JP4405127B2

JP4405127B2 - Physical condition discrimination system

Info

Publication number: JP4405127B2
Application number: JP2002015307A
Authority: JP
Inventors: 浩明庭本; 和利滝本; 徹和辻
Original assignee: Sharp Corp
Current assignee: Sharp Corp
Priority date: 2002-01-24
Filing date: 2002-01-24
Publication date: 2010-01-27
Anticipated expiration: 2022-01-24
Also published as: JP2003210419A

Description

【０００１】
【発明の属する技術分野】
本発明は体調判別システムに関し、より詳細には被験者、特に高齢者の日常の活動をプライバシーを確保しながら最も自然な形で集積して、被験者の体調を判別し、介護・看護・医療・福祉機関の活動に役立てる体調判別システムに関するものである。
【０００２】
【従来の技術】
わが国において、６５歳以上の高齢者の人口に占める割合が今後５０年の間に３０％を超えると言われている。このような状況下で、要介護老人の対策が現在注目・検討されているが、一方で多くの健康な高齢者が存在していることも見過ごしてはならない。健康な高齢者の中でも一人暮らしの高齢者は、空間的・人間関係的に孤立していることが多く、これらの人の健康状態を把握することは大変重要なことではあるが、特定多数の人の健康状態を把握・管理するには人手が圧倒的に不足している。
【０００３】
このため例えば、自治体の中にはペンダント型無線式緊急通報システムを導入しているところもあるが、これは緊急時のみに使用されるものであり高齢者の日常の健康状態を把握するものではない。また、宅内に多種のセンサを配置して高齢者の生活活動を監視するサービスシステムが警備会社などから提供されているが、システム導入および維持の費用が高いため高額所得者にしか普及していないのが現状である。さらに特開平１０−２４８０９３号公報では、家庭用電気製品にセンサを取付けて、電気ポットの給湯スイッチのオン・オフなどを検出して生活者の安否などを監視するシステムが提案されているが、電気製品の設置場所および情報量に制限があり、加えて電気製品をほとんど使わない人や夜間に電気製品の電源プラグをコンセントから抜いておく人などの場合にはこのシステムでは十分な監視はできない。
【０００４】
【発明が解決しようとする課題】
本発明はこのような従来の問題に鑑みてなされたものであり、低い導入費用で、しかも高齢者などの被験者の日常の活動をプライバシーを確保しながら最も自然な形で集積でき、さらには被験者の体調を判別できるシステムを提供することをその目的とするものである。
【０００６】
【課題を解決するための手段】
本発明によれば、被験者の活動を検知する検知手段と、該検知手段からの検知信号を受信し記憶する集積手段と、該集積手段に記憶された検知信号を解析し被験者の体調を判別する解析判別手段とを有し、前記検知手段は家庭用電気機器（以下「家電機器」と記すことがある）であり、前記解析判別手段は、前記家庭用電気機器の使用時刻から隣接するイベントの時間間隔を算出し、この隣接するイベントの時間間隔を横軸とするとともに、それぞれの隣接イベントの時間間隔における個数を全個数で割ったものを縦軸としたヒストグラムを作成し、このヒストグラムから線形回帰を行って直線の傾きの絶対値を隣接イベント分布係数として求める解析部と、予め累積した検知信号を解析して得られた基準隣接イベント分布係数を記憶する記憶部と、検知信号の解析結果の隣接イベント分布係数と前記基準隣接イベント分布係数とを比較して被験者の体調を判別する評価部とを備えることを特徴とする体調判別システムが提供される。
【０００８】
前記体調判別システムは解析判別手段による体調の判別結果を送信する送信手段をさらに備え、介護・看護・医療・福祉機関に前記判別結果を送信するのが望ましい。
【００１１】
【発明の実施の形態】
本発明者等は、低い導入費用で、しかも高齢者などの被験者の日常の活動をプライバシーを確保しながら最も自然な形で集積でき、さらには被験者の体調を的確に判別できるシステムを提供できないか鋭意検討を重ねた結果、日常生活で違和感なく利用され且つ広く普及している家電製品を検知手段として用い、家電製品の使用状況を検知・集積し、これを解析することにより被験者の体調を判別すればよいことを見出し本発明をなすに至った。
【００１２】
本実施形態に係る体調判別システム及び体調判別方法の大きな特徴は、家庭用電気機器の使用時刻から隣接イベント分布係数又は使用時間を算出し、予め検知・算出した前記家庭用電気機器の基準隣接イベント分布係数又は使用時間と比較して被験者の体調を判別することにある。
【００１３】
図１に、この体調判別システムの概説図を示す。宅内に配設された家電機器１で検知された信号は集積手段２に送られ、ここで時系列データとして集積・記憶される。集積手段２に記憶された時系列データは所定時間ごとあるいは送信要求があったときに解析判別手段３に送られる。解析判別手段では収集された時系列データを解析・判別を行う。解析判別手段３による判別結果は送信手段４を用いて必要により被験者本人５および介護・看護・医療・福祉機関６に送られ、健康状態の自己管理や当該機関による活動に活用される。この場合、被験者本人５およびその家族、前記機関６などからの要求によっても、判別された体調が送信手段４を介して要求者に送信されるようにしてもよい。以下各手段について個別に説明する。
【００１４】
まず前記家電機器としては、例えばエアコン、テレビ、冷蔵庫、電子レンジ、マットなどの従来公知の家電製品を用いることができ、これらを複数個組み合わせて使用するのが被験者の体調をより詳細に検知する上で望ましい。家電製品における検知信号としては、例えばスイッチのオン・オフや扉の開閉、温度の設定、運転モード、コンプレッサ回転数、リモコン操作、消費電力などが挙げられる。
【００１５】
前記各種家電機器で検知された検知信号は集積手段に送られ、ここで時系列データとして記憶される。検知手段から集積手段への送信方法について特に限定はないが、新たな配線工事を必要としない点から無線や電灯線搬送を用いて送信するのが好ましい。
【００１６】
集積手段に集積・記憶された時系列データは解析判別手段に送信される。集積手段から解析判別手段への送信方法に特に限定はなく、電話回線やＣＡＴＶ回線などを利用すればよい。なお図１では集積手段２と解析判別手段３を別体としているが、両者を一体として形成してももちろん構わない。また、集積手段２、解析判別手段３、送信手段４のすべてを被検者宅に設けてもよく、あるいは被検者宅とは別の所（例えば集中管理センターなど）に設けてもよい。
【００１７】
次に、解析判別手段における解析・判別方法について説明する。解析判別手段では、被験者の体調と関連づけられる解析結果を基準データとして記憶部に記憶しておき、時系列データの解析結果と基準データとの比較を行い、一致又は近似している基準データから被験者の体調を判別する。この方法の概説図を図２に示す。集積手段２から解析判別手段３へ送られてきた時系列データはまず解析部３１で解析される。具体的解析手法については後述する。そして被験者の体調と関連づけられる解析結果は基準データとして記憶部３２に記憶される。一方、順次送られてくる時系列データの解析結果は評価部３３においてこの基準データと比較され、一致又は近似している基準データから被験者の体調が推測・判別され、出力部３４で出力される。
【００１８】
ここでまず、時系列データの解析方法として家電機器の使用時刻から算出した隣接イベント分布係数を用いる場合を説明する。なお、隣接イベント分布係数は、統計分布においてサンプルのバラツキ具合を定量化した値であって、家電機器が散漫に使用されたか、集中して使用されたかを示すものである。
【００１９】
本発明者等は、１００人の被験者（独居者）の家の種々の家電製品に検知手段を配設し、その使用状況を１ヶ月にわたって調査した。この結果、家電製品の使用についての隣接イベント分布係数と被験者の体調との間に密接な相関関係があるという新たな知見を得た。以下、家電機器として冷蔵庫を用いた場合を例に説明する。
【００２０】
冷蔵庫に扉開閉検知器を設け、冷蔵庫の扉が開閉されると、開閉された日時と「開」又は「閉」の信号を解析部３１（図２に図示）に送るようにする。扉開閉検知器と解析部とは例えば電力線で接続し、電力線上に構築されたネットワークの基に扉開閉検知器から解析部へデータを送る。図３に、解析部に送る時系列データの一例を示す。図３では、例えば最初の行で、’０１年１０月２１日の１３時９分１４秒に冷蔵庫の扉が開かれたことを示し、次の行で、同日の１３時９分１８秒に扉が閉じられたことを示している。
【００２１】
この冷蔵庫の扉の開閉時刻データから隣接イベント分布係数を次のようにして算出する。まず、冷蔵庫の扉が開かれた（イベント）時刻を抽出し、隣接するイベントの時間間隔を算出する。例えば図３では、１３時９分１４秒、１３時９分５３秒、１３時２０分１１秒、１３時２１分９秒に冷蔵庫の扉が開かれているから、隣接イベントの時間間隔は３９秒、１０分２８秒、５８秒である。
【００２２】
つぎに、この隣接イベントの時間間隔を横軸とし、個数を縦軸としてヒストグラムを作成する。ここで、被験者１００人の隣接イベントの時間間隔を実際に算出したところ、この時間間隔は１秒から数十時間まで広い範囲にわたっていることがわかったので、縦軸・横軸共にlogスケールを用いることにし、さらに被験者の違いや日付けの違いが顕著に現れないようにするため、縦軸として、それぞれの隣接イベントの時間間隔における個数を全個数で割ったものを用いことにする。具体的には、横軸の目盛間隔を例えばＩ＝１０^0.5秒として、０〜１０^0.5秒間の隣接イベント時間間隔であった個数Ｎ（１０^0.5）、１０^0.5〜１０^1.0秒間の隣接イベント時間間隔であった個数Ｎ（１０^1.0）、・・・、１０^b-0.5〜１０^b秒間の隣接イベント時間間隔であった個数Ｎ（１０^b）をそれぞれ算出する。次に全個数をＭとして、Ｐ（１０^b）＝Ｎ（１０^b）／Ｍから各隣接イベント時間間隔の個数割合を算出する。以上のようにして算出した隣接イベント時間間隔（ＩＥＩ：Inter Event Interval）を横軸とし、その個数割合（Ｐ（ＩＥＩ））を縦軸としてlogスケールのヒストグラムを作成する。そして作成したヒストグラムから線形回帰を行い、その直線の傾きの絶対値を隣接イベント分布係数とする。ヒストグラムの一例を図４に示す。このヒストグラムは、後述する活力のある被験者（得点２４点、隣接イベント分布係数０．９５）と活力のない被験者（得点３点、隣接イベント分布係数１．３８）のものである。
【００２３】
一方、被験者７８人について２９日間の実験後に「ＰＯＭＳ検査」を行い、隣接イベント分布係数との関係を調べた。なお、「ＰＯＭＳ検査」は検査直前の１週間の間の体調に関するアンケートであるので、隣接イベント分布係数についても１週間分データに基づいて算出した。「ＰＯＭＳ検査」の各項目について得点の高い方から１０人、低い方から１０人を抽出して２つのグループとし、これら２つのグループと隣接イベント分布係数との間の相関関係を調べた結果、得点の高いグループ（活力のあるグループ）は平均得点が２４．５点、隣接イベント分布係数が１．００であったのに対し、得点の低いグループ（活力のないグループ）は平均得点が３．０９点、隣接イベント分布係数が１．２３であった。また一般に「ｐ値」と呼ばれる確率を示す値を算出した結果０．００３と非常に小さい値であった。このことから被験者の活力があるほど隣接イベント分布係数は小さくなり、被験者の活力がないほど隣接イベント分布係数は大きくなることがわかる。なおｐ値は、確率を示す値であって、被験者の体調と隣接イベント分布係数との間に相関関係がないと仮定した場合に、現実にそのような調査結果が得られる確率である。つまり、ｐ値が十分に小さい場合には、相関関係がないとした仮説が棄却され、両者に相関関係があるということになる。
【００２４】
また、家電機器としてテレビを用いた場合にも、冷蔵庫を用いた場合と同様に被験者の日々の体調を判別できる。冷蔵庫の場合における扉の開閉時刻の代わりに、テレビの場合には遠隔操作器（以下「リモコン」と記すことがある）からの信号送信時刻を用いる。すなわち、電源のオン・オフ、チャンネルの切換、音量の増・減といったリモコンからテレビに送られる信号の送信時刻を用いる。検知データの統計学的処理、隣接イベント分布係数の算出方法、及び「ＰＯＭＳ検査」については前記と同様であるのでここではその説明を省略し、その結果だけを下記に示す。
【００２５】
「ＰＯＭＳ検査」の”怒り”項目について得点の高い方から１０人、低い方から１０人を抽出して２つのグループとし、これら２つのグループと隣接イベント分布係数との間の相関関係を調べた結果、得点の高いグループ（”怒り”のあるグループ）は平均得点が２５．１点、隣接イベント分布係数が１．１４であったのに対し、得点の低いグループ（”怒り”のないグループ）は平均得点が０．０点、隣接イベント分布係数が１．２７で、ｐ値は０．０３と小さい値であった。このことから被験者の”怒り”があるほど隣接イベント分布係数は小さくなり、被験者の”怒り”がないほど隣接イベント分布係数は大きくなることがわかる。
【００２６】
以上から明らかなように、被験者の体調と隣接イベント分布係数との関係を基準データとして予め記憶させておき、日々の冷蔵庫の扉開閉操作やテレビのリモコン操作などから算出される隣接イベント分布係数を基準データと比較することにより被験者の日々の体調を判別できるようになる。なお、前記実施例では家電機器として冷蔵庫およびテレビを用いているが、家電機器としてこれらに限定されるものではなく、従来公知の家電機器を利用することができる。また、複数の家電機器の使用状態から複数の隣接イベント分布係数を算出することにより、被験者の体調をより的確に判別できるようになる。
【００２７】
次に、体調の解析方法として、家電機器の使用時間を用いる場合について説明する。隣接イベント分布係数を用いた場合と同様に、７８人の被験者（独居者）の家の種々の家電製品に検知手段を配設し、その使用状況を２９日間にわたって調査した。この結果、家電製品の使用時間と被験者の体調との間に密接な相関関係があるという新たな知見が得られた。以下、家電機器として冷蔵庫を用いた場合を例に説明する。なお、冷蔵庫の場合の使用時間とは冷蔵庫の扉が開けられてから閉じられるまでの時間をいう。
【００２８】
冷蔵庫に扉開閉検知器を設け、冷蔵庫の扉が開閉されると、開閉された時刻と「開」又は「閉」の信号を検知する。そして、冷蔵庫の使用時間に基づいて下記式から平均使用時間Ｔを算出する一方、前記の「ＰＯＭＳ検査」を行い”うつ”と”混乱”の体調項目と冷蔵庫の使用時間との関係を調べた。結果を表１に示す。
Ｔ＝１／Ｍ×Σ（ｔ_i）
（式中、Ｔ：平均使用時間、Ｍ：扉開閉回数、ｔ_i：各使用時間）
【００２９】
【表１】

【００３０】
表１によれば、「ＰＯＭＳ検査」の結果、被験者のうち４３人が”うつ”状態の可能性が低く、３４人が”うつ”状態の可能性が高かった。そして”うつ”状態の可能性の低い被験者４３人の平均使用時間が１２．０秒であるのに対し、”うつ”状態の可能性の高い３４人の平均使用時間は１５．３秒であった。このことから平均使用時間が長いと”うつ”状態である可能性が高く、平均使用時間が短いと”うつ”状態である可能性が低いと考えられる。そこで前出のｐ値を算出したところ０．００３４と非常に小さい値を示し、使用時間と”うつ”とは相関関係にあることが認められた。
【００３１】
また、「ＰＯＭＳ検査」の結果、被験者のうち４５人が”混乱”状態の可能性が低く、３２人が”混乱”状態の可能性の高かった。そして”混乱”状態の可能性が低い４５人の平均使用時間は１２．０秒であるのに対し、”混乱”状態の可能性の高い３２人の平均使用時間は１４．７秒であった。このことから平均使用時間が長いと”混乱”状態である可能性が高く、平均使用時間が短いと”混乱”状態のある可能性が低いと考えられる。そこで前記と同様にｐ値を算出したところ０．０１２と小さい値を示し、使用時間と”混乱”とは相関関係にあることが認められた。
【００３２】
したがって、被験者の体調と使用時間との関係を基準データとして予め記憶させておき、日々の冷蔵庫の扉開閉操作から算出される使用時間を基準データと比較することにより被験者の日々の体調を判別できるようになる。なお、前記実施例では家電機器として冷蔵庫を用いているが、家電機器としてこれらに限定されるものではなく、従来公知の家電機器をもちろん利用することができる。また、複数の家電機器の使用状態から複数の隣接イベント分布係数を算出することにより、被験者の体調をより的確に判別できるようになる。
【００３３】
【発明の効果】
本発明の体調判別システムでは、前記検知手段は家庭用電気機器であり、前記解析判別手段は、前記家庭用電気機器の使用時刻から隣接するイベントの時間間隔を算出し、この隣接するイベントの時間間隔を横軸とするとともに、それぞれの隣接イベントの時間間隔における個数を全個数で割ったものを縦軸としたヒストグラムを作成し、このヒストグラムから線形回帰を行って直線の傾きの絶対値を隣接イベント分布係数として求める解析部と、予め累積した検知信号を解析して得られた基準隣接イベント分布係数を記憶する記憶部と、検知信号の解析結果の隣接イベント分布係数と前記基準隣接イベント分布係数とを比較して被験者の体調を判別する評価部とを備えるので、低い導入費用で、しかも高齢者などの被験者の日常の活動をプライバシーを確保しながら最も自然な形で集積でき、さらには被験者の体調を的確に判別できる。
【図面の簡単な説明】
【図１】本発明の体調判別システムの概略構成図である。
【図２】解析判別手段の一例を示す概説図である。
【図３】冷蔵庫の扉開閉についての時系列データの一例である。
【図４】隣接イベント時間間隔のヒストグラムである。
【符号の説明】
１家電機器（検知手段）
２集積手段
３解析判別手段
４送信手段
３１解析部
３２記憶部
３３判別部
３４出力部[0001]
BACKGROUND OF THE INVENTION
The present invention also relates to the physical condition determination system, and more particularly to an integrated subject, especially in the elderly of daily activities the most natural while ensuring the privacy form, to determine the physical condition of the subject, nursing care and nursing and medical care - help in the activities of the welfare institutions are those related to the physical condition discrimination system.
[0002]
[Prior art]
In Japan, it is said that the proportion of elderly people over the age of 65 will exceed 30% over the next 50 years. Under such circumstances, measures for elderly people requiring care are currently attracting attention and consideration, but on the other hand, it should not be overlooked that there are many healthy elderly people. Among healthy elderly people, those who live alone are often isolated in terms of space and relationships, and it is very important to understand the health status of these people, but a large number of people There is an overwhelming shortage of human resources to understand and manage the health status of people.
[0003]
For this reason, for example, some local governments have introduced a pendant type wireless emergency call system, but this is only used in an emergency and is not intended to grasp the daily health of the elderly. Absent. Service systems that monitor the activities of elderly people by placing various sensors in the home are provided by security companies, etc., but they are popular only for high-income people due to the high cost of system installation and maintenance. is the current situation. Furthermore, in Japanese Patent Laid-Open No. 10-248093, a system is proposed in which a sensor is attached to a household electric appliance to detect on / off of a hot water supply switch of an electric pot and monitor the safety of a consumer. There is a limit to the location and amount of information on electrical appliances. In addition, this system does not provide sufficient monitoring for people who rarely use electrical appliances or who unplug electrical appliances from electrical outlets at night. .
[0004]
[Problems to be solved by the invention]
The present invention has been made in view of such conventional problems, and can be accumulated in the most natural form while ensuring privacy while keeping daily activities of subjects such as elderly people at a low introduction cost and ensuring privacy. It is an object of the present invention to provide a system capable of discriminating the physical condition.
[0006]
[Means for Solving the Problems]
According to the present invention, the detection means for detecting the activity of the subject, the accumulation means for receiving and storing the detection signal from the detection means, and analyzing the detection signal stored in the accumulation means to determine the physical condition of the subject. An analysis discriminating means, wherein the detection means is a household electric appliance (hereinafter, sometimes referred to as “home appliance”) , and the analysis discriminating means detects an event that is adjacent from the use time of the household electric appliance. A time interval is calculated, and the time interval between the adjacent events is plotted on the horizontal axis, and a histogram is created by dividing the number of each adjacent event in the time interval by the total number. serial to store an analyzing unit for obtaining the absolute value of the slope of the line as an adjacent event distribution coefficient by performing a regression, a reference neighbor event distribution coefficients obtained by analyzing the detection signal obtained by accumulating pre Me And parts, physical condition determination system characterized in that it comprises an evaluation unit which compares the adjacent event distribution coefficient and the reference neighbor event distribution coefficients of the analysis results of the sensing signals to determine the physical condition of a subject is provided.
[0008]
Preferably, the physical condition determination system further includes a transmission means for transmitting a physical condition determination result by the analysis determination means, and transmits the determination result to a care / nursing / medical / welfare organization.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Is it possible for the present inventors to provide a system that can accumulate daily activities of subjects such as the elderly in the most natural form while ensuring privacy and that can accurately determine the physical condition of the subject at a low introduction cost? As a result of intensive studies, household appliances that are widely used in daily life and widely spread are used as detection means, and the usage status of household appliances is detected and collected, and the physical condition of the subject is determined by analyzing this. As a result, the present invention has been found.
[0012]
Major feature of the physical condition determination system and physical condition determination method according to the present embodiment calculates the from the use time of the home electric appliance adjacent event distribution coefficient or use time, the reference neighbor in advance detected and calculated the household electric equipment It is to determine the physical condition of the subject in comparison with the event distribution coefficient or use time.
[0013]
FIG. 1 shows a schematic diagram of this physical condition discrimination system. Signals detected by the home electric appliance 1 disposed in the house are sent to the accumulating means 2 where they are accumulated and stored as time series data. The time-series data stored in the accumulating unit 2 is sent to the analysis discriminating unit 3 every predetermined time or when there is a transmission request. The analysis discriminating means analyzes and discriminates the collected time series data. The determination result by the analysis determination means 3 is sent to the subject person 5 and the care / nursing / medical / welfare institution 6 as necessary using the transmission means 4, and is utilized for self-management of health status and activities by the institution. In this case, the determined physical condition may be transmitted to the requester via the transmission unit 4 even in response to a request from the subject himself / herself 5 and his / her family, the organization 6 and the like. Each means will be described individually below.
[0014]
First, as the household electrical appliances, for example, conventionally known household electrical appliances such as an air conditioner, a television, a refrigerator, a microwave oven, and a mat can be used, and a combination of a plurality of these senses the physical condition of the subject in more detail. Desirable above. As detection signals in home appliances, for example, switch on / off, door opening / closing, temperature setting, operation mode, compressor rotation speed, remote control operation, power consumption, and the like can be given.
[0015]
Detection signals detected by the various household electrical appliances are sent to the accumulating means, where they are stored as time series data. Although there is no particular limitation on the transmission method from the detection means to the accumulation means, it is preferable to perform transmission using radio or power line conveyance because no new wiring work is required.
[0016]
The time series data accumulated and stored in the accumulating means is transmitted to the analysis discriminating means. There is no particular limitation on the transmission method from the accumulating means to the analysis discriminating means, and a telephone line or CATV line may be used. In FIG. 1, the accumulating means 2 and the analysis discriminating means 3 are separated from each other, but they may of course be formed integrally. Further, all of the accumulating unit 2, the analysis discriminating unit 3, and the transmitting unit 4 may be provided in the subject's home, or may be provided in a place different from the subject's home (for example, a centralized management center).
[0017]
Next, an analysis / discrimination method in the analysis discrimination means will be described. In the analysis discriminating means, the analysis result associated with the physical condition of the subject is stored in the storage unit as reference data, the analysis result of the time-series data is compared with the reference data, and the subject is determined based on the reference data that matches or approximates. Determine the physical condition. A schematic diagram of this method is shown in FIG. The time series data sent from the accumulating unit 2 to the analysis discriminating unit 3 is first analyzed by the analyzing unit 31. A specific analysis method will be described later. And the analysis result linked | related with a test subject's physical condition is memorize | stored in the memory | storage part 32 as reference | standard data. On the other hand, the analysis result of the time-series data sent sequentially is compared with this reference data in the evaluation unit 33, and the physical condition of the subject is estimated / discriminated from the reference data that matches or approximates, and is output from the output unit 34. .
[0018]
First, the case where the adjacent event distribution coefficient calculated from the usage time of the home appliance is used as the time series data analysis method will be described. The adjacent event distribution coefficient is a value obtained by quantifying the degree of sample variation in the statistical distribution, and indicates whether home appliances are used in a diffuse manner or in a concentrated manner.
[0019]
The inventors of the present invention installed detection means on various home appliances in the homes of 100 subjects (single persons), and investigated the usage status for one month. As a result, we obtained new knowledge that there is a close correlation between the distribution coefficient of adjacent events regarding the use of home appliances and the physical condition of the subject. Hereinafter, the case where a refrigerator is used as a home appliance will be described as an example.
[0020]
A door open / close detector is provided in the refrigerator, and when the door of the refrigerator is opened / closed, the date / time of opening / closing and an “open” or “closed” signal are sent to the analysis unit 31 (shown in FIG. 2). The door open / close detector and the analysis unit are connected by, for example, a power line, and data is sent from the door open / close detector to the analysis unit based on a network constructed on the power line. FIG. 3 shows an example of time-series data sent to the analysis unit. In FIG. 3, for example, the first line shows that the refrigerator door was opened at 13: 9: 14 on October 21, 2001, and the next line at 13: 9: 18 Indicates that the door has been closed.
[0021]
The adjacent event distribution coefficient is calculated from the opening / closing time data of the refrigerator door as follows. First, the time when the refrigerator door is opened (event) is extracted, and the time interval between adjacent events is calculated. For example, in FIG. 3, since the refrigerator door is opened at 13: 9: 14, 13: 9: 53, 13:20:11, 13: 21: 9, the time interval between adjacent events is 39 Seconds, 10 minutes 28 seconds, and 58 seconds.
[0022]
Next, a histogram is created with the time interval between adjacent events as the horizontal axis and the number as the vertical axis. Here, when the time interval between adjacent events of 100 subjects was actually calculated, it was found that this time interval spans a wide range from 1 second to several tens of hours, so the log scale is used for both the vertical and horizontal axes. In addition, in order to prevent the difference between subjects and the date from appearing prominently, the vertical axis is obtained by dividing the number of adjacent events in the time interval by the total number. Specifically, as the scale intervals of the horizontal axis for example I = 10 ^0.5 sec, the number N (10 ^0.5) was adjacent event time intervals of 0 ^0.5 seconds, adjacent event time interval of 10 ^0.5-10 ^1.0 seconds N (10 ^1.0 ),..., 10 ^b-0.5 to 10 ^b seconds, and the number N (10 ^b ) that is the adjacent event time interval is calculated. Next, assuming that the total number is M, the number ratio of each adjacent event time interval is calculated from P (10 ^b ) = N (10 ^b ) / M. The thus adjacent event time interval is calculated by (IEI: Inter Event Interval) and the horizontal axis, to create a histogram of log scale and the number ratio of (P (IEI)) as the vertical axis. Then, linear regression is performed from the created histogram, and the absolute value of the slope of the straight line is set as the adjacent event distribution coefficient. An example of the histogram is shown in FIG. This histogram is for subjects with vitality (score 24 points, adjacent event distribution coefficient 0.95) and subjects with no vitality (score 3 points, adjacent event distribution coefficient 1.38) described later.
[0023]
On the other hand, 78 subjects were subjected to a “POMS test” after 29 days of experiment to examine the relationship with the adjacent event distribution coefficient. Since the “POMS test” is a questionnaire related to physical condition during the week immediately before the test, the adjacent event distribution coefficient was also calculated based on the data for one week. As a result of examining the correlation between these two groups and the adjacent event distribution coefficient by extracting 10 people from the highest score and 10 people from the lowest score for each item of “POMS examination”, The group with high score (active group) had an average score of 24.5 points and the adjacent event distribution coefficient was 1.00, while the group with low score (group with no vitality) had an average score of 3. 09 points and the adjacent event distribution coefficient was 1.23. Further, a value indicating a probability generally called “p value” was calculated, and as a result, it was a very small value of 0.003. This shows that the adjacent event distribution coefficient decreases as the subject's vitality increases, and the adjacent event distribution coefficient increases as the subject's vitality decreases. The p value is a value indicating a probability, and is a probability that such a survey result is actually obtained when it is assumed that there is no correlation between the physical condition of the subject and the adjacent event distribution coefficient. That is, when the p-value is sufficiently small, the hypothesis that there is no correlation is rejected, and both are correlated.
[0024]
In addition, even when a television is used as the home appliance, the daily physical condition of the subject can be determined as in the case where the refrigerator is used. Instead of the opening / closing time of the door in the case of a refrigerator, a signal transmission time from a remote controller (hereinafter sometimes referred to as “remote control”) is used in the case of a television. That is, the transmission time of a signal sent from the remote controller to the television, such as power on / off, channel switching, volume increase / decrease, is used. Since the statistical processing of the detection data, the calculation method of the adjacent event distribution coefficient, and the “POMS inspection” are the same as described above, the description thereof is omitted here, and only the result is shown below.
[0025]
For the “anger” item of “POMS test”, 10 people from the highest score and 10 from the lowest score were extracted into two groups, and the correlation between these two groups and the adjacent event distribution coefficient was examined. As a result, the group with high score (group with “anger”) had an average score of 25.1 points and the adjacent event distribution coefficient was 1.14, whereas the group with low score (group without “anger”) The average score was 0.0, the adjacent event distribution coefficient was 1.27, and the p-value was as small as 0.03. From this, it can be seen that the adjacent event distribution coefficient decreases as the subject's “anger” increases, and the adjacent event distribution coefficient increases as the subject's “anger” does not exist.
[0026]
As is clear from the above, the relationship between the physical condition of the subject and the adjacent event distribution coefficient is stored in advance as reference data, and the adjacent event distribution coefficient calculated from the daily refrigerator door opening / closing operation or the TV remote control operation is calculated. The daily physical condition of the subject can be discriminated by comparing with the reference data. In addition, although the refrigerator and the television are used as the home appliances in the embodiment, the home appliances are not limited to these, and conventionally known home appliances can be used. In addition, by calculating a plurality of adjacent event distribution coefficients from the usage states of a plurality of home appliances, the physical condition of the subject can be determined more accurately.
[0027]
Next, the case where the usage time of household appliances is used as a physical condition analysis method will be described. As in the case of using the adjacent event distribution coefficient, detection means were arranged in various home electric appliances in the homes of 78 subjects (single occupants), and the usage situation was investigated over 29 days. As a result, a new finding that a close correlation between the physical condition of the use time and the subject of consumer electronics product has been obtained, et al. Hereinafter, the case where a refrigerator is used as a home appliance will be described as an example. In addition, the use time in the case of a refrigerator means the time after the door of a refrigerator is opened until it is closed.
[0028]
A door open / close detector is provided in the refrigerator, and when the door of the refrigerator is opened and closed, the time at which the door is opened and closed and an “open” or “closed” signal are detected. Then, while calculating the average usage time T from the following formula based on the usage time of the refrigerator, the above-mentioned “POMS test” was performed to examine the relationship between the physical condition items of “depression” and “confused” and the usage time of the refrigerator. . The results are shown in Table 1.
T = 1 / M × Σ (t _i )
(Wherein, T: average usage time, M: number of door opening / closing, t _i : each usage time)
[0029]
[Table 1]

[0030]
According to Table 1, as a result of the “POMS test”, 43 of the subjects were less likely to be “depressed” and 34 were more likely to be “depressed”. The "depression" average use time possibilities have low test subjects 43 who states whereas a 12.0 seconds, "depression" likely state 34 people average usage time of 15.3 seconds Met. From this, it can be considered that if the average usage time is long, there is a high possibility of being in a “depression” state, and if the average usage time is short, the possibility of being in a “depression” state is low. Therefore, when the above-mentioned p value was calculated, it was found to be a very small value of 0.0034, and it was recognized that the usage time and “depression” are correlated.
[0031]
As a result of the “POMS test”, 45 of the subjects were less likely to be in a “confused” state and 32 were more likely to be in a “confused” state. And the average usage time of 45 people with low possibility of “confused” status was 12.0 seconds, whereas the average usage time of 32 people with high possibility of “confused” status was 14.7 seconds. . From this, it is considered that if the average usage time is long, there is a high possibility of being in a “confused” state, and if the average usage time is short, the possibility of being in a “confused” state is low. Therefore, when the p-value was calculated in the same manner as described above, it was as small as 0.012, and it was recognized that the usage time and the “confused” were correlated.
[0032]
Therefore, the daily physical condition of the subject can be determined by storing the relationship between the physical condition of the subject and the usage time in advance as reference data, and comparing the usage time calculated from the daily door opening / closing operation of the refrigerator with the reference data. It becomes like this. In addition, although the refrigerator is used as a household appliance in the said Example, it is not limited to these as a household appliance, Of course, a conventionally well-known household appliance can be utilized. In addition, by calculating a plurality of adjacent event distribution coefficients from the usage states of a plurality of home appliances, the physical condition of the subject can be more accurately determined.
[0033]
【The invention's effect】
In physical condition determination system of the present invention, the detection means is a household electric appliance, it said analyzing determination means calculates a time interval between events adjacent from the use time of the household electric appliance, the time of the adjacent event Create a histogram with the horizontal axis as the interval and the vertical axis divided by the total number of each adjacent event in the time interval, and perform linear regression from this histogram to find the absolute value of the slope of the line events and analysis section for determining the distribution coefficients, a memory unit for storing a reference adjacent event distribution coefficients obtained by analyzing the detection signal obtained by accumulating pre Me, analysis results of the detection signals neighboring event distribution coefficient and the reference neighbor event distribution so by comparing the coefficient and an evaluation unit to determine the physical condition of the subject, low in initial cost, yet the daily activities of the subject, such as the elderly-flops While ensuring the Ibashi it can be integrated in the most natural way, and further can determine accurately the physical condition of the subject.
[Brief description of the drawings]
FIG. 1 is a schematic configuration diagram of a physical condition discrimination system according to the present invention.
FIG. 2 is a schematic diagram illustrating an example of an analysis determination unit.
FIG. 3 is an example of time-series data for opening / closing a refrigerator door.
FIG. 4 is a histogram of adjacent event time intervals.
[Explanation of symbols]
1 Home appliances (detection means)
2 Accumulating means 3 Analysis discriminating means 4 Transmitting means 31 Analyzing section 32 Storage section 33 Discriminating section 34 Output section

Claims

A detecting means for detecting the activity of the subject; an accumulating means for receiving and storing a detection signal from the detecting means; and an analysis discriminating means for analyzing the detection signal stored in the accumulating means to determine the physical condition of the subject. And
The detection means is a household electrical appliance ;
The analysis determination means includes
The time interval between adjacent events is calculated from the use time of the household electrical equipment, and the time interval between the adjacent events is plotted on the horizontal axis, and the number of each adjacent event in the time interval is divided by the total number. Create a histogram with a vertical axis, perform linear regression from this histogram, and calculate the absolute value of the slope of the straight line as an adjacent event distribution coefficient,
A storage unit for storing a reference adjacent event distribution coefficients obtained by analyzing the detection signal obtained by accumulating Me pre,
Physical condition determination system characterized in that it comprises an evaluation unit which compares the adjacent event distribution coefficient and the reference neighbor event distribution coefficients of the analysis results of the sensing signals to determine the physical condition of the subject.

The physical condition determination system according to claim 1, further comprising a transmission unit that transmits a determination result of the physical condition of the subject by the analysis determination unit, and transmits the determination result to a care / nursing / medical / welfare organization.