Drug Treatment Effect Model Based on MODWT and Hawkes Self-Exciting Point Process.

Area of Science:

• Pharmacodynamics
• Biomedical data analysis
• Time series modeling

Background:

• Precision medicine requires robust long-term drug effect monitoring.
• Current models lack robustness for instant drug treatments.
• Accurate modeling is crucial for assessing drug benefits and side effects.

Purpose of the Study:

• To develop a time series model for monitoring drug effects in precision medicine.
• To capture complex drug behaviors and time-lagged effects.
• To improve the robustness of pharmacodynamic assessments.

Main Methods:

• Utilized a sin function for baseline heart rate trends.
• Employed a Hawkes self-exciting point process model for drug effects.
• Incorporated cumulative Gamma distribution for time lag effects.
• Applied Maximal Overlap Discrete Wavelet Transformation for signal decomposition.
• Analyzed real-world heart rate and drug (Liquemin) data.

Main Results:

• The model effectively described baseline trends and drug effects with low noise.
• A selected scale variable (s4) showed significant cumulative drug effect characterization (Pearson correlation = 0.22).
• The model demonstrated accuracy in real-world data analysis.

Conclusions:

• The developed Hawkes process model accurately characterizes cumulative drug effects.
• This approach enhances pharmacodynamic monitoring in precision medicine.
• Future work can extend the model to multiple variables and drug types.