Biopathological validation of laboratory results using Artificial intelligence techniques

The biopathological validation of laboratory result is defined as a cognitive process carried out by the Pathology Specialist Physician which is based not only on his/her knowledge and clinical experience, but also, on the technological means available and aims to provide the patient or his doctor, with a rigorous result with the highest quality, ensuring a level of excellence in healthcare provision.

The biopathological validation of a laboratory result is based on the following assumptions:

●       The laboratory result under evaluation by the Clinical Pathology Specialist was determined according to the procedures approved by the quality management system in force, with regards to the analytical equipment involved and the responsibilities of the various professionals intervening in the process;

●       The information generated by the analysis devices (alarms) was properly processed and, when applicable, is made available to the Pathology Specialist Physician, in order to be considered in the act of biopathological validation;

●       For each test/parameter ranges of values (maximum and minimum) or critical results are defined.  Thus, for each result that presents itself out of normality, the system signals the occurrence to the Clinical Pathology Specialist, through the biopathological validation panel;

●       The detection/confirmation of a critical value associated with a certain parameter, assumes that the Emergency Clinical Pathology Specialist is immediately contacted;

●       The detection/confirmation of a critical value associated with a certain parameter, in a patient without a clinical history or clinical context requires that the pathologist contacts the patient or the prescribing physician to gather more information about the clinical history, clinical context or other relevant factors;

●       When previously defined in the procedure, the validation of the result must be complemented with written information addressed to the prescriber.

During the act of biopathological validation of a laboratory result, the Clinical Pathology Specialist considers the following criteria or factors in his/her evaluation/ interpretation:

●       Reference values, critical values or criteria that requires confirmation associated with the laboratory parameter under validation, when applicable;

●       The alarms transmitted by the analytical equipments to the Laboratory Management and Information System (hereinafter referred to as LMIS) and associated with the determination of some parameters, such as, information related to the occurrence of haemolysis, lipemia, fibrin, clots, aggregations, etc .;

●       Factors that determine the need to repeat the test, carry out a complementary evaluation or further investigation;

●       Other technical-scientific criteria associated with the laboratory parameter or its method of determination published in the medical literature, defined by protocols with suitable entities or recommended by the company that commercializes the method of determination (e.g.: algorithms or specific grids for a given parameter);

●       Limitations of the analysis methods in question (false positives and negatives);

●       Patient Clinical history and clinical laboratorial context, when applicable;

●       History of previous results, when applicable;

●       Other conditions, whenever the Clinical Pathology Specialist decides so.

With the aim of virtually assist the Clinical Pathology Specialist in the act of biopathological validation of laboratory results, the innovative solution should be based on a software or computing matrix that through Artificial Intelligence techniques executes a biopathological validation, according to the technical criteria and procedures in force,  but devoid of errors or gaps inherent to the human observer.

The solution to be identified should also contain tools capable of analysing new patterns or identifying new correlations between variables, offering suggestions or alarms that contribute to a possible extension of the technical criteria for biopathological validation. It should also be able to suggest the most appropriate written information to be addressed to the doctor, according to the specific situation of the patient in question.

Finally, it should enable changes to be made to the standards and values contained in the current computer system, under the conditions defined in the company’s Quality Manual.

In short, and once accomplished the previously mentioned conditions, the solution should allow the result to be delivered to the patient/client without a human observer having to manually perform the act of biopathological validation, within the limits and technical conditions considered safe.

The seeker is a Portuguese company founded many decades ago that owns a network of laboratories specialized in clinical pathology, Genetics, and anatomical pathology. It has more than 50 specialists in clinical pathology and genetics and performs more than 60 thousand analysis per day corresponding to approximately 7000 patients per day.

The  Laboratory Management and Information System (LMIS) is an IT tool used to register and manage all the information related to the process of each patient who performs laboratory analysis either in the laboratory or at the collection points of the company in question. This system also manages the documentation inherent to each episode, namely, the emission of the patient identification tags for each analysis tube, sending of results to the patient (by e-mail or by SMS, printed or communicated to other computer platforms) and consulting and providing specific collection instructions for the technician and the patient.

On the other hand, this system works as a repository of all clinical information concerning the patient, namely the laboratory results of each analytical parameter analysed in the patient’s sample(s) and in the different episodes that occurred. It also has several management modules that allow the billing of the analysis to the various entities involved or the collection of various types of maps or statistical analyses, in order to support the management of the company.

Figure 1 intends to outline the transmission and communication of data between the various information systems existing in the company. According to the scheme, the LMIS has a module for attendance and registration of the patient – Attendance/Registration Module, where all the information concerning the patient is registered, such as the prescribing entity, the social security number, the tax number, the address, the analytical parameters in the prescription/credential, the type of biological product to be collected, among other information. The registration of this information allows the subsequent emission of the patient’s identification tags for each analysis tube, through the Label Emission Module.

All analysis tubes have an unique identification for each patient.

Most measuring devices are in network with LMIS.  Thus way and once the analytical results of the various measuring devices have been determined, the information is transmitted by network to LMIS. In cases where it is not possible to transmit analytical results in a fully automated way, the analytical results are entered manually into the LMIS, and are validated three times by different people when they are introduced.

Ranges of values (maximum and minimum) or critical results for each parameter are defined at LMIS.

The analytical results of each patient are grouped in a results presentation grid (Analytical Results Module) which gathers the information of each patient from the different measuring devices, according to the parameters analysed.

For each unusual result, the system signals the occurrence to the Clinical Pathology Specialist. The system also has some validation rules applied to the analytical results that are made available in this grid. Warnings (e.g.: value highlighted in red if outside the reference range) or alarms (e.g., alert message to the doctor) can be sent.

Thus, the act of biopathological validation of each laboratory result is performed by the Clinical Pathology Specialist based on the visualization of the results grid and considering all the technical criteria or factors previously mentioned.

On the other hand, the validation of the result may be complemented with written information addressed to the prescribing physician. Currently, there are already some messages that have been previously configured that are triggered either by conditions that are verified for their activation, or because the technician and / or the doctor places them through the manual introduction of acronyms which automatically lead to the appearance of the pre-defined message in full. In addition, the system allows messages to be written in the free field for each parameter being validated. However, it does not have the capacity to suggest to the technician or doctor, certain information that contributes to complement the analytical result, i.e., the initiative always comes from the human operator.

Besides the analytical parameters to be determined, the LMIS can transmit data (e.g., age and gender) to measuring devices. However, this dual control is not a feature of all measuring devices.

Thus, and considering that the company’s existing LMIS has the capacity to gather all the necessary data for further analysis with an innovative perspective, the solution to be identified may exist in two different scenarios:

Scenario A – maintain the current LMIS and interconnect the new technological tool to this system. As previously mentioned, the solution to be identified could be a ready-made software or a computing matrix that processes this information using artificial intelligence techniques.

Scenario B – replace the current LMIS with a ready-made technological solution. This should combine the functions of the current system and additionally have a biopathological validation module using artificial intelligence techniques.

Placing this challenge on Inocrowd’s technological platform will check if there are ready-made solutions already on the market covering all or only a portion of the desired laboratory parameters, but also if the existing marketed solutions have the versatility to be adjusted to all Seeker needs.

On the other hand, existing solutions may present different levels of efficiency and circuits regarding the transmission of information between the various measuring devices and the LMIS.

The main benefit is social and economic due to the reduction of human error and efficiency gains. Additionally, there are advantages in terms of the technical quality of the operations performed and the working conditions of pathologists.

The solution to be found should achieve the following objectives:

a)      Virtually assist the Clinical Pathology Specialist in the act of biopathological validation of laboratory results allowing the result to be delivered to the patient without a human observer having to perform this act manually providing the clinician with better management of his time;

b)     Improve the accuracy and quality of the biopathological validation of laboratory results using Artificial Intelligence techniques to eliminate the errors or gaps inherent in the human observer, identify new correlations between variables or suggest information to complement the biopathological validation act.

Requirements for Scenarios A and B:

●       Technological tool (software or matrix) intended for the biopathological validation act that uses Artificial Intelligence techniques to analyze and process laboratory results;

●       Enumeration of parameters and analytical methods included in or applicable to the new solution.

●       Detail of the rules, calculation methods, algorithms, mathematical models or periodicity of analysis applied to the different parameters evaluated;

●       Detail of the specific Artificial Intelligence techniques and the process inherent to the evolution and maturation of the algorithms (quality control), when applicable;

●       Bibliographic references associated with values, rules, calculations, algorithms included in the solution;

●       Detail of the rules associated with the parameterization of written information addressed to the doctor, according to each type of parameter;

●       Description of alerts management for the pathologist;

●       Justification of the analytical methods and parameters to which the solution does not apply or other limitations related to the proposed solution;

●       Detail of the communication form(s) used in the different stages of the process, the interconnection with the current LMIS (Scenario A) and the interconnections between the new solution and the existing equipment in the central laboratories and collection points;

●       Solution with the possibility of interconnection with technological tools or measuring devices that may be acquired in the future;

●       Solution with the versatility to adjust to the real needs of the Seeker, in particular, add or modify rules, pre-defined values, algorithms, etc. associated with certain analytical parameters;

●       The solution that does not change the procedures of operations and hierarchy relations already established;

●       The solution must detail the associated costs and justify them, as well as the intended conditions for their payment.

Specific requirements for Scenarios B:

●       A Ready-made technological solution that simultaneously includes the functionalities of the current LMIS complemented with a module for biopathological validation of laboratory results using artificial intelligence techniques, according to the requirements described above.

Exclusion Criteria:

●       Solution with high maintenance costs;

●       The solution that does not allow changing the values inherent to each parameter, adding, or adapting rules or adding new measuring devices.