us.graph.md - concept

The goal of problem-solving is to instantiate this model, which in reality derives from the knowledge of each specific case, obtained through the agent's actions and the assumptions contained in the domain and task models. From this point of view, problem-solving is no longer a process of input and output (io), nor a means of selecting actions (as in Newell's theory of the level of knowledge), nor is it a process of transforming the model (as in actuation components).

The level of knowledge provides the means to "rationalize" the behavior of a system from the point of view of an external observer. It is based on the following model of the behavior of an agent: the agent possesses knowledge, and part of this knowledge constitutes the agent's goals. The agent can perform a set of actions and choose the actions according to the principle of rationality.

The old generation of expert systems reflects in its structure the specific framework of use during which expert knowledge leads to appropriate behavior for a limited number of problem situations. A knowledge system is useful to the extent that it contributes to the construction of a meaningful process model, such as the result of an interaction between agents, systems, and the world.

Notions such as domain model, task, problem-solving method, mechanism, inference, or role of knowledge, provide additional structure to the concept of knowledge. The main points cited as a basis indicate that the level of knowledge is a model for rationalizing observed behavior. The models of level of knowledge used in the investigation of these systems are specific points of view, based on their use and respective level. The perspective of knowledge continues to increase in importance. It is considered today as an essential tool in modern life, but, according to leading researchers, some of the issues should be addressed in more depth by people who use a complex and refined technology such as this.

high level abstraction

The increase in the amount of diagnostic information has required clinicians to make a huge effort to work with a complex set of relationships. The idea behind using graph.md is to help relate this basic information to provide specialists with relevant information. In this sense, process mapping aims to create protocols to facilitate data analysis and allow the identification of information related to each diagnosis. Knowledge is formalized through conceptual maps. The automated analysis of information banks (images, exams, records, and clinical data) through this model allows the degree of relationship between a specific context of a patient and the most relevant possibilities, proposing a standardization and simplification of the diagnostic process. The goal is to help people keep their medical care records up to date and, at the same time, contribute to the development of new technologies applied to practical solutions. This scenario allows the provision of pattern recognition resources and the application of protocols to support medical diagnosis defined by specialists in the knowledge base. The knowledge synthesis model was developed by graph.md and is available to healthcare institutions and other areas that are interested in entering the world of super apps.

To this end, we have created collective spaces within the platform for knowledge integration and construction (Synthesis), based on the ideas of Nonaka (the world's leading expert on Knowledge Management and creator of the knowledge managers of the future project at ICS University in Japan (https://www.ics.hub.hit-u.ac.jp).

Considering the significant increase in Graph DBs expected for the coming years, the expected impacts cannot yet be estimated, but they will certainly be very positive for companies that invest in this transformation.

Without a doubt, graph.md, through the knowledge synthesis project, will be focused on the high technological level of this new scenario.

We are currently focused on developing new features for the graph.md super app.

Blockchain is an important technology used to develop the digital signatures part that provides support for the authenticity of information stored in knowledge bases.

We intend to make a new version of the solution available annually. However, it should be noted that this project is under continuous development (for life).

Low level abstraction

In the conceptual model of the synthesis knowledge base, four main structures are defined: Maps, Concepts, Patients, and Diagnoses.

According to the structure, the maps are associated with the possibilities of information RelationshipMapConcept present in the process of a conventional consultation between the doctor and the patient, for example, anamnesis, as the anamnesis process occurs and the doctor identifies that a concept should be associated with the patient (e.g.: symptom). This concept is directly related to the patient through the PatientRelationshipConcept class. The proposed intelligent agent model has the function of identifying when the same set of relationships exists in the two classes: PatientRelationshipConcept and DiagnosisRelationshipConcept. When this similarity exists, a context is identified. This diagnostic context can trigger an action or consultation. The selection of diagnoses associated with patients who have the same symptoms (concepts) as the patient being diagnosed allows the selection of probable concepts that the patient under analysis may have. The goal now is to ratify or exclude these alternatives, precisely by the concepts that do or do not associate with the suggested possibilities.

It is important to emphasize that a specialist, in the diagnostic process, may not yet be sure of what is affecting the patient, but he or she may already know what the patient does not have. In this sense, two types of relationships between concepts and patients are made available: positive, which assertively associates the patient with the concept; and negative, which does not explicitly associate the concept in question, that is, it states that the patient does not have a certain concept. For the model, the importance is to be able to filter out what is not relevant and show only what should be shown. This idea is associated with the parameters of precision and response of a query (Recall and Precision). For explanatory purposes, these two concepts can also be associated with the idea of false positive and false negative. Establishing the importance at the time of associating a certain symptom with the patient. The possibilities would be:

(a) the patient does have a symptom related to him or her; correct association

(b) the patient has a symptom that is not related to him or her; non-association error

(d) the patient does not have a symptom that is not related to him/her; correct non-association

It is observed that the diagnostic process is carried out by the specialist doctor based on the patient's anamnesis and test results. The importance of the scientific methodology in collecting information is fundamental, as explained above, for the usability of an intelligent system that must be non-imperative, notifying the user through messages, alerting possible details that may have gone unnoticed, but that does not induce the user's decision, only helping him/her to reach a more efficient diagnosis. In this sense, a significant part of the effort to build the diagnostic model must be focused on validating the consistency of the data entered, seeking to minimize the occurrences of states (b) and (c).

get in touch

contact@graph.md to get more information on graph md points of view.

Av. João Pessoa, 731 - Sala 802 - Cep: 90.040-000 - Porto Alegre, Brazil