Automated data import/export, validation, review & preparation

Integration with multiple commercial and open source applications on different platforms (Windows or Linux), used for data collection, analysis and reporting

Regulatory Compliant with audit trailed tracking of data, analysis and reports

Integration capability with multiple analysis (NCA, POP-PK, Simulations, etc.) and reporting tools (R, SAS, NONMEM, etc.)

Facilitate data and workflow standardization

Supports cloud based computations.

For more information, product demonstration and evaluation; please visit our website or contact us any time.

Sincerely,
The MF-Pharm Team

After several months of development efforts, the MashFrame team is pleased to announce the release of MashFrame Pharmacology Suite (MF-Pharm) 2.0.

For more information, product demonstration and evaluation; please visit our website or contact us any time.

Sincerely,

Harjeet Singh Arora
Managing Director / CEO
Rudraya Corporation
www.rudraya.com

Rudraya Corporation, Plainsboro, NJ
Global Pharmacometrics, Pfizer Primary Care Business Unit, New London, CT

Introduction:
Success of a Meta-Analysis depends on quick, efficient and standardized data: 1) Collection, 2) Validation, 3) Preparation and 3) Exploration. Typically a Pharmacometrician will use combination of comma separated spreadsheets and tools like R1 with lots of manual time consuming efforts for cleaning and preparation of the data. We report herein on a tool that integrates and automates data validation, preparation, and exploration steps for a model-based meta-analysis of Young Maniac Rating Scale (YMRS2) in bipolar patients, which can easily be adapted and reused for any other Meta-Analysis.

Objectives:
To develop interactive and modular framework to automate the workflow for validation, preparation and exploration of any kind of literature data. Illustrate an application using this framework for preparing data for model-based meta-analysis of YMRS in bipolar patients.

Methods:
Data Collection and Extraction: Articles published in bibliographic databases were searched to identify clinical trials using specific criteria such as appropriate patient population, inclusion/exclusion criteria, etc. with YMRS as the efficacy endpoint. Data was then extracted from literature using data specifications defined in an internally developed Literature Information and Knowledge Explorer (LIKE) repository application.

Data Preparation: Raw data extracted from articles was received as a Comma Separated Value (CSV) file with two header rows. This file was first uploaded to MashFrame Cabinet (a shared network repository of files) for further use within the MashFrame Framer application. The previously indentified manual workflow was decomposed into individual steps to clean, prepare and summarize the data to be used for the meta-analysis. These steps are summarized in Table 1. Each step was performed as a self contained and independent software module called a MashObject.

Table 1: Steps needed to clean, validate and prepare data for Meta-Analysis

Each object in the workflow were joined to create a MashFrame by using open standards based Extensible Markup Language called MashFrame XML (MF-XML). Each object was designed to provide flexibility, automation; reusability and interactive user interface. A MashFrame with details of few MashObjects is summarized in Figure 1.

Figure 1: MashFrame with few screenshots of user interfaces and outputs

The MashObjects can be re-arranged/re-ordered in any manner by simply dragging and dropping to a frame to meet any specific need. The MashFrame Engine was used to interpret and process MF-XML associated with each MashObject by distributing individual step actions to three MashFrame Process Servers. Figure 3 shows the high-level architecture of MashFrame Engine.

Exploring the Data: By using a combination of the Steps (i.e. MashObjects) listed in Table 1, for each study, available YMRS vs. time or dose data were identified and extracted. Longitudinal response data was obtained from twenty five studies representing 5498 subjects treated with the drugs of interest. MashObjects implementing R Visualizations, through the statistical programming language R4, were used to create charts. These MashObjects allowed pharmacometricians to automatically create and update standardized R scripts by simply providing chart attributes via a form field data entry. Figure 1 shows some of the charts produced using this R Visualizations MashObject.

Preparing Nonlinear Mixed Effect Model software (NONMEM3) ready Data: Following an initial graphical exploration of the data, a NONMEM data MashObject was used to create dataset ready for further NONMEM analysis. This MashObject automatically validated and created a data file which can be directly used by NONMEM software.

Data Sharing: Data and Visualizations were shared with the team by automatically creating MashFrame Widgets using the MashFrame Views application. These MashFrame Widgets can be rendered on desktops or even mobile devices in real time. Figure 1 shows some of the MashFrame Widgets displayed on a pharmacometrician’s desktop.

Results:
Steps needed to collect, validate, prepare and explore Meta-Analysis data of YMRS meta-data in Bipolar Patients were converted to interactive, modular and reusable software objects called MashObjects. These MashObjects were linked together to generate a integrated MashFrame developing a reusable automated framework to: Fix data Errors; Generate Unique ID’s; Normalize Time, Dose and Response Values; Calculate Arm; Generate Visualizations; and Data Validation Reports. Final cleaned data and numerical and graphical summaries were then made available to users via Desktop Widgets enabling rapid deployment and sharing of results with the project team. Flexibility in using MashFrame and Objects allowed rapid reconfiguration for changes in understanding of the underlying data. Some of the key benefits of workflow automation tool like MashFrame are summarized in figure 2.

Figure 2: MashFrame Benefits

Figure 3: MashFrame Engine High Level Architecture

Conclusions:
MashObjects created for Bipolar YMRS data could be easily re-arranged and re-used to clean, validate and prepare Meta-Analysis dataset from the ~ 700 indications in the LIKE data repository. This kind of open standards based modular approach allows end users to reuse objects for future analysis without going through a time consuming and potentially expensive software development lifecycle. Once a particular step is structured as an XML object ,it can be used, maintained or updated by researchers, without requiring any IT resource. Familiar modeling and simulation tools such as R, NONMEM, etc. can be directly integrated as individual MashFrame objects. These MashObjects can be deployed on distributed computing systems. MashFrame Widgets based approach to rapid reporting of results provides enhanced collaboration within research teams.

References:
[1] R Development Core Team (2006). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org.
[2] Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for mania: reliability, validity and sensitivity. Br J Psychiatry. 1978;133:429-435
[3] Beal, et al., NONMEM Users Guides: Parts I-VIII. NONMEM Project Group: University of California, San Francisco, 1998.
[4] http://www.r-project.org