 |
It defines the disease dynamics of patients with pseudonym identifiers. |
 |
Determines the optimal HCT treatment sequence for patients with pseudonymized identifiers. |
 |
High-level answers formed from the low-level data of the database based on the combined patient data, which support the filling of the EBMT forms provided by the partner clinic. |
 |
High-level answers formed from the low-level data of the database based on unified patient data, which support the clinical trials running at the partner clinic. |
 |
A visual overview of the history of some patients. |
 |
Relational database combined from various data sources of the patient care system. |
No, the project company (Poiesy BV) is registered in the Netherlands, founded by a company also registered in the Netherlands (Szegedi Holding BV).
The startup company is currently 100% owned by Szegedi Holding BV, which may change after the involvement of investors. According to our plans (to simplify the administration) only companies (not individuals) can be owners.
No, the partner hospital that provides its data for the development of the models can use Poiesy services free of charge (and help determine its effectiveness through testing). Other services provided by the data collection module (searchable database, automatic EBMT form filling, patient history overviews) will be available only to them.
Yes, the procedures are subject to uniform EU regulations. Therapeutic decisions are determined by legislation, and those are determined by international standards.
The structure of health records and the code bases used may differ from institution to institution (even when international code bases are used, e.g. ICD, SNOMed, LOINC). However, the EBMT forms we use as data input eliminate this heterogeneity, as this format is used uniformly in all European haemato-oncology institutions.
Primarily those that are necessary for the complete completion of the EBMT forms (both in terms of diagnostics, therapy and follow-up). Since we would like to examine whether there are any data beyond the EBMT forms that are important for prediction models, we will also analyze additional data. If necessary, we will request that the structure of the EBMT forms be expanded with this data.
No, because they were developed only in the partner institution, for the purpose of data collection. It is in the interest of the employees here to record medication and patient care data in this, since the hospital system is not suitable for this. Integrating these software with the IT systems of other institutions would place an excessive burden on us. In the future, perhaps…
The Poiesy API communicates with the hospital's HIS system: it receives the content of EBMT forms and returns disease dynamics characteristics, and it receives the therapy calendar and returns warnings or an optimized version. The integration step can be avoided by installing the Poiesy Desktop application within the hospital's closed network, which manages the logistics of EBMT forms and the surgical calendar.
There are several options for this:
This habit needs to be changed (it is the only one). As soon as new test results are obtained or a new therapeutic intervention has been performed, it must be recorded on the forms and the prediction calculation must be run again. The frequency of updates is the decision and responsibility of the user hospital.
Rapidly changing parameters (laboratory values, vital parameters) can be requested in addition to the EBMT form structure if necessary. These are mostly well-structured data, not dependent on the data-providing laboratory or country.
The Hungarian partner company producing the anonymized data (Bioinformat-X): has ethical permission for the patient data of the South Pest Central Hospital and the National Institute of Hematology (DPC) between 2025.06.01 and 2027.12.31.
Supervisor: Gergely Kriván MD PhD
Participants:
In connection with molecular genetic tests, a consent statement is requested from all patients, in which they can separately stipulate the handling of their data and samples, their use in research and their (anonymized) transfer to third parties. In the project, we only use the data of patients with appropriate consent.
Sensitive patient data is not disclosed to Poiesy employees. The Hungarian partner company (Bioinformat-X) encounters sensitive data when creating the anonymized database, but even there only doctors (who are also IT specialists) can see this. The in-hospital services of Poiesy's data collection module naturally allow for patient identification, but this data does not leave the hospital server.
In general, patients are treated in the order they arrive. This can be modified by the duration of pre-treatment (conditioning) and finding a suitable donor. The order thus established can be overridden by an unexpected, urgent deterioration in condition: in this case, a patient in serious condition is moved to the front of the queue. Poiesy helps to find these patients so that their emergency care can be planned before their condition deteriorates. On the other hand, it helps to find those times in the surgical calendar that do not endanger patients who have been placed in the back in this way.
We use our own developed programs for data collection, mapping, and conversions. We use AI and NLP models to encode long text descriptions into structured data. We generate our prediction models with our own developed neural networks and classical statistical or big-data methods. The procedure for optimizing the patient care sequence is performed by our quantum computer routine. The user interfaces are built according to the standard web architecture: MariaDB + PHP + HTML + JS.
English
Nederlands
Deutsch
Magyar