From care for these patients does not end the

From a clinical point of view, patients with certain diseases or syndromes are at high risk for re-admittance to the hospital. It is therefore of crucial importance that the structured and well organized care for these patients does not end the moment of discharge. PCT models developed in this research can serve as a decision support before initial discharge of the patient. They can provide indication of readmission risk as well as potential diagnoses that could appear in future admission.

So, PCT models with data- or expert-driven structure induce clusters of diseases that are frequently diagnosed together. In other words, predicting risk of diagnoses on higher level of hierarchy means that there is a risk of all diseases on lower levels. Doctors can interpret such clusters and exploit this information in order to conduct additional examinations before initial discharge.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

In further text, we will interpret one of the PCT models in medical context (Figure 2). In general, readmission as indicated by the PCT clusters (depicted in Figure 2) generally covers three patient populations (clusters) that comply with common medical knowledge suffering from infection (or sepsis), premature development state, and or immunosuppression. Additionally, PCT model reveals two more clusters represent mixtures of previously defined groups. They cannot be interpreted by standard medical concepts, but they show frequent comorbidities (diseases that occur together) and this can be motivation for future research.

caption{K-means (k=5) hierarchy}

extbf{MetaLabel 1} covers a patient population characterized by sepsis and or infection (MetaLabel 1.3; 1.1; 1.2) related (ICD V4589; ICD V427) or non-related to a surgical intervention. MetaLabel 1.4 could suggest premature development related to orofacial hypotonia (ICD 7833; ICD 32723). MetaLabel 1 1-5 could indicate a patient population post liver transplantation, complicated by sepsis and immunosuppression. All ICD codes as part of MetaLabel 1 can coincide in this patient population.

extbf{MetaLabel 5} suggests a patient population with premature development. All meta-labels can be related with a premature development status. This status is characterized by immature immunity, immature pulmonary functioning and cerebral palsy which result in a failure to thrive (ICD 78341)

extbf{MetaLabel 4} suggests a patient population characterized by an immature or suppressed immunity which is related to leukemia (MetaLabel 4.1; 4.3) or premature development (MetaLabel 4.2). This situation results in infection (MetaLabel 4.4 and 4.5). Furthermore, these patients require enteral feeding by gastrostomy catheter making them at risk for diarrhea.

extbf{MetaLabel 3, 2} consists of a mixture of patients described above. No unique patient population can be defined for these MetaLabels but partly covers infection and premature development. Some ICD combinations seem odd at first notice (acidosis/hypopotassemia and pancytopenia and unspecified essential hypertension).

Overall, these results indicate that patients hospitalized because of premature status, a compromised immune system, in status post transplantation, severe infection (sepsis) or immature cerebral functioning are at high risk for new episodes of hospital admittance.

Hierarchies obtained from other cluster models used in this research are presented in Appendix.