Upgrade to Pro — share decks privately, control downloads, hide ads and more …

感染症の数理モデル7

 感染症の数理モデル7

Daisuke Yoneoka

August 24, 2024
Tweet

More Decks by Daisuke Yoneoka

Other Decks in Research

Transcript

  1. ⽬次 1. 感染症のコンパートメントモデル 2. 基本再⽣産数 3. 最終流⾏規模 4. R実装 5.

    ⼈⼝の異質性とSIR 6. 再⽣産⽅程式とエボラ vs インフル 7. R0 の推定⽅法(流⾏初期) 8. 内的増殖率の検定 9. Effective distance 10. 分岐過程 (Branching process) 11. ⼤規模流⾏確率と⽔際対策 本書の内容をカバーします。 具体的なコードなどは右の本 詳細なプログラムなどは https://github.com/objornstad/epimdr/tree/ master/rcode (結構間違ってる。。。) 2/48
  2. 分岐過程 • Galton-Watson processの定式化が⼀番有名 を⾮負の整数値をとるiidな確率変数: (ただし ) 46 <latexit sha1_base64="GL0Z3QrIIO4C2/WhI0xbfQk3Ms4=">AAACeHicfVBNbxMxFHQWCiUUSOHIxRChtjSKdiu+LhUVcOCCKBJpo2bT1VvnJbXqj5X9FhGt9syv4Qq/hb/CCW+aSrRFPMnSaGbs55m8UNJTHP9qRdeur9y4uXqrfXvtzt17nfX7B96WTuBAWGXdMAePShockCSFw8Ih6FzhYX76ttEPv6Dz0prPNC9wrGFm5FQKoEBlnUdHWUXbSc13eepLnVVyN6mPq6OM6mFQerLOOt24Hy+GXwXJEnTZcvaz9dbrdGJFqdGQUOD9KIkLGlfgSAqFdTstPRYgTmGGowANaPTjapGl5k8CM+FT68IxxBfs3zcq0N7PdR6cGujEX9Yaspfrf8mjkqavxpU0RUloxNmuaak4Wd50wyfSoSA1DwCEk+G7XJyAA0GhwQuLmrfJWuVDmncYUjr8EKiPBTog655WKbiZlqYOqWdpr0H/M8LXc2NA7XaoPLlc8FVwsNNPXvSff3rW3XuzLH+VPWSP2SZL2Eu2x96zfTZggn1j39kP9rP1O+LRRrR1Zo1ayzsP2IWJdv4ALL/B1w==</latexit>

    Zt+1 = Zt X i=1 Xt,i <latexit sha1_base64="dnBFeKrJ/oDLZmTXmiBMLy/SMRU=">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</latexit> Xt,i <latexit sha1_base64="itvO7QQ8osiOx0s2uq1sUfmGpdI=">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</latexit> P(Xt,i = m) = pm Zt は世代tの感染者数 Xt,i は世代tの個体iが感染させた数 N世代後に感染者数はどうなるか?が⽬的 • その定義よりZはマルコフ連鎖(もっと⾔うと普通は状態0を吸 収状態とする吸収的マルコフ連鎖を仮定) • 世代tの感染者数Zt の期待値は、ある個体がうつす 感染数の期待値から <latexit sha1_base64="/QJrU3OHsX386wyBoDoLP9nW310=">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</latexit> p0 > 0, p0 + p1 < 1 <latexit sha1_base64="6p2Tnm+ggx9c59MScxarcZT1nDM=">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</latexit> E[Zt] = E[Xt,i]t
  3. ⼀⼈の感染者が⼊ってきたときの伝播リスク ⼀⼈が平均R0 ⼈にうつすとすると感染者数は 第t-1世代時の総感染者数は R0 <1なる感染症において、総感染者数の情報を得ることが感染性を知 るうえでとても⼤事 でも、Spatially heterogeneousなpopulationで これやると推定値にバイアスが⼊る

    (Birello et al. 2024) 47 <latexit sha1_base64="DnbtaWbc3n81nGFLZ0YCTE/7GZM=">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</latexit> 1, R0, R2 0 , R3 0 , . . . <latexit sha1_base64="I/0wJq8xSxF3i5ymZMMtAvaZW3k=">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</latexit> In ⌘ n 1 X t=0 Rt 0 = 1 Rn 0 1 R0 <latexit sha1_base64="qrynYPzU6iu7RXDTR3fx8npLUqI=">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</latexit> R0 ⇡ 1 1 In 今、MERSではR0 は⼩さ いので、R0 nはめっちゃ ⼩さいはず
  4. 伝播リスク (Cont.) 今、1感染者だけが輸⼊されたとする 最終規模は ⼀⼈の⼈が何⼈にうつすかの確率分布 (offspring dist)を負の⼆項分布 (Poisson-Gamma mixture:ポワソン分布の平均がガンマ分布に従う場合のMarginalが負の⼆項) メリット:⼤きい分散(∝1/k,つまりスーパスプレッダー)が表現可能

    最終規模の分布は以下 (Nishiura et al. (2012)) 48 <latexit sha1_base64="pm1TwQ0noTY5pZ5jkhIVALzNKM8=">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</latexit> P(Xt = x) = (b + x) x! (b) ✓ R0 R0 + b ◆x ✓ 1 + R0 b ◆ b <latexit sha1_base64="T/R4CEK8s0u5Y3UnW1hiAjHE71Q=">AAAC33icfVFdb9MwFHUyPkb56oA3BDJUSB3bqmQaHy8VE/DAC6Igug01XWQ7TurNiSPbQUSWn+EJ8cpP40/wG3DaTKIr4kqWjs4519f3GJecKR0Evzx/7cLFS5fXr3SuXrt+42Z349aBEpUkdEwEF/III0U5K+hYM83pUSkpyjGnh/j0VaMffqZSMVF81HVJpznKCpYygrSj4u63Uf/TsN6EQxglqUTERKUUSWxOhoE9NvXOro04TXU/mosn1mC7VUeSZTO9aU394ExeNGNrPsTBFrat49jgetnh5FVTvRPauNsLBsG84CoIW9ADbY3iDe9FlAhS5bTQhCOlJmFQ6qlBUjPCqe1ElaIlIqcooxMHC5RTNTXzyCx85JgEpkK6U2g4Z//uMChXqs6xc+ZIz9R5rSG3cf4veVLp9PnUsKKsNC3IYlZacagFbL4AJkxSonntACKSuedCMkMuHe0+amlQc7cWgiu3zWvqtpT0raPelVQiLeRjEyGZ5aywbuss2m7Q/4zoy5nRoU7HRR6eD3gVHOwOwqeDJ+/3evsv2/DXwV3wEPRBCJ6BffAGjMAYEPDbu+Pd8+77yP/qf/d/LKy+1/bcBkvl//wDnYjojg==</latexit> P(Y = y) = Qy 2 j=0 j b + y y! ✓ b R0 + b ◆by ✓ R0b R0 + b ◆y 1 <latexit sha1_base64="EQojc5Q800JgptYSdGmloPbTJsM=">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</latexit> Y = 1 X t=0 Zt
  5. 使い⽅ データが揃い、負の⼆項分布のパラメータが(最尤)推定できると Q1. 例えば、1例輸⼊されたときに、⼆次感染の発⽣確率は? Q2. ⼆次感染で終わる(絶滅)確率は? ( と書く) Q3. 総感染者数が8⼈以上になる確率は?

    49 <latexit sha1_base64="tmBDkU20ZGE8b/aXy0bJusbL1/c=">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</latexit> P(X = m) = pm 統計家向け: ここがXの確率⺟関数(PGF) の形になっていることに気づこう <latexit sha1_base64="zdM3ub26NHPLf88BSULhbww9BNg=">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</latexit> GY (s) = E[sY ] = 1 X y=0 p(y)sy <latexit sha1_base64="pm1TwQ0noTY5pZ5jkhIVALzNKM8=">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</latexit> P(Xt = x) = (b + x) x! (b) ✓ R0 R0 + b ◆x ✓ 1 + R0 b ◆ b <latexit sha1_base64="wE/DP7Md1drzpMZLwfps7BG5Hng=">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</latexit> P(Y 8) = 1 7 X k=1 P(Y = l) = 1 7 X k=1 Qy 2 j=0 j b + y y! ✓ b R0 + b ◆by ✓ R0b R0 + b ◆y 1 <latexit sha1_base64="+6WHC/trHMFVEDbf/7IJsjth4R8=">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</latexit> P(Z2 = 0|X0 = 1) = 1 X k=1 P(Z2 = 0|Z1 = k, X0 = 1)P(Z1 = k|X0 = 1) = 1 X k=1 {P(X = 0)}k pk = 1 X k=1 pk 0 pk <latexit sha1_base64="+XbQA5R3whGRkSGQ+iOxrL/sAjc=">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</latexit> P(X1 > 0) = 1 P(X1 = 0) = 1 ✓ 1 + R0 k ◆ b ⇡ 22.7% (19.3 25.1%)
  6. ⼤規模な流⾏が起こらない確率 (⼀⼈から出発した集団の)絶滅確率(⼤規模な流⾏が起こらない確率)をqとすると収束定理より COVID-19の場合,R=1.6, b = 0.1とすると,q = 0.9226くらい つまり,⼤規模流⾏が起こる確率は1-0.9226 =

    7.7%くらい. これは1⼈の感染者が⼊った場合.もしn(=10)⼈の感染者が⼊国したら? qnが絶滅確率になる(i.e., ⼤規模流⾏が起こる確率は,1- qn) ちなみに q =0.5533くらいなので,⼤規模流⾏が起こる確率は46.7%くらい 50 <latexit sha1_base64="dRL7+hGzi5v/Q9PRWYPhD5GVXiw=">AAACdHicfVBdSxtBFJ1sv2z6Fe1jLQwNgi0Sdktb+yKK+tCXUguNCtkY7k7uxsH5WGfuloQlT/6avtZf4x/x2dmYQtXSCwOHc8+9d87JCiU9xfFFI7p3/8HDRwuPm0+ePnv+orW4tO9t6QR2hVXWHWbgUUmDXZKk8LBwCDpTeJCd7NT9g5/ovLTmB00K7GsYGZlLARSoQev1Kd/gqS/1oBpvxNOjVJqcJrxYHb89PRoPWu24E8+K3wXJHLTZvPYGi43NdGhFqdGQUOB9L4kL6lfgSAqF02ZaeixAnMAIewEa0Oj71czHlK8EZshz68IzxGfs3xMVaO8nOgtKDXTsb/dqci3T/2r3Sso/9ytpipLQiOtbeak4WV7nwofSoSA1CQCEk+G7XByDA0EhvRuH6t1krfLBzS4Glw6/BupbgQ7IundVCm6kpZkG16N0rUb/E8L4jzCgZjNEntwO+C7Yf99JPnU+fv/Q3tqeh7/AXrE3bJUlbJ1tsS9sj3WZYGfsF/vNzhuX0XLUjlaupVFjPvOS3aiocwUQyb/2</latexit> q = 1 X x=0 p(x)qx ←の右辺は,Xの負の二項分布の確率母関数に なっていることに注意 Def of PGF <latexit sha1_base64="oGKFVnKDzth2ADr8GO0GSBF6Lec=">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</latexit> f(q) = E[qX ] = 1 X x=0 p(x)qx 収束定理の応用 <latexit sha1_base64="jx1xtdvnjIysvk0hVNKdgYd9clU=">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</latexit> q = f(q) <latexit sha1_base64="lScAOhK4RqzZsUttEJcF41e/VRk=">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</latexit> q = P(9n 1; Zn = 0|X0 = 1) <latexit sha1_base64="SPvzUfp7usuqeWoIT4jsHuNUGMQ=">AAACgXicfVBdTxNBFJ0uKFC/ij7yMqExaRHrLgE0MUQiPPhiRGOBpFub2end7YT5WGbuGprN/gd/Da/4N/w3zpaaCBhvMsnJuefeO+ckuRQOw/BXI1hYvHd/aXml+eDho8dPWqtPj50pLIc+N9LY04Q5kEJDHwVKOM0tMJVIOEnODur+yXewThj9Fac5DBXLtEgFZ+ipUWvjnO7RWEKKnXicWsbLqCqjF19GYSd6ed59lVSxFdkEu9+SUasd9sJZ0bsgmoM2mdfRaLXxLh4bXijQyCVzbhCFOQ5LZlFwCVUzLhzkjJ+xDAYeaqbADcuZqYo+98yYpsb6p5HO2L8nSqacm6rEKxXDibvdq8nNRP2rPSgwfTMshc4LBM2vb6WFpGhoHRIdCwsc5dQDxq3w36V8wnw66KO8cajejcZI590cgndp4aOnPuVgGRq7UcbMZkroyrvO4s0a/U/ILv4IPWo2feTR7YDvguOtXrTb2/m83d5/Pw9/mayRddIhEXlN9skHckT6hJMf5JJckZ/BQtANwmDrWho05jPPyI0K3v4GRyfDBQ==</latexit> q = ✓ 1 1 + R0(1 q)/b ◆b
  7. ⽔際対策 これまでの計算は⽔際対策何も無しの場合(⾃然状態) 実際は⼊国制限などの対策があるよね ⼊国者数:N そのうちの感染割合:b ⽔際対策の効果 (⼊国リスクの相対的減少) :a 51 <latexit

    sha1_base64="ihbGBmTVqGay20h70nDaLwuaK5w=">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</latexit> n = (1 a)bN ⽔際対策をすり抜けて市中感染に寄与してしまう⼈数 ⼤規模流⾏が起こる確率は <latexit sha1_base64="iwsas9R9Zkvsvu33cZVN14FhKtI=">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</latexit> 1 qn = 1 q(1 a)bN ←何%くらい感染者が⼊るのを防げるか? ここに⾊々モデルを⼊れることで拡張可能