. Here, the combination {A} of the daily action patterns is the same as the daily action pattern A, but is regarded as one of the combinations of the daily action patterns. For example, a combination of daily action patterns is derived as follows.
Day 1 {A}
Day 2 {A, D}
Day 3 {B}
Day 4 {C}
Day 5 {D, A}
Day 6 {A, B, C, D}
Day 7 {A, D}
Day 8 {C}
The above-described derivation result in step S14 is an example of a combination of daily action patterns up to the eighth day in the past several months. Based on the derivation result in step S14 described above, a combination of daily activity patterns can be extracted as follows.
{A}, {B}, {C}, {A, D}, {A, B, C, D}
In this embodiment, a daily activity pattern group is derived based on the past several months of action patterns stored in step S12 (step S13), and the daily activity patterns are stored for the past several months stored in step S12. Based on the behavior pattern and the daily behavior pattern group derived in step S13, a combination of daily behavior patterns is derived (step S14). However, a daily activity pattern may be derived based on the activity patterns of each of the past several months stored in step S12, and a combination of the daily activity patterns may be derived. That is, the derivation of the group of daily action patterns in step S13 and the derivation of the combination of daily action patterns in step S14 may be performed simultaneously. When performed simultaneously, the number of combinations of daily action patterns and daily action patterns increases in the process of comparing the action patterns of a certain day in the past several months with the daily action patterns. Is a combination of the following daily behavior patterns.
Day 1 {A}
Day 2 {B}
Day 3 {C}
Day 4 {D}
Day 5 {B}
Day 6 {B, C, D}
Day 7 {B}
Day 8 {D}
Next, a combination of the daily action pattern group derived in step S13 and the daily action pattern derived in step S14 is stored in the daily action pattern storage unit 30 (step S15).
[0029]
Next, with reference to the flowchart shown in FIG. 3, a process for determining whether or not the behavior pattern on the observation day has an abnormality in the behavior pattern abnormality monitoring system will be described. First, the data processing unit 26 reads the living activity data including the moving location and time of the subject on the observation date stored in the living activity data storage unit 28 (Step S20). Next, the same processing as in step S11 in FIG. 2 is performed on the living activity data on the observation day read in step S20 to derive an action pattern symbol string on the observation day (step S21). The action pattern symbol string on the observed date is stored in the action pattern symbol string 29 (step S22).
[0030]
Next, the data processing unit 26 reads the combination of the daily activity pattern group and the daily activity pattern stored in the daily activity pattern storage unit 30 (Step S23). Based on the action pattern symbol string on the observation day stored in step S22, a combination of action patterns on the observation day is derived (step S24). A function (second function) capable of determining the similarity between the combination of the behavior patterns on the observation day derived in step S24 and the combination of the daily behavior patterns read in step S23 is used. Then, the similarity is measured (step S25).
[0031]
For example, suppose that {A, B, C} is derived as a combination of action patterns on the observation day. Next, for example, the combination of the daily action patterns read in step S23 is {A}, {B}, {C}, {A, D}, {A, B, C, D} (actual data In this case, the number becomes much larger than this, and becomes several tens or more.), And the similarity between the derived behavior pattern on the observation day and the read combination of the daily behavior pattern is sequentially measured.
[0032]
For example, here, the combination of the i-th daily activity pattern is M _i Then, in this example, i is 1 to 5. Therefore, the combination M of the i-th daily action pattern _i And the combination of the action patterns on the observation day is as follows.
M ₁ = {A}
M ₂ = {B}
M ₃ = {C}
M ₄ = {A, D}
M ₅ = {A, B, C, D}
Further, assuming that the combination of the behavior patterns on the observation day is N, N = {A, B, C}. Combination of action pattern N on observation day and combination M of i-th daily action pattern _i The similarity of R _i Then R _i Is measured by Equation 1 which is a second function.
[0033]
(Equation 1)

That is, the similarity R ₅ Is measured by Expression 2.
[0034]
(Equation 2)

Similarly, the similarity R ₁ ~ R ₄ Is measured, R ₁ = 0, R ₂ = 0, R ₃ = 0, R ₄ = 0.
[0035]
When it is determined that the comparison has been performed for all the combinations of the daily action patterns (step S26), the data processing unit 26 determines whether the action pattern on the observation day is daily (step S27). ). That is, the value of the highest similarity among the combinations of the daily action patterns measured in step S25 is extracted. Then, it is determined whether or not the value of the extracted similarity exceeds a threshold. When it is determined that the value of the similarity exceeds the threshold, it is determined that the behavior pattern on the observation day has a similarity to any one of the combinations of the daily behavior patterns in the combination of the daily behavior patterns. Is determined to be the target. That is, according to the above-described example, the similarity R measured in step S26 _i Is R ₁ = 0, R ₂ = 0, R ₃ = 0, R ₄ = 0, R ₅ = 0.75, and the highest similarity value is 0.75. Here, for example, when the threshold value is set to 0.6, since the highest similarity value 0.75 exceeds the threshold value 0.6, the behavior pattern on the observation day is determined to be everyday. .
[0036]
On the other hand, if the data processing unit 26 determines that the value of the similarity to any of the combinations of the daily action patterns does not exceed the threshold, the data processing unit 26 determines that the action pattern on the observation day is similar to any of the combinations of the daily action patterns. It is determined that there is no sex, and it is determined that it is not everyday. Next, the determination result in step S27 is transmitted to terminals 1 to 3 via communication control unit 32 and network 16. Then, the terminals 1 to 3 output the determination result by display or the like (step S28). That is, if it is determined in step S27 that it is everyday, an output to that effect is output. On the other hand, if it is determined in step S27 that the subject is not ordinary, it is determined that the subject has an abnormality, and an output indicating that the subject is abnormal is output.
[0037]
According to the behavior pattern change monitoring system according to the first embodiment, when the user stays at a specific place such as a toilet during a predetermined time, the specific place is set as a staying place and the specific If the user has not stayed at the place, the action pattern symbol string is derived with the place where the user stayed the longest during the predetermined time as the stay place. Therefore, since an action pattern symbol string corresponding to a small change can be derived, it is possible to reliably grasp an action pattern including a movement to a place where an abnormal change is likely to occur, such as a toilet. In addition, a group of daily behavior patterns is derived based on the derived behavior pattern symbol string, and a person having various daily behavior patterns or a person who cannot limit the daily behavior patterns based on the derived daily behavior pattern group. It is possible to derive a highly accurate combination of daily action patterns that can cope with the above. Therefore, since it is possible to measure the similarity between the action pattern symbol string and the combination of the highly accurate daily action pattern on the observation day corresponding to the fine change, it is possible to quickly and reliably discover the change and the change in the action pattern at an early stage. it can.
[0038]
Next, a behavior pattern change monitoring system according to a second embodiment of the present invention will be described with reference to FIGS.
[0039]
In the behavior pattern change monitoring system according to the second embodiment, the behavior pattern change monitoring system according to the first embodiment derives a daily action pattern based on all living activity data for the past several months. On the other hand, a daily activity pattern group is derived based on living activity data on a predetermined day of the week for the past several months. Therefore, it is a system that is desirable to be used for a person whose daily behavior pattern is regular for each day of the week.
[0040]
The behavior pattern change monitoring system according to the second embodiment has the same configuration as the behavior pattern change monitoring system according to the first embodiment shown in FIG. In the behavior pattern change monitoring system according to the second embodiment, in the behavior pattern change monitoring system according to the first embodiment, a daily action pattern group based on all living action data is stored in the daily action pattern storage unit 30. Is stored in the daily activity pattern storage unit 30 on a predetermined day of the week. The other points are the same as those of the behavior pattern change monitoring system according to the first embodiment.
[0041]
First, living activity data collected in the residence system 10 is transmitted to the in-center management server 12 via the network 14 and stored. Note that the operation in the house system 10 is the same as the operation in the house system 10 according to the first embodiment, and thus a detailed description is omitted.
[0042]
Next, a process for deriving a daily activity pattern on a predetermined day of the week in the behavior pattern abnormal monitoring system will be described with reference to a flowchart shown in FIG. The data processing unit 26 reads the living activity data including the moving place and time of the subject on the predetermined days of the past several months stored in the living activity data storage unit 28 (step S30). Next, an action pattern symbol string read for each of the predetermined days of the past several months read in step S30 is derived (step S31). The action pattern symbol string stored on the action pattern symbol string storage unit 29 is stored in the action pattern symbol string storage unit 29 (step S32). The derivation of the action pattern symbol string is the same as the derivation of the action pattern symbol string according to the first embodiment, and a detailed description will be omitted.
[0043]
Next, the data processing unit 26 derives a daily activity pattern for a predetermined day of the week based on the action pattern symbol string for a predetermined day of the past several months stored in step S32 (step S33). That is, using the function (first function) capable of determining the similarity, the similarity of the action pattern symbol string stored in step S32 for each of the predetermined days of the past several months is calculated. measure. Based on the measured similarity, several types of daily activity patterns (daily activity pattern groups) on a predetermined day of the week are derived. Next, based on the action pattern symbol string of the predetermined days of the past several months stored in step S32 and the daily action pattern group of the predetermined days of the day derived in step S33, the predetermined number of the past months is determined. A combination of daily action patterns on a predetermined day of the week corresponding to each day of the week is derived (step S34).
[0044]
For example, it is assumed that ten types of daily activity patterns A, B, C, D, E, F, G, H, I, and J on Monday are derived in step S33. Next, the action pattern of each day of Monday in the past months is compared with the daily action patterns A to J on Monday derived in step S13 or a combination of the daily action patterns A to J on Monday, A combination of similar Monday's daily action patterns is derived. That is, when the behavior pattern on a certain Monday in the past months is similar to both the daily behavior patterns A and D on Monday, the combination of the similar daily behavior patterns on Monday is used as the combination of the similar daily behavior patterns on Monday. The combination {A, D} is derived, and the behavior pattern on a certain Monday in the past several months is classified into the combination {A, D} of everyday behavior patterns on Monday. On the other hand, if a combination {A} of Monday's daily action patterns is derived as a combination of Monday's daily action patterns similar to another Monday's action pattern in the past months, Is classified as a combination {A} of Monday's daily action patterns. Here, the combination of Monday's daily action patterns {A} is the same as Monday's daily action pattern A, but is regarded as one of the combinations of Monday's daily action patterns. For example, a combination of daily activity patterns on Monday is derived as follows.
First Monday {A}
Second Monday {A, D}
Third Monday {B}
4th Monday {C}
Fifth Monday {D, A}
6th Monday {A, B, C, D}
7th Monday {A, D}
8th Monday {C}
The above-described derivation result in step S34 is an example of a combination of daily activity patterns up to the eighth Monday in the past several months. Based on the derivation result in step S34, a combination of Monday's daily activity patterns can be extracted as follows.
{A}, {B}, {C}, {A, D}, {A, B, C, D}
In this embodiment, a daily activity pattern group on a predetermined day is derived based on the activity patterns on a predetermined day of the past several months stored in step S32 (step S33), and in step S32, Based on the stored action patterns on the predetermined days of the past several months and the daily action patterns on the predetermined days derived in step S33, a combination of the daily action patterns on the predetermined days is derived (step S34). . However, based on the behavior patterns of each of the predetermined days of the past several months stored in step S32, the daily activity patterns of the predetermined days are derived, and the combination of the daily behavior patterns of the predetermined days is derived. May be. That is, the derivation of the group of daily activity patterns on the predetermined day of the week in step S33 and the derivation of the combination of daily activity patterns on the predetermined day of the week in step S34 may be performed simultaneously. In the case of performing simultaneously, in the process of comparing the behavior pattern on a certain day of the week and the daily activity pattern on the predetermined day in the past months, the daily behavior pattern on the predetermined day and the daily action pattern on the predetermined day Since the number of combinations increases, under the same conditions as in the above-described example, the following combinations of daily activity patterns on a predetermined day of the week are obtained.
First Monday {A}
Second Monday {B}
Third Monday {C}
4th Monday {D}
Fifth Monday {B}
6th Monday {B, C, D}
7th Monday {B}
8th Monday {D}
Next, the daily activity pattern storage unit 30 stores the combination of the daily activity pattern group of the predetermined day of the week derived in step S33 and the daily activity pattern of the predetermined day of the week derived in step S34 (step S35).
[0045]
Next, with reference to a flowchart shown in FIG. 5, a process for determining whether or not there is an abnormality in the behavior pattern on the observation day in the behavior pattern abnormality monitoring system will be described. In steps S40 to S42, the data processing unit 26 performs the same processing as in steps S20 to S22 in FIG. 3 to derive an action pattern symbol string on the observation date (steps S40 and S41), and the derived action The pattern symbol string is stored in the action pattern symbol string 29 (step S42).
[0046]
Next, the data processing unit 26 reads the combination of the daily activity pattern group on the observation day and the daily activity pattern on the observation day stored in the daily activity pattern storage unit 30 (step S43). Based on the action pattern symbol string on the observation day stored in step S42, a combination of action patterns on the observation day is derived (step S44). A function (second function) capable of determining the similarity between one of the combination of the behavior pattern on the observation day derived in step S44 and the combination of the daily activity pattern on the day of the week on the observation day read in step S43. And the similarity is measured (step S45).
[0047]
For example, suppose that {A, B, C} is derived as a combination of action patterns on the observation day. Next, for example, the combination of the daily action patterns on the day of the week on the observation day read in step S43 is {A}, {B}, {C}, {A, D}, {A, B, C, D}. (In actual data, the number is much larger than this. It becomes several tens or more.), And the combination of the activity pattern on the derived observation day and the daily activity pattern on the day of the week on the read observation day is The similarity is measured sequentially.
[0048]
For example, here, the combination of daily activity patterns on the day of the i-th observation day is M _i Then, in this example, i is 1 to 5. Therefore, the combination M of daily activity patterns on the day of the i-th observation day _i And the combination of the action patterns on the observation day is as follows.
M ₁ = {A}
M ₂ = {B}
M ₃ = {C}
M ₄ = {A, D}
M ₅ = {A, B, C, D}
Further, assuming that the combination of the behavior patterns on the observation day is N, N = {A, B, C}. Combination N of action pattern on observation day and combination M of daily action pattern on day of i-th observation day _i The similarity of R _i Then R _i Is measured by Equation 1 which is a second function. That is, the similarity R ₅ Is measured by Expression 2. Similarly, the similarity R ₁ ~ R ₄ Is measured, R ₁ = 0, R ₂ = 0, R ₃ = 0, R ₄ = 0.
[0049]
When it is determined that the comparison has been performed for all combinations of the daily activity patterns on the day of the week on all observation days (step S46), the data processing unit 26 determines whether the behavior pattern on the observation day is daily. Is performed (step S47). That is, the highest similarity among the similarities measured in step S45 is acquired, and it is determined whether or not the acquired similarity exceeds a threshold. When it is determined that the similarity exceeds the threshold, the behavior pattern on the observation day is determined to be daily. That is, according to the above example, the similarity R measured in step S46 _i Is R ₁ = 0, R ₂ = 0, R ₃ = 0, R ₄ = 0, R ₅ = 0.75, and the highest similarity value is 0.75. Here, for example, when the threshold value is set to 0.6, since the highest similarity value 0.75 exceeds the threshold value 0.6, it is determined that the behavior pattern on the observation day is everyday. .
[0050]
On the other hand, when it is determined that the obtained highest similarity does not exceed the threshold, the data processing unit 26 determines that the behavior pattern on the observation day is not ordinary. Next, the terminals 1 to 3 output the determination result in step S47 on the display unit (step S48). That is, if it is determined in step S47 that it is everyday, an output to that effect is output. On the other hand, if it is determined in step S47 that the subject is not ordinary, it is determined that the subject has changed, and an output indicating that the change has occurred is output.
[0051]
According to the behavior pattern abnormality monitoring system according to the second embodiment, a daily behavior pattern group on a predetermined day of the week is derived to grasp a change and an abnormality in the behavior pattern of the subject. Therefore, it is possible to derive a highly accurate combination of daily action patterns of a person having a regular daily action pattern for each day of the week, for example, a person who is scheduled to go out on a predetermined day of the week. Therefore, it is possible to quickly and surely detect a change and an unusual change in the behavior pattern of a person whose daily behavior pattern is regular for each day of the week.
[0052]
Next, a behavior pattern change monitoring system according to a third embodiment of the present invention will be described with reference to FIGS.
[0053]
In the behavior pattern change monitoring system according to the third embodiment, the behavior pattern change monitoring system according to the first embodiment derives a daily action pattern based on all living activity data for the past several months. On the other hand, a daily activity pattern group is derived based on living activity data in a specific time zone for the past several months. Therefore, the system is suitable for a person whose daily activity pattern is regular for each specific time zone.
[0054]
The behavior pattern change monitoring system according to the third embodiment has the same configuration as the behavior pattern change monitoring system according to the first embodiment shown in FIG. In the behavior pattern change monitoring system according to the third embodiment, in the behavior pattern change monitoring system according to the first embodiment, a daily action pattern group based on all living action data is stored in the daily action pattern storage unit 30. Is stored in the daily-action-pattern storage unit 30 in a specific time slot. The other points are the same as those of the behavior pattern change monitoring system according to the first embodiment.
[0055]
First, living activity data collected in the residence system 10 is transmitted to the in-center management server 12 via the network 14 and stored. Note that the operation in the house system 10 is the same as the operation in the house system 10 according to the first embodiment, and thus a detailed description is omitted.
[0056]
Next, with reference to the flowchart shown in FIG. 6, a process for deriving a daily activity pattern in a specific time zone in the behavior pattern abnormal monitoring system will be described. The data processing unit 26 reads the living activity data including the moving place and time of the subject in a specific time zone for the past several months stored in the living activity data storage unit 28 (step S50). Next, an action pattern symbol string read in step S50 for each day of a specific time zone for the past several months is derived (step S51). The action pattern symbol string stored in the action pattern symbol string storage unit 29 is stored in the action pattern symbol string storage unit 29 (step S52). The derivation of the action pattern symbol string is the same as the derivation of the action pattern symbol string according to the first embodiment, and a detailed description will be omitted.
[0057]
Next, the data processing unit 26 derives a daily action pattern in a specific time zone based on the action pattern symbol string in a specific time zone for the past several months stored in step S52 (step S53). That is, using a function (first function) capable of determining similarity, the degree of similarity of the action pattern symbol string of each day in a specific time zone of the past several months stored in step S52. Is measured. Based on the measured similarity, a daily activity pattern group in a specific time zone is derived. Next, based on the action pattern symbol string of the specific time zone for the past several months stored in step S52 and the daily action pattern group of the specific time zone derived in step S53, the past several months are obtained. A combination of daily action patterns corresponding to each specific time zone is derived (step S54).
[0058]
For example, it is assumed that ten types of daily action patterns A, B, C, D, E, F, G, H, I, and J in the morning are derived in step S53. Next, the action pattern in the morning of each day in the past months is compared with the daily action patterns A to J in the morning derived in step S53 or the combination of the daily action patterns A to J in the morning. Then, a combination of similar morning daily action patterns is derived. That is, if the behavior pattern in the morning of a certain day in the past several months is similar to both the daily activity patterns A and D in the morning, the combination of similar daily activity patterns in the morning The combination {A, D} of daily activity patterns in the morning is derived, and the activity pattern in the morning of a certain day in the past several months is classified into the combination {A, D} of daily activity patterns in the morning. On the other hand, when a combination {A} of the daily activity pattern in the morning is derived as a combination of the daily activity pattern in the morning similar to the activity pattern in the morning of another day in the past several months, The behavior pattern in the morning on another day in the past months is classified into the combination {A} of the daily behavior pattern in the morning. Here, the combination {A} of the morning daily activity patterns is the same as the morning daily activity pattern A, but is regarded as one of the combinations of the morning everyday activity patterns. For example, a combination of daily activity patterns in the morning is derived as follows.
Morning of the first day {A}
Morning of the second day {A, D}
Morning of the third day {B}
4th morning {C}
Morning of the fifth day {D, A}
Morning of the sixth day {A, B, C, D}
Morning of the 7th day {A, D}
Morning of the 8th day {C}
The above-described derivation result in step S54 is an example of a combination of everyday action patterns until the morning of the eighth day in the past several months. Based on the derivation result in step S54 described above, a combination of daily activity patterns in the morning can be extracted as follows.
{A}, {B}, {C}, {A, D}, {A, B, C, D}
In this embodiment, a daily action pattern group is derived based on the action patterns in a specific time zone for the past several months stored in step S52 (step S53) and stored in step S52. A combination of daily activity patterns in a specific time zone is derived based on the activity patterns in a specific time zone for the past several months and the daily activity pattern group in the specific time zone derived in step S53 (step S54). . However, based on the behavior pattern of a specific time zone of each of the past several months stored in step S52, a daily activity pattern of a specific time zone is derived, and the daily activity pattern of the specific time zone is derived. A combination may be derived. That is, the derivation of the daily activity pattern group in the specific time zone in step S53 and the derivation of the combination of the daily activity pattern in the specific time zone in step S54 may be performed simultaneously. If performed at the same time, in the process of comparing the behavior pattern of a specific time zone on a certain day in the past several months with the daily activity pattern of a specific time zone, the daily activity pattern of a specific time zone and the specific time Since the combinations of daily activity patterns of the obi increase, under the same conditions as in the above-described example, the combinations of the daily activity patterns in the specific time zone (morning) are as follows.
Morning of the first day {A}
Morning of the second day {B}
Morning of the third day {C}
Morning of the fourth day {D}
5th morning {B}
Morning of the 6th day {B, C, D}
Morning of the 7th day {B}
Morning of the 8th day {D}
Next, the daily activity pattern storage unit 30 stores the combination of the daily activity pattern group of the specific time zone derived in step S53 and the daily activity pattern of the specific time zone derived in step S54 (step S55).
[0059]
Next, with reference to the flowchart shown in FIG. 7, a process for determining whether or not there is an abnormality in the behavior pattern on the observation day in the behavior pattern abnormality monitoring system will be described. In steps S60 to S62, the data processing unit 26 performs the same processing as in steps S20 to S22 in FIG. 3 to derive an action pattern symbol string on the observation date (steps S60 and S61), and The pattern symbol string is stored in the action pattern symbol string 29 (step S62).
[0060]
Next, the data processing unit 26 reads a combination of a daily activity pattern group in a specific time zone and a daily activity pattern in a specific time zone stored in the daily activity pattern storage unit 30 (step S63). Based on the action pattern symbol string on the observation day stored in step S62, a combination of action patterns in a specific time zone on the observation day is derived (step S64). It is possible to determine the similarity between any one of the combination of the behavior pattern in the specific time zone on the observation date derived in step S64 and the combination of the daily activity pattern in the specific time zone read in step S63. The similarity is measured by using a function (second function) that can be performed (step S65).
[0061]
For example, suppose that {A, B, C} is derived as a combination of action patterns in a specific time zone on the observation day. Next, for example, the combination of the daily activity patterns in the specific time zone read in step S63 is {A}, {B}, {C}, {A, D}, {A, B, C, D}. (In actual data, the number will be much larger than this. It will be several tens or more.) Then, the activity pattern of the specific time zone and the read daily activity of the specific time zone on the derived observation day will be described. The degree of similarity with the combination of patterns is sequentially measured.
[0062]
For example, here, the combination of the daily activity patterns in the ith specific time zone is M _i Then, in this example, i is 1 to 5. Therefore, the combination M of the daily action patterns in the ith specific time zone _i And the combination of the behavior pattern of the specific time zone on the observation day is as follows.
M ₁ = {A}
M ₂ = {B}
M ₃ = {C}
M ₄ = {A, D}
M ₅ = {A, B, C, D}
Further, assuming that a combination of action patterns in a specific time zone on the observation day is N, N = {A, B, C}. Combination N of action pattern in specific time zone on observation day and combination M of daily action pattern in i-th specific time zone _i The similarity of R _i Then R _i Is measured by Equation 1 which is a second function. That is, the similarity R ₅ Is measured by Expression 2. Similarly, the similarity R ₁ ~ R ₄ Is measured, R ₁ = 0, R ₂ = 0, R ₃ = 0, R ₄ = 0.
[0063]
Next, when it is determined that the comparison has been performed for all the combinations of the daily activity patterns in all the specific time zones (step S66), the data processing unit 26 determines that the activity pattern in the specific time zone on the observation date is the daily activity pattern. It is determined whether the target is appropriate (step S67). That is, the highest similarity among the similarities measured in step S65 is acquired, and it is determined whether or not the acquired similarity exceeds a threshold. When it is determined that the acquired similarity exceeds the threshold, it is determined that the behavior pattern in a specific time zone on the observation day is daily. That is, according to the above-described example, the similarity R measured in step S26 _i Is R ₁ = 0, R ₂ = 0, R ₃ = 0, R ₄ = 0, R ₅ = 0.75, and the highest similarity value is 0.75. Here, for example, when the threshold value is set to 0.6, since the highest similarity value 0.75 exceeds the threshold value 0.6, the behavior pattern in a specific time zone on the observation day is a daily routine. It is determined that there is.
[0064]
On the other hand, when it is determined that the acquired highest similarity does not exceed the threshold, the data processing unit 26 determines that the behavior pattern in a specific time zone on the observation date is not ordinary. Next, the terminals 1 to 3 output the determination result in step S67 on the display unit (step S68). That is, if it is determined in step S67 that it is everyday, an output to that effect is output. On the other hand, if it is determined in step S67 that the subject is not ordinary, it is determined that the subject has changed, and an output indicating that the change has occurred is output.
[0065]
According to the behavior pattern change monitoring system according to the third embodiment, a daily action pattern group in a specific time zone such as morning or afternoon is derived, and the change and the change in the behavior pattern of the subject are grasped. Therefore, it is possible to derive a highly accurate daily behavior pattern of a person whose daily activity pattern is regular for each specific time zone, for example, a person who goes out in the afternoon and a day when he / she does not go out. Therefore, it is possible to quickly and surely detect a change and an unusual change in the behavior pattern of a person whose daily behavior pattern is regular in a specific time zone.
[0066]
Note that the first function used in the above embodiment may be a function based on statistical processing, a Markov model, or a neutral network.
[0067]
Further, in the above-described embodiment, a daily action pattern group is derived using a function capable of determining the degree of similarity, and a daily action pattern combination is derived based on the derived daily action pattern group. By measuring the similarity, it is determined whether or not the behavior pattern on the observation day is daily. However, when used for people who have different daily behavior patterns compared to the past due to changes in the living environment, etc., the daily behavior patterns become extraordinary or extraordinary behavior patterns May occur, so that the group of daily action patterns may be corrected. That is, the daily activity pattern group can be re-derived based on the activity pattern symbol strings of the past several months including the activity pattern symbol strings of the behavior patterns determined to be not the daily activity patterns. Is also good. Further, an action pattern determined to be a daily action pattern may be excluded from the daily action pattern group.
[0068]
【The invention's effect】
According to the present invention, when staying at a specific place such as a toilet during a predetermined time, the specific place can be set as a staying place, and when not staying at a specific place during the predetermined time. In, since the place where the user stayed the longest during the predetermined time can be set as the place of stay, it is possible to derive an action pattern symbol string corresponding to a small change. In addition, a group of daily behavior patterns is derived based on the derived behavior pattern symbol string, and it is not possible to limit a person having various daily behavior patterns or a daily behavior pattern based on the derived daily behavior pattern group. It is possible to derive a highly accurate combination of daily action patterns that can be applied to a person or the like. Therefore, since it is possible to measure the similarity between the action pattern symbol string on the observation day corresponding to the minute change and the combination of the highly accurate daily action pattern, it is possible to early and surely discover the change and the change in the action pattern. it can.
[0069]
In addition, when used for a person who has a regular daily activity pattern for each day of the week, for example, a person who regularly goes out on a specific day of the week, the daily activity pattern for each day of the week is analyzed by analyzing the behavior pattern for each day of the week. A group can be derived, and a combination of daily behavior patterns can be derived based on the derived daily behavior pattern group. In addition, when the daily activity pattern is used for a person having regularity in a specific time zone such as, for example, morning, afternoon, etc., a daily activity pattern group for each time zone obtained by analyzing the activity pattern for each time zone is derived. A combination of daily activity patterns can be derived based on the derived daily activity pattern groups. Therefore, even if a person has a regular behavior pattern every day of the week or a person who has regularity depending on the time of day, it is possible to quickly and surely find a change and an abnormality in the behavior pattern.
[0070]
In addition, when used for a person or the like whose daily behavior pattern is different from the previous one due to a change in the living environment, based on the behavior pattern symbol string including the behavior pattern symbol string of the behavior pattern determined to be unusual. Daily behavior pattern groups can be derived, and the behavior patterns determined to be everyday can be excluded from the daily behavior pattern group. In addition, it is possible to quickly and surely detect a change and an abnormality in the behavior pattern.
[Brief description of the drawings]
FIG. 1 is a block diagram of a behavior pattern change monitoring system according to a first embodiment of the present invention.
FIG. 2 is a flowchart illustrating a process for deriving a daily action pattern in the action pattern change monitoring system according to the first embodiment of the present invention;
FIG. 3 is a flowchart showing a process for determining whether or not the behavior pattern on the observation day has an abnormality in the behavior pattern abnormality monitoring system according to the first embodiment of the present invention;
FIG. 4 is a flowchart of a process for deriving a daily activity pattern on a predetermined day of the week in the behavior pattern change monitoring system according to the second embodiment of the present invention;
FIG. 5 is a flowchart showing a process for determining whether or not there is an abnormality in the behavior pattern on the observation day in the behavior pattern abnormality monitoring system according to the second embodiment of the present invention.
FIG. 6 is a flowchart of a process for deriving a daily activity pattern in a specific time zone in the behavior pattern change monitoring system according to the third embodiment of the present invention;
FIG. 7 is a flowchart showing a process for determining whether or not there is an abnormality in the behavior pattern on the observation day in the behavior pattern abnormality monitoring system according to the third embodiment of the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 ... House system, 12 ... Center management server, 14, 16 ... Network, 18 ... House management server, 20 ... Data processing part, 22 ... Living activity data storage part, 24 ... Communication control part, 26 ... Data processing part , 28: Life activity data storage unit, 29: Behavior pattern unit symbol string storage unit, 30: Daily activity pattern storage unit, 32: Communication control unit

Claims (6)

Living activity data storage means for storing a plurality of days of living activity data including a moving place and time of a person detected based on a reaction of the human sensor,
Regarding the living activity data for a plurality of days stored by the living activity data storage means, an action pattern symbol string deriving means for deriving an action pattern symbol string for each day,
Daily action pattern deriving means for deriving a daily action pattern group based on the action pattern symbol string derived by the action pattern symbol string deriving means,
Daily action pattern combination deriving means for deriving a combination of daily action patterns based on a daily action pattern group derived by the daily action pattern deriving means,
A comparison means for comparing a behavior pattern symbol string derived on the observation day derived by the behavior pattern symbol string derivation means, and a combination of daily behavior patterns derived by the daily behavior pattern combination derivation means,
Determining means for determining whether the action pattern on the observation day is a daily action pattern based on the comparison result by the comparing means,
Output means for outputting a determination result by the determination means,
A behavior pattern change monitoring system, comprising: The action pattern symbol string deriving means determines whether or not the user has stayed at a specific place during a predetermined time, and if it is determined that the user has stayed at the specific place, the stay at the specific place during the predetermined time is determined. 2. The behavior pattern change monitoring system according to claim 1, wherein the place is a place, and when it is determined that the user has not stayed at the specific place, the place where the user has stayed the longest during the predetermined time is set as a stay place. The daily activity pattern deriving unit derives a daily activity pattern group for a predetermined day of the week based on an activity pattern symbol string for a predetermined day of the week derived by the activity pattern symbol string deriving unit. The behavior pattern change monitoring system according to claim 2. The daily activity pattern deriving means derives a daily activity pattern group in a specific time zone based on an activity pattern symbol string in a specific time zone derived by the activity pattern symbol string deriving means. The behavior pattern abnormality monitoring system according to claim 1 or 2. The daily action pattern deriving means derives a daily action pattern group based on an action pattern symbol string including an action pattern symbol string of an action pattern determined to be not a daily action pattern by the determination means. The behavior pattern change monitoring system according to any one of claims 1 to 4. The method according to any one of claims 1 to 5, wherein at least one of the behavior patterns derived as the daily behavior pattern by the daily behavior pattern deriving unit is excluded. Behavior pattern change monitoring system.

Images