Related Experiment Videos
Methods to estimate the optimal threshold for normally or log-normally distributed biological tests
Jérôme Jund1, Muriel Rabilloud, Martine Wallon
1Department of Biostatistics, Hospices Civils de Lyon, Lyon, France. dim.jjund@ch-annecy.fr
Summary
Determining optimal screening test thresholds is crucial for clinical utility. New methods provide accurate threshold estimation for normally distributed data, particularly for diagnosing congenital toxoplasmosis.
Area of Science:
- Biostatistics
- Medical Diagnostics
- Epidemiology
Background:
- Establishing optimal thresholds for screening and diagnostic tests is essential for their effective clinical implementation.
- Test performance is influenced by the distribution of results in healthy and diseased populations.
Purpose of the Study:
- To present methods for estimating optimal screening or diagnostic test thresholds.
- To maximize population utility by determining thresholds for normally or log-normally distributed test results.
Main Methods:
- The study proposes point and interval estimation methods for test thresholds.
- Methods were evaluated using simulation for bias, coverage probability, symmetry, and confidence interval width.
- Assumptions include normally or log-normally distributed data with equal variances in healthy and diseased groups.
Main Results:
- The proposed methods are asymptotically nonbiased with satisfactory coverage probability for sample sizes >= 50.
- Optimal threshold determination for antibody load in congenital toxoplasmosis diagnosis was demonstrated.
- The methods are easy to implement with few constraints.
Conclusions:
- The presented methods offer a robust approach to determining optimal test thresholds.
- Accurate sample size determination is critical for achieving desired precision.
- These methods can enhance the clinical utility of diagnostic tests, as shown in congenital toxoplasmosis screening.