Clinical Care And Artificial-intelligence-guided Resistance Training Improve Functionality In Lower Back Pain

Abstract

PURPOSE: To examine the role of artificial intelligence (AI) guided resistance training relative to clinical variables in subjects who suffer lower back pain (LBP). METHODS: 69 subjects (M24:F45) with LBP completed an 8-week intervention. Subjects self-selected clinical care enrollment (physical therapy, chiropractic, or no care) for their LBP prior to enrollment. Enrolled subjects were randomly assigned to either an 8-week AI-guided moderate-intensity core resistance training protocol or to no interventional training. This created four study groups (CONTROL (n = 9), TRAIN (n = 13), CLIN (n = 26), or COMBO (n = 21)). The CONTROL group received no treatment or training, the TRAIN group received training but no treatment, the CLIN group received treatment but no training, and the COMBO group received both treatment and training. AI-guided training was performed on a Tonal exercise trainer with supervision. Participants were evaluated using functional tests (Biering-Sorenson’s exam, 6-minute walk test) and patient-reported outcomes (PROMIS Physical Function, PROMIS Pain Interference, Tampa Scale of Kinesiophobia, Visual Analog Pain Scale) at baseline, 4-weeks, and 8-weeks. Wilcoxson matched pairs analysis was used for comparison of metrics within each group at baseline and 8 weeks. Significance level was set at α ≤ 0.05. RESULTS: TRAIN, CLIN, and COMBO groups increased Biering-Sorenson endurance from baseline to 8 weeks (23 + 8 sec, P < 0.02) and increased distance traveled in the 6-minute walk test (0.02 + 0.06 miles, P ≤ 0.03). The TRAIN, CLIN, and COMBO groups improved scores in both PROMIS Physical Function (1.8 + 0.8, P < 0.05) and Pain Interference (-3.3 + 1.4, P ≤ 0.02) after the 8-week intervention. For the TRAIN group, PROMIS scores also improved at the 4-week midterm evaluation (P ≤ 0.04). Tampa Scale of Kinesiophobia and Visual Analog Pain Scale scores did not change for any group (P > 0.05). No significant changes were observed in the CONTROL group for any outcome measure (P > 0.05). CONCLUSIONS: Core-targeted, AI-guided resistance training and physical clinical care increased functional outcomes and patient satisfaction in individuals with LBP. Combining physical clinical care with AI-guided resistance training did not have a complimentary effect on improving functional or patient satisfaction outcomes.