University of Twente Student Theses
Utilizing the qSOFA-score & process mining techniques to predict ICU admissions
Gerritsen, J.W.S. (2024) Utilizing the qSOFA-score & process mining techniques to predict ICU admissions.
|
PDF
3MB |
| Abstract: | - Objective: The graduation project aimed to investigate patient event logs of patients going into the ICU and generate an ICU admission predictor using the qSOFA score and process mining techniques. - Methods: Data was simulated with qSOFA scores that were linked to patient outcomes and the data was attached to a patient journey event log dataset of general practitioner patients. A lab test in the patient journey dataset was used as a substitute for the ICU. After combining the datasets, they were analysed using the “Perform Predictions of Business Process Features” plugin from Prom Lite 1.4 to analyse the results. - Results: 10000 patients were in the dataset of which 2106 went to the ICU. With the qSOFA-scores attached, it was concluded that qSOFA ≥ 2 was the most accurate predictor with a 73.8% accuracy which is not in line with background research. Looking at two control metrics the kappa statistic and the RMSE which were 0.2379 and 90.8% accordingly, the metrics both suggest that the predictor cannot be taken as scientifically valid as the kappa statistic should be in the range of 0.81-1.00 and the RSSE which should be between 0%-10%. - Conclusion: Even though the resulting ICU admission predictor, qSOFA ≥ 2, had a strong accuracy, it was not in line with background research and cannot be taken as scientifically valid. The simulated data is highly likely to not be an accurate representation of real qSOFA-to-ICU outcome data. However, utilizing process mining techniques on real data could still be beneficial to the ICU management field. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Creative Technology BSc (50447) |
| Link to this item: | https://purl.utwente.nl/essays/102755 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page