Jacob Powell, PhD

Predicting which traumatic brain injury patients will meaningfully improve during treatment remains a major challenge in military medicine. In a new study, researchers evaluated whether machine learning models could forecast clinically significant improvement among 790 active-duty service members completing a four-week interdisciplinary Intensive Outpatient Program at the National Intrepid Center of Excellence. They compared XGBoost models built from demographic data, total symptom scores, and symptom-cluster scores to identify which factors best predicted improvement in posttraumatic stress and post-concussion symptoms. Notable results include: 🔹 Symptom-cluster models outperformed total-score models (area under the curve 79% vs. 75%; accuracy 72% vs. 68%). 🔹 The strongest predictors were PTSD arousal, avoidance, and reexperiencing sub-scores, followed by education level and cognitive post-concussion symptoms. This work highlights how targeted machine learning tools could help clinicians anticipate outcomes and allocate treatment resources more effectively. 𝗥𝗲𝗮𝗱 𝗺𝗼𝗿𝗲: link.springer.com/article/10… 𝗦𝘁𝘂𝗱𝘆: Prediction of Clinically Significant Improvements During the Interdisciplinary Intensive Outpatient Program for Traumatic Brain Injury Using Machine Learning 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Rujirutana Srikanchana, @JakeRPowell, Jacob Powell, Treven Pickett, Thomas DeGraba, and Chandler Sours Rhodes @WRNMMC_DHA @USUhealthsci @UofMaryland @UMmedschool 𝗦𝘂𝗯𝗺𝗶𝘁 𝘆𝗼𝘂𝗿 𝗺𝗮𝗻𝘂𝘀𝗰𝗿𝗶𝗽𝘁: link.springer.com/journal/10…

Mild Traumatic Brain Injury and Career Stage Associate with Visible Perivascular Spaces in Special Operations Forces Soldiers @jmihalik79 @GfellerCenter @UNCEXSS @UNC_HMSC @JakeRPowell @UNCpublichealth @genevaUSA @uncfammed @USASOCNews @BMESociety Link: rdcu.be/dXhGG