Print

Print


We (www.mdlogix.com) have just been awarded funding from NIDA for below
2-year project ("A Tool for Network Research on HIV Among Drug Users"), and
are looking for a PhD or postdoctoral level person with social network
expertise interested in joining us fulltime.

I will be at Sunbelt meeting in Cancun if anybody would like to meet with me.

Allen Tien

A.      SPECIFIC AIMS

Phase I focused on the problem of linking members of network samples in
large or unbounded populations, such as drug abusers in communities (or
members of large corporations).  A software program ("SocioMetrica
LinkAlyzer") that matches and merges records from egocentric nomination
data to construct sociometric data (containing more information on links
between population members) was created.  LinkAlyzer uses attributes such
as gender, age, physical features, and behaviors to determine if different
records represent the same actual person (for analysis and for
intervention) or the same type of person (for making population
inferences).  In Phase I the LinkAlyzer program was used to process a HIV
drug abuse data set with 15,000 records and 100 million potential matches,
generating a network model.  Analytic results based upon that model suggest
that matching records on a range of characteristics that represent the same
types of person is sufficient for inferences about social networks.  The
public health value of LinkAlyzer is substantial, because while it is
relatively easy and inexpensive to gather egocentric nomination data (there
are many existing data sets of this type), it is more difficult and costly
to obtain or generate sociometric data.  LinkAlyzer is a tool that enables
users to more easily apply sociometric methods to egocentric data,
increasing their capability to use network models to make population
inferences and address public health or other problems.

Phase II will create and further validate a SocioMetrica package of
applications.  Assessment interfaces and functions will be added to assist
in gathering personal/egocentric data from subjects.  Under the General
Public Licence, code for the estimation of an expanded set of social
network measures will be implemented.  In this engineering process, a
Common Data Model and platform will be created that makes it easier for
social network researchers work with different programs and to implement
their own network measures and algorithms.  Due to the computational
demands of LinkAlyzer matching functions and estimating certain social
network measures with large datasets, an Internet architecture is planned
to provide users with optional (and highly secure) access to low-cost
processing power from servers with large amounts of RAM and multiple
CPUs.  This Web services architecture will also simplify the business of
software distribution, maintenance, and user support, and will facilitate
value-added multi-site project operations and management.  Iterative
testing and validation will be done with consultants and potential
customers using existing projects and data sets.  The Specific Aims of
Phase II are:

       1.        To create, test, and validate user interfaces for directly
gathering network data;

       2.        To further validate, extend, and refine LinkAlyzer operations;

       3.        To create a platform that will facilitate the development
and use of network estimation functions;

       4.        To implement an Internet distributed client-server
architecture;

       5.        To implement multi-site project and data management and
reporting functions.


=======================================

Allen Y. Tien, MD, MHS
President and Research Director
Medical Decision Logic, Inc.
724 Dulaney Valley Road, Suite 2
Towson, MD 21204

&

Clinical Associate Professor
Department of Psychiatry
West Virginia University
Morgantown, WV

web site address: http://www.mdlogix.com

tel: 410-828-8948 or 410-821-5618
fax: 410-828-8948