Assessing the Adequacy of a Prediction Model

Authors

Abhaya Indrayan, Max Healthcare, New DelhiFollow
Sakshi Mishra Ms, Max HealthcareFollow

Comments

Key Points:

1. Many prediction models claim to have good predictive performance but that may not be so in most cases.
2. An analysis of the recent 20 articles on prediction models indicates that the current method of assessing predictive accuracy is primarily area under the ROC curve, supported by methods such as calibration plot and decision curve analysis. Most of these methods are inadequate to assess the predictive performance of the models.

We discuss the inadequacy of these methods for assessing the predictive performance and provide remedying suggestions

Disciplines

Biostatistics | Medical Biomathematics and Biometrics

Suggested Citation

Indrayan, Abhaya and Mishra, Sakshi Ms, "Assessing the Adequacy of a Prediction Model" (September 2024). COBRA Preprint Series. Working Paper 123.
https://biostats.bepress.com/cobra/art123