Medical data analysis

AltraBio provides the complete range of data analysis services relevant to medical data:

  • machine learning methods;
  • data quality audit, data-mining;
  • power estimation, survival estimation, log rank test, Cox model, complex statistical models;
  • database design and implementation.

Characterization of patient populations is of a key importance for evaluation of treatment response as well as for establishment of novel diagnostic and predictive criteria. In the "classical" data analysis approach, the goal is to identify variables with values that are significantly different between treatment and control groups. This constitutes an important constraint as, under such hypothesis, both treatment and control groups are considered to be homogeneous.

In contrast, AltraBio's initial analysis does not rely on the information on patient (sample) assignment (i.e., the “a priori” knowledge of the population structure). This enables characterization of subpopulations within phenotypically defined groups (e.g., within a group of drug non-responders or within a group of patients sharing a diagnosis based on histological or cytogenetic criteria). In addition, panels of biomarkers defining such subpopulations can be identified. Application of this principle can lead to identification of new diagnostic and predictive patient classes that are distinct from previously established categories and are based on understanding of the molecular mechanisms of pathology or drug response.

More information...