System and method for parallelizing grid search method facilitating determination of PK-PD parameters

Patent metadata

Field	Value
Publication number	US-9891958-B2
Application number	US-201514663968-A
Country	US
Kind code	B2
Filing date	Mar 20, 2015
Priority date	Mar 21, 2014
Publication date	Feb 13, 2018
Grant date	Feb 13, 2018

Abstract

Disclosed is a system and method for parallelizing grid search technique facilitating determination of PK-PD parameters. The method may comprise determining number of grids. The method may further comprise creating grid points based upon the number of grids (N) and a number of parameters (p). The method may further comprise distributing the grid points amongst number of threads. The method may further comprise evaluating an objective function value corresponding to each grid point in order to compute an objective function value associated with each of the grid points. Further, the method may comprise identifying a grid point having minimum objective function value. The grid point having the least objective function value may indicate the estimated initial PK-PD parameters.

First claim

We claim:

1. A method for searching Pharmacokinetic (PK)-Pharmacodynamic (PD) parameters in a parameter space, wherein the method comprising:
receiving, by a processor, a number of grids (N) induced over a parameter space, number of parameters (p), lower bound value (LBi) and upper bound value (UBi) of each parameter;
determining, by the processor, total number of grid points (n) based upon the number of grids (N) and the number of parameters (p), wherein each grid point is separated from an adjacent grid point based upon a step size, and wherein each of the total number of grid points (n) is represented by a set of coordinates;
computing, by the processor, plurality of grid points corresponding to the set of coordinates associated therewith based upon the lower bound value (LBi), the upper bound value (UBi), and the number of grids (N);
employing a parallelized grid search method based on the set of co-ordinates and initial PK-PD parameters to generate a plurality of final estimates for a parameter space;
executing a parallel implementation of a grid search to automatically assign one or more iterations of the parallelized grid search method based on a specified scheduling; and
parallelly computing, by the processor, an objective function value for each grid point of the plurality of grid points, wherein a grid point having a minimum objective function value indicates the PK-PD parameters.

2. The method of claim 1 , wherein the number of grids (N) determined, is user-defined or a predefined number, based on the range of the parameters.

3. The method of claim 1 , wherein the step size is determined by (UBi−LBi)/(N+1), wherein the UBi and LBi are upper bound value and lower bound value of parameter i, and N is the number of grids.

4. The method of claim 1 , wherein the co-ordinate values is computed by using following equation: LBi+(ri+1) (UBi−LBi)/(N+1), wherein the ri=0,1,2,3, N−1 and indicates co-ordinate values of the grid points.

5. The method of claim 1 , wherein the PK-PD parameters are searched based on a model selected from a group comprising of a single compartmental model, a multi-compartmental model, an algebraic model, and a differential equation model.

6. A system for searching Pharmacokinetic (PK)-Pharmacodynamic (PD) parameters in a parameter space, the system comprises:
a processor;
a memory coupled to the processor, wherein the processor executes a plurality of modules stored in the memory, and wherein the plurality of modules comprises:
a receiving module to receive a number of grids (N) induced over a parameter space, number of parameters (p), lower bound value (LBi) and upper bound value (UBi) of each parameter;
a determining module to determine total number of grid points (n) based upon the number of grids (N) and the number of parameters (p), wherein each grid point is separated from an adjacent grid point based upon a step size, and wherein each of the total number of grid points (n) is represented by a set of coordinates; and
a computing module to:
compute plurality of grid points corresponding to the set of coordinates associated therewith based upon the lower bound value (LBi), the upper bound value (UBi), and the number of grids (N),
employ a parallelized grid search method based on the set of co-ordinates and initial PK-PD parameters to generate a plurality of final estimates for a parameter space;
execute a parallel implementation of a grid search for automatically assigning one or more iterations of the parallelized grid search method based on a specified scheduling; and
parallelly compute an objective function value for each grid point of the plurality of grid points, wherein a grid point of the plurality of grid points having a minimum objective function value indicates the PK-PD parameters.

7. The system of claim 6 , wherein the number of grids (N) determined, is user-defined or a predefined number, based on range of the parameters.

8. The system of claim 6 , wherein the step size is determined by (UBi−LBi)/(N+1), wherein the UBi and LBi are upper bound value and lower bound value of parameter i, and N is the number of grids.

9. The system of claim 6 , wherein the co-ordinates is computed by using following equation: LBi+(ri+1) (UBi−LBi)/(N+1), wherein the ri=0, 1, 2, 3, N−1.

10. A non-transitory computer readable medium embodying a program executable in a computing device for searching Pharmacokinetic (PK)-Pharmacodynamic (PD) parameters in a parameter space, the program comprising:
a program code for receiving a number of grids (N) induced over a parameter space, number of parameters (p), lower bound value (LBi) and upper bound value (UBi) of each parameter;
a program code for determining total number of grid points (n) based upon the number of grids (N) and the number of parameters (p), wherein each grid point is separated from an adjacent grid point based upon a step size, and wherein each of the total number of grid points (n) is represented by a set of coordinates;
a program code for computing plurality of grid points corresponding to the set of coordinates associated therewith based upon the lower bound value (LBi), the upper bound value (UBi), and the number of grids (N);
a program code for employing a parallelized grid search method based on the set of co-ordinates and initial PK-PD parameters to generate a plurality of final estimates for a parameter space;
a program code for executing a parallel implementation of a grid search to automatically assign one or more iterations of the parallelized grid search method based on a specified scheduling; and
a program code for parallelly computing an objective function value for each grid point of the plurality of grid points, wherein a grid point having a minimum objective function value indicates the PK-PD parameters.

11. The method of claim 1 , wherein the objective function value is computed by using following equation:

WSRS
=

∑

i
=
1

#
⁢

⁢
Observations

⁢

⁢

w
i

⁡

(

y
observed

(
i
)

-

y
predicted

(
i
)

)

2

wherein y observed (i) is observed value of dependent variable, and y predicted (i) is predicted value of the dependent variable.

12. The system of claim 6 , wherein the objective function value is c

Assignees

• Tata Consultancy Services Ltd

Inventors

• Agrawal Nishant Kumar
• Narayanan R
• Nambiar Manoj Karunakaran
• Abdulrazak Rihab
• Pandey Ambuj
• Das Shyam Sundar

Classifications

• G06F9/5044 Primary

  considering hardware capabilities · CPC title

• G06F19/709

  Physics · mapped topic

• G06F17/30864

  Physics · mapped topic

• G06F17/30867

  Physics · mapped topic

• G06F19/704

  Physics · mapped topic