Quantitative Analytics for Biomedical and Other Complex Data

Your Subtitle text
Descipher LLC - Analytic Dynamics Home Page


OVERVIEW

We have over two decades of experience in clinical science, pharmaceuticals, health information technology, and modeling of clinical measurements. With the "new" electronic medical record (EMR) on the horizon, it may be possible to put more hard science into the practice of medicine.  It is certainly possible to improve clinical trials right now in drug development and academic clinical research.
  • Pathodynamics

  • This involves the modeling of dynamic patterns reflected in clinical measurements over time.   Models are built using applied mathematics such as stochastic differential equations to represent the biology and statistical methods to estimate the parameters and random components of the models.  Sometimes just the display of dynamic data in motion reveals very informative patterns.

    This approach is intended to provide a top-down view of the biological system as opposed to the bottom-up genomic approach.  Our aim is to make clinical medicine more quantitative and therefore more scientific through dynamic modeling.  We have developed algorithms for pathodynamic modeling and data display.

  • Biomedical Signal Processing

  • There are two fundamental types of continuous-type clinical signals:  Those that are measured sporadically from tissue or fluid samples like blood and those that are measured repeatedly over short time intervals such as 24-hour electrocardiograms (ECG/EKG) and other electrophysiological signals.  The latter can also be sporadically measured such as a heart rate or QT-interval measured over several seconds and then repeated monthly or annually.

    Discrete-type clinical signals are also abundant.   These can be found in vital signs or physical examination findings such a cough or rash being present or absent.  Clinical adverse events such as those collected in drug clinical trials to monitor drug safety are typically discrete.

    Pathodynamics approaches or standard signal processing methods may apply here.  We have developed algorithms for analyzing the dynamics of 24-hour Holter (ECG) QT intervals.
  • Computer-aided Diagnostics

  • For decades computer scientists and physicians have been using so-called artificial intelligence (AI) techniques in attempts to create diagnostic software.   For the most part, it has not worked.  This failure has been partly due to the complexity of the subject matter, the medical domain, and partly due to the lack of good electronic medical records.

    We have developed an algorithm that matches medical test results with diseases. It is designed to overcome many of the drawbacks of the standard medical expert systems or AI approaches.
Website Builder