Smartphone-derived mobility metrics identify distinct recovery phenotypes following ACL reconstruction: associations with function, pain, and psychological health

BACKGROUND: Recovery after anterior cruciate ligament (ACL) reconstruction is typically monitored through patient-reported outcomes and periodic clinical assessments, which may not fully capture real-world functional performance. Smartphone-derived mobility data provide a passive and objective approach for evaluating daily activity and recovery. PURPOSE: To examine associations between smartphone-derived mobility metrics and recovery outcomes in adults 3-12 months after primary ACL reconstruction and to identify distinct recovery phenotypes using mobility-based clustering. METHODS: This cross-sectional study included 120 adults who were 3-12 months post-primary ACL reconstruction and had at least 14 days of retrospective smartphone mobility data. Mobility variables included daily step count, activity volume and intensity (e.g., active minutes and walking bout duration), sedentary behavior, and walking-related metrics. Outcomes included pain (Numeric Rating Scale [NRS]), functional status (Knee Injury and Osteoarthritis Outcome Score [KOOS] Pain, Activities of Daily Living, and Sport/Recreation subscales), knee-related quality of life (ACL Quality of Life questionnaire [ACL-QoL]), and psychological measures including the Pain Catastrophizing Scale (PCS), Patient Health Questionnaire-9 (PHQ-9), and Generalized Anxiety Disorder-7 (GAD-7). Associations were evaluated using Pearson correlation and multivariable linear regression analyses. Recovery phenotypes were identified using unsupervised k-means clustering of mobility metrics. RESULTS: Higher daily step counts were independently associated with better KOOS-ADL scores (β = 0.48, p < 0.001), higher ACL-QoL scores (β = 0.45, p < 0.001), and lower pain intensity (β = - 0.39, p = 0.001). Three recovery phenotypes emerged: a high-activity/low-pain phenotype with superior function and minimal psychological distress, a moderate-activity/moderate-pain phenotype with intermediate outcomes, and a low-activity/high-pain phenotype characterized by reduced mobility, poorer function, greater pain, and higher pain catastrophizing. Significant gradients were observed across phenotypes for pain (NRS: 1.72-5.01), function (KOOS-Pain: 88.94-62.37), and pain catastrophizing (PCS: 8.52-24.91; all p < 0.001). CONCLUSIONS: Smartphone-derived mobility metrics were significantly associated with recovery outcomes following ACL reconstruction. Mobility-based clustering identified distinct recovery phenotypes associated with meaningful differences in pain, function, and pain catastrophizing, highlighting the potential of passive mobility monitoring to support personalized rehabilitation strategies. However, given the variability in evaluation periods and uncontrolled surgical factors, future research employing standardized, long-term follow-ups is warranted.