a-callus button
Ρομποτική Τεχνολογία Αποκατάστασης-1

Prediction of Rehabilitation functional gain with the creation of a new tool – Prediction of Functional Independence Measure (FIM©) at discharge

Παρακαλώ συμπληρώστε τα στοιχεία σας και θα επικοινωνήσουμε μαζί σας το συντομότερο.

33rd Annual Congress of the Hellenic Neurosurgical Society -
4th Congress of the SΕENS - 28-29 June 2019
Hotel Makedonia Palace, Thessaloniki

Objectives:  An attempt to identify the determining factors of discharge FIM and to use them in order to predict that size. The aim is the rational management of the rehabilitation program and the early placement of a therapeutic goal, especially in patients with brain damage.

Material & Method: The data used as a sample for the estimation process and modeling are derived from 200 patients in our Center in 2018. For the estimation, the variables used are determinants for measuring FIM at discharge. The survey follows a double approach to comparing and safeguarding the results. A Multivariate Crossectional Regression Estimation and an Artificial Neural Network (ANN) was developed to adapt to a non-linear environment.

Results: The first indications of this regression give us a number of factors that are statistically significant (FIM at admission, comorbidity, infections, bladder catheter, pressure ulcers, and length of stay) helping the final estimation. Correspondingly, the results are obtained through the assessment of ANN and MSE.

Conclusion: Both methods offer two good prediction tools. In particular, the Neural Network architecture, without rejecting any parameter, learns the system to think and produce / predicts a score of FIM at discharge, using the parameters it judges right through its training. In our case the algorithm seems to work well by throwing the estimation error constantly.