Intelligent autoverification

Why use an expert system?

Valab® is an expert system used to assist the biological validation of medical test results. An expert system reproduces the reasoning of a human expert, reasoning which is based on choices taking into account the largest possible number of parameters.

When a decision is based on one or two criteria only, the reasoning is relatively easy to master. To evaluate the coherence and/or the acceptability of a medical test result however, multiple criteria come into play: the age and sex of the patient, his/her state of health, if he/she is treated in a hospital or at home, the previous results obtained for the same test, etc.

The reasoning rapidly becomes extremely complex. The assessment of the coherence / acceptability of the result must also take into account the results of all the other tests in the report. A classic algorithmic system used to assess the coherence / acceptability of a result will quickly be overwhelmed by so many combinatorial choices, only an expert system like Valab can handle such complexity.

Valab helps where and how?

In a typical laboratory process, the test results go through steps of technical validation and biological validation on the screen before they are released. Valab steps in after the technical validation step and before biological validation on the screen.

The LIS sends each report to Valab for biological verification just before the step of biological validation on the screen. Valab assesses the coherence and/or acceptability of each result of the report taking into account not only the demographic and contextual (medical and technical) parameters of the result itself, but also the influence of the other results in the report (inter-parametric verification) and their anteriority and kinetics if available.

In less than one second, Valab returns to the LIS the results of its verification and an indication on the validity of the report. The reports comprising a set of results considered to be coherent and/or acceptable by Valab can then be authorised for release. The reports containing results considered to be incoherent or inacceptable by Valab will be examined more closely on the validation screens of the laboratory to understand the origin of this incoherence / inacceptability.

The examination on the screen of the reports containing results judged to be incoherent or inacceptable by Valab is greatly facilitated as the results concerned are indicated by special indicators (or "flags"). In this way, Valab allows to perform an initial filtering of the reports considered a priori to be validatable and clearly shows up the incriminated results in the reports containing incoherent or inacceptable values.

The following schematic shows the implementation logic of Valab in the laboratory:

Specialities and tests

The standard version of Valab supports four specialities covering the basic tests of the laboratory:

  • biochemistry
  • hematology
  • coagulation
  • blood gases

Additional tests, fully modeled and ready to use, can be added to the basic tests. These additional tests complete the range of basic specialities or belong to other specialities: markers, hormones, serology, therapeutic drugs, toxicology, urinalysis . . .

If your need for specialities and tests is not covered by the basic or additional specialities and tests, you can create new ones yourself by using the Auto-Expert module provided with Valab.

Other types of tests cover specific needs. Duplicate tests allow to exploit the results of a test obtained under different technical conditions, linked tests inherit the verification result assigned to another test, and blind tests allow the systematic validation of certain values (temperature, weight, …) if required

Operating modes and integrated tools

Autoverify mode is the main function of Valab. In this mode, Valab receives and automatically verifies the reports sent by the LIS. Through the Valab interface, the verification results can be displayed in real-time on the Valab workstation.

Simulation mode allows to analyse with total safety the verification results of the reports processed without compromising the real results sent to the LIS. It is possible in this way to study the influence of the verification rules applied by Valab (positive and negative influences). SImulation mode is also used during the development of new expertise models created with the Auto-Expert module.

The Auto-Expert module allows to create new specialities and expertise models for new tests. Many tests already created with Auto-Expert and ready to use are available for download on this site.

Consultative mode (mandatory validation)

In normal operation, Valab indicates by "flags" only the non-validated results in the verified reports. However the laboratory may want to filter certain results "to be seen systematically" even if their inherent coherence would have allowed their validation.

A global "mandatory validation" option allows to mark all the results validated by Valab as being "to be seen by a specialist ". This option allows global monitoring of the operation of Valab, after first-time installation for example.

This "mandatory validation" mechanism also allows to perform individual filtering at different levels according to specific criteria:

  • at the individual test level - each result corresponding to the tests concerned will be flagged
  • at the request originator level - all the requests coming from the same prescriber or from the same department will be flagged
  • at the therapeutic and clinical information level - for example, all the reports concerning patients affected with diabetes can be systematically flagged
  • at the complementary information level - for example, all the results obtained in a context of "venous blood" can be systematically flagged

Global consultative mode and analysis of the Valab operating statistics are used to optimise Valab after the initial installation.

Safety and traceability

Safety and traceability are ensured by the creation of user accounts and tracking of the modifications made. As a complement to the traceability of the LIS, the communication history and verification results can be archived by Valab for a given period of time.

Communicating with the LIS through a serial link or network connection, any communication errors are analysed and categorised by severity level.

Activity statistics indicate the rejection rates per test and allow to monitor the operation of Valab and make optimisation adjustments.

Ergonomics and functionality

Valab benefits from many setup options allowing to respond to the needs of the users. The Valab interface is available in French and in English (modifiable by a simple menu option), other languages will be added later.

Valab handles the reception of complete and partial reports in cumulative / non-cumulative mode and many options complete its functionality (check on the calculations associated with differential requests and blood counts, ratio check for the results of duplicate tests, . . .).

A robust and flexible client-server architecture

Valab 11.01 is based on a client-server architecture allowing operation that combines both robustness and flexibility of use. Setup and monitoring of operation are performed through the Valab interface (the client) but the expertise engine (the server) can work independently of this interface. Valab can now run as a Windows® service, allowing automatic restart in the event of shutdown / restart of the work session.

The multilab option allows to connect several LIS's or databases to the Valab workstation. Simultaneous multi-user access is possible thanks to the remote control option.

Whether performed directly on the Valab workstation or from another workstation in remote control, setup of the autoverification parameters is possible while Valab is working in production. Simulation mode and the Auto-Expert module can also be used while Valab is working in production.