Using predictive process monitoring to assist thrombolytic therapy decision-making for ischemic stroke patients

Abstract

Background: Although clinical guidelines provide the best practice for medical activities, there are some limitations in using clinical guidelines to assistant decision-making in practical application, such as long update cycle and low compliance of doctors with the guidelines. Driven by data of actual cases, process mining technology provides the possibility to remedy these shortcomings of clinical guidelines. Methods: We propose a clinical decision support method using predictive process monitoring, which could be complementary with clinical guidelines, to assist medical staff with thrombolytic therapy decision-making for stroke patients. Firstly, we construct a labeled data set of 1191 cases to show whether each case actually need thrombolytic therapy, and whether it conform to the clinical guidelines. After prefix extraction and filtering the control flow of completed cases, the sequences with data flow are encoded, and corresponding prediction models are trained. Results: Compared with the labeled results, the average accuracy of our prediction models for intravenous thrombolysis and arterial thrombolysis on the test set are 0.96 and 0.91, and AUC are 0.93 and 0.85 respectively. Compared with the recommendation of clinical guidelines, the accuracy, recall and AUC of our predictive models are higher. Conclusions: The performance and feasibility of this method are verified by taking thrombolytic decision-making of patients with ischemic stroke as an example. When the clinical guidelines are not applicable, doctors could be provided with assistant decision-making by referring to similar historical cases using predictive process monitoring.

Process mining applications in the healthcare domain: A comprehensive review

“…In the context of healthcare, forecasting patient flows can help managers allocate money and human resources to the many health services provided by the hospital (Benevento, Aloini, et al, 2019; Duma & Aringhieri, 2017, 2020; Kempa‐Liehr et al, 2020; Van Der Spoel et al, 2012). Also, the prediction of complications, comorbidities, severity of a disease on the basis of the patients' characteristics can help physicians to choose the best treatments and drugs (Back et al, 2020; H. Xu, Pang, Yang, Li, & Zhao, 2020; H. Xu et al, 2021). In some cases, the problem of predicting the process behavior reduces to a classification problem (Benevento, Aloini, et al, 2019; Duma & Aringhieri, 2017; H. Xu, Pang, Yang, Li, & Zhao, 2020; Van Der Spoel et al, 2012; H. Xu et al, 2021), that is, finding the most probable label for a given sequence of events, that can represent a process activity, a quantity, a time instant, a resource, or any other key aspect of the process.…”

“…Also, the prediction of complications, comorbidities, severity of a disease on the basis of the patients' characteristics can help physicians to choose the best treatments and drugs (Back et al, 2020; H. Xu, Pang, Yang, Li, & Zhao, 2020; H. Xu et al, 2021). In some cases, the problem of predicting the process behavior reduces to a classification problem (Benevento, Aloini, et al, 2019; Duma & Aringhieri, 2017; H. Xu, Pang, Yang, Li, & Zhao, 2020; Van Der Spoel et al, 2012; H. Xu et al, 2021), that is, finding the most probable label for a given sequence of events, that can represent a process activity, a quantity, a time instant, a resource, or any other key aspect of the process. In other cases, linear regression techniques are preferred to classifiers (Benevento, Aloini, et al, 2019; Kempa‐Liehr et al, 2020) as tools for inferring the trajectory of ongoing processes.…”

“…Kempa‐Liehr et al (2020) proposed to use Geometric regression for predicting patient postoperative length of stay. To assist medical staff with thrombolytic therapy decision‐making for stroke patients, H. Xu, Pang, Yang, Li, and Zhao (2020) proposed a clinical decision support method using decision tree and random forest classifiers for the prediction of the next activity given a sequence of events. H. Xu et al (2021) proposed a transfer learning‐based framework for predicting the long‐term recurrence risk in patients with ICE after discharge from hospitals, in order to point out high‐risk patients for intervention.…”

Acute stroke CDS: automatic retrieval of thrombolysis contraindications from unstructured clinical letters

“…Of the research groups working on CDS in the stroke domain, some identify and display contraindications to thrombolysis at the point of care (5,6), whilst others focus on predicting outcomes in the case that thrombolysis is administered versus not administered (7)(8)(9). A recent review of machine learning methods for selecting patients who might benefit most from thrombolysis treatment is provided in (10).…”