Diese Präsentation wurde erfolgreich gemeldet.
Die SlideShare-Präsentation wird heruntergeladen. ×

A data driven nomogram for breast cancer survival

Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Anzeige
Nächste SlideShare
Genetics of Breast Cancer
Genetics of Breast Cancer
Wird geladen in …3
×

Hier ansehen

1 von 14 Anzeige
Anzeige

Weitere Verwandte Inhalte

Diashows für Sie (20)

Ähnlich wie A data driven nomogram for breast cancer survival (20)

Anzeige

A data driven nomogram for breast cancer survival

  1. 1. A Data-Driven Nomogram for Breast Cancer Survival Capstone by Lisa Federer, Bret Gaulin, Michal Haskell, Andy Pollack, and Carmen Smiley
  2. 2. The Problem Image source:
  3. 3. 1 in 8 US women will be diagnosed with breast cancer during their lifetime.
  4. 4. Image source: http://www.pathophys.org/wp-content/uploads/2012/12/breastcancer-copy.png
  5. 5. Our hypothesis Certain demographic, diagnostic, and treatment parameters can reliably predict survival time for women with breast cancer. Using this knowledge, we could build a “calculator” to estimate survival time for individuals.
  6. 6. The data http://seer.cancer.gov/
  7. 7. Data exploration and variable selection • Survival time (months) • Age at diagnosis • Year of birth • Race • Origin (Hispanic recode) • Stage • Histology • Tumor extent • Number of primary tumors • Laterality • ER Status • PR Status • Radiation therapy 146 variables in SEER database 13 variables of interest
  8. 8. Classification by survival time
  9. 9. Classification model results Model Precision Recall f-score Support Logistic Regression 0 0.65 0.43 0.23 146 1 0.55 0.37 0.44 416 2 0.55 0.76 0.63 471 Average 0.56 0.56 0.54 1033 Naïve Bayes 0 0.39 0.16 0.23 162 1 0.63 0.3 0.4 402 2 0.5 0.84 0.63 466 Average 0.54 0.84 0.63 1030 Decision Tree 0 0.3 0.39 0.34 131 1 0.76 0.75 0.75 416 2 0.8 0.75 0.78 485 Average 0.72 0.7 0.71 1032 Random Forest 0 0.58 0.43 0.5 183 1 0.73 0.75 0.74 414 2 0.71 0.78 0.74 440 Average 0.7 0.7 0.7 1037 K Neighbors 0 0.53 0.33 0.41 150 1 0.83 0.82 0.82 425 2 0.77 0.88 0.82 472 Average 0.76 0.77 0.76 1047
  10. 10. Regression and survival models Cox proportional hazards model 𝜆 𝑡 = 𝑏0 𝑡 exp(𝑏1 𝑥1 + ⋯ + 𝑏 𝑁 𝑥 𝑛) Aalen’s additive model 𝜆 𝑡 = 𝑏0 𝑡 + 𝑏1 𝑡 𝑥1 + ⋯ + 𝑏 𝑁 (𝑡)𝑥 𝑇)
  11. 11. Nomogram demonstration
  12. 12. Limitations Not a substitute for medical advice
  13. 13. Limitations Missing or incomplete data in SEER database
  14. 14. Questions?

×