PIDexpert-decision support system for primary immunodeficiencies

Crina Samarghitean1, Kati Iltanen2, Kirsi Varpa2 , Merja Helminen3, Martti Juhola2, Mauno Vihinen1,3

1Institute of Medical Technology, FI-33014 University of Tampere, Finland
2Department of Computer Sciences, University of Tampere, Finland
3Tampere University Hospital, Tampere, Finland


Background

PIDexpert is a medical expert system (MES) to assist physicians in diagnosis of primary immunodeficiencies (PID). MESs are computer programs developed to contain and use knowledge from human experts in order to attain high levels of performance in a problem area [1]. PIDs are often difficult to diagnose because of similar and overlapping symptoms in many disorders and patients go often undiagnosed. PIDs represent a challenge in their diagnosis and treatment [2]. PIDexpert is designed to aid in PID diagnosis. It can also be useful in the medical education of personnel or medical students and as a research tool.

System architecture

PIDexpert includes a knowledge base, a query base, an answer database, an inference engine, and a user interface (Fig. 1)

The system contains information about primary immunodeficiencies, the clinical and laboratory parameters used in PID diagnosis. The knowledge base is built using data and facts from e.g. Immunodeficiency Resource (IDR, http://bionf.uta.fi/idr) [3], IDdiagnostics (http://bioinf.uta.fi/IDdiagnostics) [4], IDbases [5], clinical guidelines, literature and medical experts. The query base contains questions involved in the diagnosis of PIDs, concerning symptoms and signs, medical and family history, clinical and laboratory findings. The inference engine is rule-based and uses scores to compute the best matched disease profiles using a pattern recognition method.

PIDexpert status at the moment

More than 100 parameters have been used to describe the diseases. The query base and the knowledge base are under development and contributions from PID experts are needed. New parameters will be added to increase the reliability of the system. Real patient cases will serve to test the system in evaluation and testing phase (Fig.2).

Impact for the community and future work

  • PIDexpert will be a useful tool for the diagnosis of PIDs and for medical education.
  • Automation of medical algorithms can reduce errors and facilitate better treatment protocols
  • The first version of PIDexpert will be released soon

Your help is needed!

Collaborators to help in evaluating and testing PIDexpert are welcome. Anyone interested please contact us at: Crina.Samarghitean@uta.fi

 



Fig. 2 Steps in development of PIDexpert and status at the moment


References:
1. Samarghitean C, Vihinen M: Medical expert systems. Curr Bioinf 2008, 3:56-65.
2. Ochs HD, Smith CIE, Puck JM: Primary immunodeficiency diseases, a molecular and genetic approach, 2nd edn. Oxford: Oxford University Press; 2007.
3. Samarghitean C, Väliaho J, Vihinen M: IDR knowledge base for primary immunodeficiencies. Immunome Res 2007, 3:6.
4. Samarghitean C, Väliaho J, Vihinen M: Online registry of genetic and clinical immunodeficiency diagnostic laboratories, IDdiagnostics. J Clin Immunol 2004, 24(1):53-61.
5. Piirilä H, Väliaho J, Vihinen M: Immunodeficiency mutation databases (IDbases). Hum Mutat 2006, 27(12):1200-1208.