Information Technology Reference
In-Depth Information
be applied to predict if a patient will acquire or not a nosocomial infection. The
models are satisfactory and can support the decision-making of healthcare
professionals. These models allow the application of the appropriate preventive
measures, necessary to ensure the risk patients' safety and well-being.
4
Conclusion
The prevention of nosocomial infections is crucial because these infections can put at
risk the security and well-being of patients and healthcare professionals. Moreover,
their diagnoses and treatment results in additional costs for healthcare institutions.
DM technology can be applied to clinical data in order to predict the occurrence of
infections. Through the use of real data extracted from nosocomial infection forms
recorded in CHP, the DM module of this work demonstrated that it is possible to
obtain classification DM models capable of predicting if a patient with certain risk
factors will or not acquire a nosocomial infection.
In this DM study CRISP-DM methodology was followed and two classification
algorithms were used: SVM and NB. It is possible to conclude that these techniques
and the use of real clinical data allow to predict nosocomial infection cases. Good
models to predict the non-occurrence of infection were achieved (sensibilities higher
than 91.90%). The use of DM predictive models for nosocomial infection can be seen
as a great contribution to the development of CDSS to this sector.
In the future, the best predictive model can be applied to data to predict cases of
infection in real time. It is important to note that, being nosocomial infections an
extremely important problem in healthcare institutions, the methodology presented in
this paper can be adapted to generate predictive models for these infections in other
institutions. Moreover, it is also important to emphasize that the best model achieved
with this study can be applied to other specialties of CHP.
The work presented in this paper is of great worth for society because it allows the
prevention and reduction of nosocomial infections in healthcare institutions through the
promotion of an evidence-based clinical practice, decreasing the risk of complications to
patients and improving their safety and well-being.
5
Future Research Directions
The replication of the DM study with other data and DM techniques as well as the
incorporation of other variables in the predictive model would be also interesting
tasks.
The best model achieved must also be applied to present data in order to make real
time nosocomial infection predictions in new hospitalized patients. The results of
these predictions can be integrated in a Business Intelligence platform. This platform
can perform as a CDSS because the information presented can be used to support
decision-making process in the healthcare organization through the nosocomial
infection occurrence prediction.
It is also important to evaluate the usability and functionality of the DM predictive
models to find improvement opportunities related with their performance or with
important features to their users.
Search WWH ::
Custom Search