A Markov Chain Model for the Dynamics of Cognitive Decline in an Elderly USA Population Using Data from the Health and Retirement Study

By Gerardo Soto-Campos, Andrew Pavelyev and Carlos A. Soto.

Published by The International Journal of Aging and Society

Format Price
Article: Print $US10.00
Article: Electronic $US5.00

Using ten waves of the longitudinal data in the Health and Retirement Study (HRS), we built a Markov chain to describe the dynamics of cognitive decline in an USA elderly population. The cognitive states are defined in terms of thresholds of a composite of four cognition tests in HRS scaled to produce a score ranging between 0 and 35, and defined as a total cognitive function (TCF). The Markov chain consists of four states describing no cognitive impairment, NCI; mild cognitive impairment, MCI; and severe cognitive impairment, SCI. The absorbing state dead, D, is introduced in this model via the Tracker file in HRS. A practical use of the Markov chain is shown by computing the unadjusted life expectancy of people with severe cognitive decline and comparing it with independent studies for people with Alzheimer’s disease. The Markov chain estimations fall within the range of the reported values.

Keywords: Cognitive Decline, Markov Chains, Transition Rates, Life Expectancy

The International Journal of Aging and Society, Volume 2, Issue 3, pp.9-21. Article: Print (Spiral Bound). Article: Electronic (PDF File; 358.483KB).

Dr. Gerardo Soto-Campos

President and Founder, Modeling Health, LLC., Culver City, CA, USA

Gerardo Soto-Campos holds a PhD in physical chemistry from the University of Oregon. He did postdoctoral studies in chemical physics at UCLA in Los Angeles, and published several scientific papers in the general area of statistical physics. In 1999, he began to work in the medical industry and has held several positions varying in degree of complexity such as mathematical modeler, senior scientist, director of validations, and head of the diabetes modeling team at a research company based in San Francisco, CA. After June 2010, the emphasis of his former company changed from research to Internet services. He then decided to create a scientific consulting start up in April 2011, Modeling Health, LLC. He has worked on cardiovascular and metabolic modeling for many years, but lately he has been almost exclusively engaged in cognitive modeling, since he believes studies on cognitive decline and dementia will become crucially important not only in terms of research and public health, but also equally important in terms of the financial and sociological impact that the baby boomer will place on Medicare and other tax-paid services in the US.

Dr. Andrew Pavelyev

Independent Consultant, Modeling Health, LLC., Charlotte, North Carolina, USA

Dr. Andrew Pavelyev holds a MS in physics from the internationally renowned Moscow State University and a PhD in physics from the University of South Carolina. After finishing his PhD, he worked as a programmer-analyst at Policy Management Systems Corporation in South Carolina. He also designed sophisticated software solutions in the banking industry in Charlotte, NC. From 2001 to 2005 he worked as a scientist for Kaiser Permanente, a health care organization headquartered in Oakland, CA. Among his diverse functions in this position, he developed complex software architectures to model health care systems by computer. From 2006 to 2010, he joined Archimedes, Inc., a company based in San Francisco, CA, where he was in charge of maintaining the integrity of all the computer software used to run projects for external clients. He also played a key role in developing the statistics of diverse populations used in Archimedes. Since 2010, he has been an independent consultant, and in the last twelve months he has been a consultant with Modeling Health, LLC., a California startup specializing in creating solutions to health care problems based on data-driven models and computer simulations.

Prof. Carlos A. Soto

Associate Professor of Physics, Center for Research in Mathematics, Hidalgo State University, Pachuca, Hidalgo, Mexico

Professor Carlos A. Soto holds MS and PhD degrees in physics from the Center for Research and Advanced Studies of the National Polytechnic Institute in Mexico City. He has worked on different areas of theoretical physics. In 2007 he joined the scientific staff at the Center for Research in Mathematics at the Hidalgo State University in Mexico. He is currently an associate professor of physics at the Hidalgo State University, where he is also in charge of coordinating the undergraduate program in physics. Since 2011, his scientific interests have expanded to include Markov processes in statistics and their applications in multistate models.