Predicting Success on ASCP-BOC by Using Existing Data

Abstract

Abstract Introduction/Objective The MLS Program at tUAMS is housed within the College of Health Professions. Our graduates take the ASCP BOC certification exam as a graduation requirement and the vast majority of students will take the exam. Methods/Case Report For this study, we examined online MLT to MLS students who took their ASCP BOC certification exam in late 2019 thru April of 2020 and used this data to build a multiple regression model to predict ASCP scores . This model was used to predict ASCP MLS scores of students graduating in the next cohort in December of 2021. Results (if a Case Study enter NA) 103 students took the ASCP MLS certification exam in early 2022. In early 2018, the MLS program created a database to store pre-admission information that was routinely collected. In addition, we also gathered academic information as the student progressed thru the program and post-graduation. There were a total of over 50 variables included in the database. The goal of this database was to provide insight to help both faculty and students succeed. We used a simple multiple regression model to determine which variables were correlated with the ASCP BOC score. We also used this data to develop a simple predictive model to prospectively estimate a student’s ASCP BOC score. 6 variables were identified as potentially useful in a simple model. For the initial student ASCP score prediction, a blended model using the average predicted score of the six models in Table 2 was used. The performance of the model is shown in Table 3. At the individual level, the model was accurate 82% of the time. Conclusion Based on our initial findings, we believe that this model may be useful to help gauge student readiness for passing the ASCP BOC exam. Several students expressed their gratitude, including one who stated that this information spurred them to study harder. There was one interesting finding. 8 of the 12 students (67%) who were predicted to fail passed the exam on the first attempt. However, only 64% of students (11/17) predicted to pass, but with a predicted score of <430, passed. Subsequently, 2 of these 6 failures passed on their second attempt. We may need to investigate a more powerful way to communicate the results of the model for all of those in the “danger area”. 98% of students with a predicted score of 470 or above passed on their first attempt. The model is currently being refined to include 2021-2 certification scores which should improve the academic utility of the model.