Statistical Modeling of Drug Half Lives
Advisor(s)
Dr. Tarek Mahfouz
Confirmation
1
Document Type
Poster
Location
ONU McIntosh Center; Activities Room
Start Date
24-4-2026 10:00 AM
End Date
24-4-2026 10:50 AM
Abstract
A drug’s elimination half-life (t ½ ) is defined as the time it takes for the plasma concentration of its pharmacologically active form to decline by 50%. It, therefore, is the factor to consider when determining the frequency of dosing. Drugs with long t ½ remain active for extended periods of time which corresponds to less frequent dosing and enhanced patient compliance. Accurately estimating the t ½ of drug candidates is, therefore, important in the drug design process because it helps reduce experimental cost and time by identifying unsuitable candidates early and focusing the search on those with reasonable duration of action. This research aims to develop a mathematical model that can accurately predict a drug’s t ½ using only structural information. Experimentally determined t ½ data were collected from drug databases for 98 approved drugs belonging to different therapeutic groups and spanning a broad range of t ½ , from short (≤2 hours) to long (≥100 hours). The dataset was divided into a training set for model development and optimization and a test set for external validation. A comprehensive set of structural descriptors was calculated for all drugs. For the training set, different groups of descriptors were incorporated into linear and non-linear modelling approaches to identify relationships between molecular features and t ½ . Statistical analyses were applied to identify the best performing models and these were used to calculate t ½ values for the test set. Computed t ½ values for the test set were then compared to the experimental values and the best performing models are presented.
Recommended Citation
Donovan, Sean Keith and Harman, Braden William, "Statistical Modeling of Drug Half Lives" (2026). ONU Student Research Colloquium. 23.
https://digitalcommons.onu.edu/student_research_colloquium/2026/Posters/23
Open Access
Available to all.
Statistical Modeling of Drug Half Lives
ONU McIntosh Center; Activities Room
A drug’s elimination half-life (t ½ ) is defined as the time it takes for the plasma concentration of its pharmacologically active form to decline by 50%. It, therefore, is the factor to consider when determining the frequency of dosing. Drugs with long t ½ remain active for extended periods of time which corresponds to less frequent dosing and enhanced patient compliance. Accurately estimating the t ½ of drug candidates is, therefore, important in the drug design process because it helps reduce experimental cost and time by identifying unsuitable candidates early and focusing the search on those with reasonable duration of action. This research aims to develop a mathematical model that can accurately predict a drug’s t ½ using only structural information. Experimentally determined t ½ data were collected from drug databases for 98 approved drugs belonging to different therapeutic groups and spanning a broad range of t ½ , from short (≤2 hours) to long (≥100 hours). The dataset was divided into a training set for model development and optimization and a test set for external validation. A comprehensive set of structural descriptors was calculated for all drugs. For the training set, different groups of descriptors were incorporated into linear and non-linear modelling approaches to identify relationships between molecular features and t ½ . Statistical analyses were applied to identify the best performing models and these were used to calculate t ½ values for the test set. Computed t ½ values for the test set were then compared to the experimental values and the best performing models are presented.