Adeniji, Abidemi Kassim
(2012)
Incorporating Diagnostic Accuracy into the Estimation of Discrete Survival Function.
Doctoral Dissertation, University of Pittsburgh.
(Unpublished)
This is the latest version of this item.
Abstract
The Empirical distribution function (EDF) is a commonly used estimator of the population cumulative distribution function. The Survival function is estimated as the complement of the EDF. However, the clinical diagnosis of an event is often subject to misclassification, by which the event is assessed with some uncertainty. In the presence of such errors, the true distribution of the time to first event is unknown. We develop a method to estimate the true survival distribution by incorporating negative predictive values (NPV) and positive predictive values (PPV), which are assumed to be known, into a product-limit style construction of a survival function. This allows us to quantify the bias of the EDF that do not account for misclassification due to the presence of misclassified events in the observed data. We present an unbiased estimator of the true survival function and its variance. In addition to dealing with misclassified clinical outcomes, this dissertation addresses survival function estimates in the presence of misclassified and incomplete data. The product limit (KM) estimator is commonly used to estimate the survival function when follow-up time is incomplete due to drop-outs. Typically this method is employed assuming that the outcome is known with certainty. We develop a method to estimate the true survival distribution by incorporating the NPV and PPV into a Kaplan-Meier-like construction. This allows us to quantify the bias in the KM survival estimates due to the presence of misclassified events in the observed data. We present an unbiased estimator of the true survival function and its variance. Asymptotic properties of the proposed estimators are provided and these properties are examined through simulations. We demonstrate our methods using data from the VIRAHEP-C study.
Estimating the true distribution of time to an event such as time to symptom resolution among subgroups of population with certain characteristics is of public health importance. When the event is measured with error, the actual distribution cannot be estimated without bias, providing an inaccurate picture of the population. The new methods provide clinical investigators with a tool to accurately estimate the survival probabilities in the presence of misclassified events.
Share
Citation/Export: |
|
Social Networking: |
|
Details
Item Type: |
University of Pittsburgh ETD
|
Status: |
Unpublished |
Creators/Authors: |
Creators | Email | Pitt Username | ORCID |
---|
Adeniji, Abidemi Kassim | abk38@pitt.edu | ABK38 | |
|
ETD Committee: |
|
Date: |
24 September 2012 |
Date Type: |
Completion |
Defense Date: |
26 July 2012 |
Approval Date: |
24 September 2012 |
Submission Date: |
23 July 2012 |
Access Restriction: |
2 year -- Restrict access to University of Pittsburgh for a period of 2 years. |
Number of Pages: |
54 |
Institution: |
University of Pittsburgh |
Schools and Programs: |
School of Public Health > Biostatistics |
Degree: |
PhD - Doctor of Philosophy |
Thesis Type: |
Doctoral Dissertation |
Refereed: |
Yes |
Uncontrolled Keywords: |
Misclassification, Measurement error, Diagnostic testing, Product limit estimation, Generalized estimating equations, Binary classification |
Date Deposited: |
24 Sep 2012 15:46 |
Last Modified: |
15 Nov 2016 14:00 |
URI: |
http://d-scholarship.pitt.edu/id/eprint/13049 |
Available Versions of this Item
-
Incorporating Diagnostic Accuracy into the Estimation of Discrete Survival Function. (deposited 24 Sep 2012 15:46)
[Currently Displayed]
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
|
View Item |