This document discusses the relationships between various performance measures used to evaluate and compare risk prediction models, including the net reclassification improvement (NRI) and decision-analytic measures. It explains that the NRI at a risk threshold T is equal to the difference in sensitivity plus specificity (Youden index) between two models. Decision-analytic measures like net benefit and relative utility are preferable to NRI as they account for different costs of misclassifying patients with or without the outcome. The document also presents a case study comparing a ovarian tumor risk prediction model with and without the CA-125 tumor marker, finding a negative NRI but positive net benefit and relative utility at threshold T=5%.
This document discusses the relationships between various performance measures used to evaluate and compare risk prediction models, including the net reclassification improvement (NRI) and decision-analytic measures. It explains that the NRI at a risk threshold T is equal to the difference in sensitivity plus specificity (Youden index) between two models. Decision-analytic measures like net benefit and relative utility are preferable to NRI as they account for different costs of misclassifying patients with or without the outcome. The document also presents a case study comparing a ovarian tumor risk prediction model with and without the CA-125 tumor marker, finding a negative NRI but positive net benefit and relative utility at threshold T=5%.
2. 文献1
一般患者の再入院についての予測モデルの検討?
?
-Hospital Readmission in General Medicine Patients: A Prediction ModelOmar Hasan, David O. Meltzer, Shimon A. Shaykevich, Chaim M. Bell, Peter J. Kaboli, Andrew D. Auerbach,
Tosha B. Wetterneck, Vineet M. Arora, James Zhang, and Jeffrey L. Schnipper (2010).
Journal of General Internal Medicine 25(3), 211-219.
2
10. 文献2
プライマリケアにおける軽症から中程度のうつ病に関す
る持続的回復の確率の推定?-THREAD研究結果より-?
ーEstimating probability of sustained recovery from mild to moderate depression in
primary care: evidence from the THREAD studyー (2011)
Psychological Medicine 41, 141-150.
Dowrick, C.
C. Kendrick andFlach, M. Leese, J. Chatwin, R. Morriss, R. Peveler, M. Gabbay, R. Byng, M. Moore, A. Tylee,
T.
THREAD Study Group
10
18. 文献3?小児保健の専門家の予防的早期発見は
青年期以前における心理社会的問題を予測する
:TRAIL Study
-Early Findings of Preventive Child Healthcare Professionals Predict
Psychosocial Problems in Preadolescence: The TRAILS Study-
Merlijne Jaspers, Andrea F. de Winter, Gea de Meer, Roy E. Stewart, Frank C. Verhulst,
Johan Ormel, and Sijmen A. Reijneveld (2010)
Journal of Pediatrics, 157(2), 316-321.
17