せん妄の診断は精神科医や看護師による観察により行われており、生物学的指標に基づく診断は確立されていない。本研究の目的は簡易型脳波計測器を用いて取得した脳波によりせん妄時脳波の特異な脳波パターンを特定し、さらに得られた脳波パターンのみを用いて，せん妄の生物学的指標に基づく診断を確立する．
男女合計40名の脊椎手術を受けた慶應義塾大学病院整形外科入院患者に対し，簡易脳波計(Mind Wave Mobile II BMD)による脳波計測を行った．脳波計測は、閉眼状態で数を3分間のカウントアップ，その後3分間の安静閉眼の２種類の条件で行なった。測定回数は手術前1回，手術後最大3回とし，各回間で24時間以上の間隔を空けた．せん妄の有無はせん妄評価尺度(Confusion Assessment Method:CAM)により判定を行った。脳波解析には各タスク20秒毎の脳波に対して、周波数解析による正規化パワースペクトルの算出により定量化を行なった。つぎに，Wilcoxonの順位和検定(有意水準: 0.01)を行いタスク毎に有意差のある周波数を検出し，ロジスティック回帰によるせん妄陽性群と陰性群の分類を行った．
有意差検定を行った結果，カウントアップ時に7, 8, 12~15, 17, 30Hz,安静閉眼時に1~5, 7, 8, 13, 14Hzの有意差が確認された．次に有意差が確認された周波数を用いて2クラス分類の為ロジスティック回帰を行った結果，カウントアップ時の7~8Hz, 12~15Hz及び安静閉眼の7~8Hz, 13~14Hzのパワースペクトル和を用いた回帰がせん妄判定の最適モデルとなった．また，せん妄患者の脳波はα波，β波でパワーが増強されていることが確認できた．
The purpose of this study is to identify the characteristics of electroencephalogram when delirium appears using a simple electroencephalograph. In addition, delirium is identified using only the obtained EEG patterns. Although delirium has been subjectively screened using CAM, low sensitivity remains a major problem.
In this study, a simple electroencephalograph (Mind Wave Mobile II BMD) was used to evaluate delirium by CAM in a total of 40 patients undergoing spine surgery in Keio University Hospital who underwent spinal surgery. The number was counted for 3 minutes with the eyes closed, and then EEGs were obtained during 3 minutes with the eyes closed at rest. The number of measurements was one before surgery and a maximum of three after surgery. Measurements after surgery were spaced at least 24 hours each.
For the analysis, a normalized power spectrum was calculated by frequency analysis for EEG every 20 seconds for each task. Next, Wilcoxon rank sum test (significance level: 0.01) was performed to detect frequencies with significant differences for each task, and the delirium-positive group and the negative group were classified by logistic regression. As a result of the significance test, significant differences of 7, 8, 12 to 15, 17, and 30 Hz were observed at the time of counting up and 1 to 5, 7, 8, 13, and 14 Hz at the time of resting eyes closed. Next, logistic regression was performed for the two-class classification using the frequencies for which significant differences were confirmed. Regression using was the optimal model for delirium judgment. It was also confirmed that the power of the EEG of the delirium patients was enhanced by α and β wave.