us.graph.md - doctor

The focus of graph.md's knowledge synthesis project is to build a truly relevant tool to support a diagnosis. This technology aims to integrate all of a patient's medical records, providing the possibility of analyzing information for screening, remote care, and/or referral to specialized treatment centers, through virtual care, allowing for a detailed investigation of the case. Another possibility is monitoring by intelligent agents, automatically indicating the condition and the need for follow-up by the doctor himself. In this context, the technology allows the transformation of clinical information into diagnostic knowledge, based on the vision of the specialist doctor to create more efficient diagnostic protocols in different areas of medicine.

Based on the identification of the current context, the idea is to create interactive communication protocol models for each recognized diagnostic group and associated with those in progress. Each interaction carried out by the doctor can be inserted into the standard protocol, which can be automatic or require authorization for sending.

Another possibility is the recognition of interactive patterns. When the AI identifies a standard behavior in the interaction log that is not in the protocol, a request will be triggered for its inclusion. After the physician accepts, this new pattern enters the radar of events to be applied due to the similarity of context, as well as all others previously included in the communication model in question. In this way, we will have a self-construction protocol specific to each situation. Returning to the collaborative scenario, the protocols may not only be part of the physician specifically, but they can also be shared respecting the acceptance of the patient who owns the medical record. Thus, in graph.md, the physician has the benefit of receiving help in a new dynamic for him that other colleagues have already experienced. [doctor e1 audiobook]

All PATIENT data is assigned by doctors who treat several other patients in different clinics and hospitals. But shouldn't a person's clinical data be in their possession? On the other hand, different behaviors cause symptoms and diseases over time. Likewise, care over time generates health benefits. Different experiences generate different levels of complexity for each person. Even things that are supposedly simple for most people can be extremely complex: such as hydrating an athlete in search of an Olympic medal, or feeding a newborn, where minutes can make the difference between life and death.

When a person thinks about eating, it is because they are hungry. They do not think about the physiological impact on their body, just as we do not think about words, our intelligence does not allow us to think about the synapses necessary to generate words. Whenever possible, we observe at the highest level of abstraction, guided by the law of least effort. But when we experience some difficulty, we descend to the lowest levels to try to understand the problem and look for a solution, for example, because we lack energy, or because we are feeling some discomfort, or pain in some part of the body.

The PATIENT is propagated among all the doctors who have treated him throughout his life. But now this same PATIENT can centralize all his information that is not his but belongs to the doctors who have analyzed him. Observing this aspect is essential to perceive the reduction in the complexity of the system and at the same time maintain the integrity of the data. The knowledge base of a person can be the same size as that of a specialist doctor. The patient's: it is large because it contains all the events of his life related to his health. The doctor's: because it contains all the events of patients who have had a certain disease at some point. On the other hand, one more patient in the knowledge base can change the results of the algorithms, like the last straw that makes the glass overflow. There is a need to synthesize, creating a specific knowledge base to perform the momentary diagnosis that collects all the necessary information from the patient and the specialists' maps, suggests a diagnostic hypothesis and is eliminated, without first returning the new and relevant information to the patient, recording this consultation, and in the same way, enabling the specialist to generate new analyses from these new clinical cases, which in the future may be part of the knowledge bases of other determined diagnoses. Now imagine, millions of patients, with thousands of doctors, within a scenario of 10 thousand cataloged diseases that list more than 5 thousand symptoms, it doesn't seem STRANGE. [doctor e2 audiobook]

Everything is recurrent, cycles are completed in different units of time: seconds, for facial responses in a conversation; minutes for the next breath, hours to sleep; days to eat; years to learn; and a lifetime to live. The complexity of the system cannot be measured, but the fact is that there is no possibility of providing quality medical diagnostic support with PATIENT data lost in institutional databases, where they are just a record of a set of tables and relationships. The Future of AI lies in edge prediction: BiTemporality is an important concept, we always need to know the patient's current state, this is the FIRST DIMENSION; we also need to know what their clinical condition was like previously, so the SECOND DIMENSION is a journey back in time. In this sense, with the time machine at our disposal, we can perform cross-sectional analyses to identify the evolution of a certain behavior or physical symptom. Due to the large volume of information and the high flow of data, the most complex algorithms are restricted to being applied previously at the time of clinical analysis. For this reason, it is possible to use the same temporal storage structure to place the results of the processing of these algorithms in tags directly in the knowledge bases, identifying behaviors from previous clinical cases. For example, we create a node to relate two or three symptoms (grouping) and link them directly to the disease with the respective probability. Therefore, if a patient presents these symptoms, we can perform a complementary query on the symptoms that would lead to a certain diagnostic hypothesis. With these intelligent filters, it is possible to create dynamic forms to navigate through 50 concepts within a universe of 50 thousand, with at least 500 thousand relationships. Even great specialists would not be able to verify all these possibilities due to the significant increase in technologies that generate more and more clinical information. It is no longer about knowledge or intelligence; the limit becomes the processing capacity of the human brain. Another aspect is that different situations generate different needs. Therefore, the more specific the diagnosis, the further away it will be from the most common symptoms. There is a limit to how much a patient knows about their situation; they can know about their fatigue. But you haven't studied your whole life to know that there is a congenital problem in your left ventricle without the support of a specialist. In this way, the diagnosis migrates some concepts from the PATIENT's knowledge base to the specialist's distinct universe with another 50 thousand concepts. Now imagine being called for a consultation because a specialist didn't like the results of some of your tests. Here we can leave the universe of the diagnosed disease and enter the LOOP of preventive health. [doctor e3 audiobook]

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