Predicting Post-Concussion Syndrome Outcomes with Machine Learning
This work addresses a healthcare problem for patients with persistent PCS, but it is incremental as it applies existing methods to a specific medical dataset.
The paper tackled predicting outcomes for patients with persistent post-concussion syndrome (PCS) after a concussion, using machine learning models, with a random forest classifier achieving 85% accuracy and an AUC of 0.94.
In this paper, machine learning models are used to predict outcomes for patients with persistent post-concussion syndrome (PCS). Patients had sustained a concussion at an average of two to three months before the study. By utilizing assessed data, the machine learning models aimed to predict whether or not a patient would continue to have PCS after four to five months. The random forest classifier achieved the highest performance with an 85% accuracy and an area under the receiver operating characteristic curve (AUC) of 0.94. Factors found to be predictive of PCS outcome were Post-Traumatic Stress Disorder (PTSD), perceived injustice, self-rated prognosis, and symptom severity post-injury. The results of this study demonstrate that machine learning models can predict PCS outcomes with high accuracy. With further research, machine learning models may be implemented in healthcare settings to help patients with persistent PCS.