Implementation of a Decision Support System for Interpretation of Laboratory Tests for Patients

Ilya SEMENOV a  and Georgy KOPANITSAb,c,[1]

a Helix laboratory service, Saint-Petersburg, Russia

b Institute Cybernetic Center, Tomsk Polytechnic University, Tomsk, Russia

cTomsk State University for Architecture and Building, Tomsk, Russia

Abstract. The paper presents the results of the development and implementation of an expert system that automatically generates doctors’ letters based on the results of laboratory tests. Medical knowledge is expressed using a first order predicate based language. The system was implemented and evaluated in the Helix laboratory service.

Keywords. Decision support, Laboratory information system, telemedicine, first order predicates


In Russia many patients address laboratory services directly without a doctor’s referral. This causes the problem of interpretation of laboratory test results by the patients without a proper medical background.  So the patients require that the laboratory services provide not only the results of the tests but also their interpretation.  Automated decision support systems that have proved their efficiency for doctors can be a good solution for this problem [1]. The experience of development and implementation of decision support systems for doctors show the efficiency of such solutions for the doctors. However, developers face problems when it comes to the decision support for patients. They require different approach in data presentation and interpretation [2; 3].

The goal of this paper is to present a research and development of a decision support system for the patients of a laboratory service.

To achieve this goal we have developed a decision support system that solves a classification problem and defines the following parameters based on the results of laboratory tests:

  • Diagnosis (group of diagnoses)
  • Recommendations to run other laboratory tests
  • Recommendation to refer to a specialist doctor

1. Methods

To achieve the described above goal a decision support system must solve a classification problem where we must associate a vector of test results to a set of diagnoses and find a set of recommendations associated with this diagnosis. To organize a communication between the system and an expert we have implemented a knowledge representation language (KRL) that is based on the first order predicate logic[4].

After the knowledge representation language was implemented we have developed an interface to allow experts to fill in the knowledge base. For the pilot project to test the feasibility of the approach we have chosen a limited set of laboratory tests that could be interpreted by the system.  We have invited 3 laboratory doctors and 3 specialists (gynaecologist, urologist and general practitioner) to fill in the system’s knowledge base.

On the next step we have developed a classification algorithm that has the following possible outcomes:

Found a set of diagnosis that can be associated to the results of the laboratory test

No diagnosis found

Found a set of diagnosis, but the system requires extra test or vital signs to chose the proper diagnosis form this set.

The knowledge representation language, knowledge base and the classification algorithm were developed as a Doctor Ease decision support system, which was implemented in the Helix laboratory service. After the system has been implemented we made a qualitative research to evaluate the acceptance of the system among the patients with 100 participants. 

2. Results

The developed decision support system consists of the following modules:

  • Data base;
  • Data extraction system
  • Knowledge base;
  • Inference engine;
  • Knowledge base editor;
  • Explanation system
  • Results generator

A structural scheme of the system is presented in the figure 1.

 Схема ЭС

Figure 1. Structural scheme of the decision support system

Data base with a dynamical structure is made to store facts (test results) and intermediate inference results. Facts are taken from a laboratory information system (LIS).

Knowledge base of the DoctorEase system is made to store expert knowledge and inference rules.

Inference engine uses facts form the DB and knowledge and rules form the knowledge base to solve the classification task.

Knowledge base editor allows experts do define new knowledge and rules.

Explanation system analyses the sequence of the rules’ application to explain how the system achieved the result.

The decision support system has two main use cases: knowledge acquisition and decision support.  Knowledge acquisition mode allows defining inference rules, which are complex objects and each of them add its element to the resulting inference provided that the application of the conditions of the rule to the fact returns a positive value.  The knowledge is defined as a test result and its reference value associated to a set of diagnosis.

In the decision support mode the system generates recommendations applying a set of knowledge and rules to the facts that are derived form a LIS data base.

DoctorEase decision support system allows crating queries in the language that is closed to natural. The knowledge representation language is based on the first order logic and the predicates and relationships have meaningful names in Russian so the experts can define knowledge and rules using the terminology they are used to.

Knowledge base organization

The structure of the knowledge base of the system is presented in the figure 2.

объектная модель эс

Figure 2. Object model of DoctorEase

On the first step we define a configuration of the laboratory test, which is a complex object consisting of the parameters that are sufficient to make an inference. The configuration consists of laboratory test and inference rules.

A direct rule is an object that is defined for each parameter of a laboratory test along with the conditions for processing these parameters.

Each rule has a list of exclusion rules, which can exclude direct rules from the inference provided that their conditions are true.

Laboratory test is a template that consists of laboratory tests’ components. For example a Complete blood count consists of 22 components.

Laboratory tests are grouped into “orders”, which are commercial units that can be ordered by the patients.

Each rule has a set of conditions that work with comparison operators: =, <>, includes (>=  or =<), excludes (>=  and =<).

Conditions are associated with each other by logical operators and, or and not.

Inference process

After the system has received a notification that the laboratory test results are available it starts the inference according to the following algorithm:

1.    Patient’s order is analyzed to find if there exists a configuration for this order.

2.    Fact (test results) are loaded to the decision support system’s data base

3.    The inference engine defines a sequence of rules from the knowledge base to be applied to the facts

4.    Exclusion rules are applied to the facts to exclude non valid rules from the inference

5.    Result blocks are added to the result file according to the rules’ sequence


The system was implemented in the Helix laboratory service in Saint-Petersburg, Russia. At the moment it generates about 3500 reports a day. The implementation of the system allowed increasing the number of patients who refer to a doctor after laboratory tests. A qualitative study with 100 patients demonstrated a high acceptance of the system. The majority (82%) of the patients reported that they trust the system and follow its advice to visit a doctor if necessary.

3. Discussion

The paper presents a process of development and implementation of a decision support system for laboratory service patients. The system allows patients reading and understanding medical records in natural language. For the laboratory service the system allowed increasing the level of satisfaction of the patients and the number of patients who came back to the laboratory service for more detailed testing.

Current research is focused on the extension of the knowledge representation language by adding an ability to work with fuzzy sets [5]. This will provide experts with flexibility when they define knowledge and rules. We also are studying the possibility to validate the reports that are produced by DoctorEase to enable the system acquiring knowledge based on its experience.

4. References

[1]           L. Ahmadian, M. van Engen-Verheul, F. Bakhshi-Raiez, N. Peek, R. Cornet, and N.F. de Keizer, The role of standardized data and terminological systems in computerized clinical decision support systems: literature review and survey, Int J Med Inform 80 (2011), 81-93.

[2]           G. Kopanitsa, Standard based multiclient medical data visualization, Stud Health Technol Inform 180 (2012), 199-203.

[3]           G. Kopanitsa, Evaluation study for a multi-user oriented medical data visualization method, Stud Health Technol Inform 200 (2014), 158-160.

[4]           R.S. Michalski, Pattern recognition as rule-guided inductive inference, IEEE Trans Pattern Anal Mach Intell 2 (1980), 349-361.

[5]           K. Boegl, K.P. Adlassnig, Y. Hayashi, T.E. Rothenfluh, and H. Leitich, Knowledge acquisition in the fuzzy knowledge representation framework of a medical consultation system, Artif Intell Med 30 (2004), 1-26.

[1] Institute Cybernetic Center, Tomsk Polytechnic University, 634050 Lenina 30, Tomsk, Russia,

Share and Enjoy:
  • Добавить ВКонтакте заметку об этой странице
  • Мой Мир
  • Facebook
  • Twitter
  • LiveJournal
  • MySpace
  • В закладки Google
  • Google Buzz
  • Яндекс.Закладки
  • LinkedIn
  • Technorati
  • Digg
  • MisterWong.RU
  • Сто закладок