A retrospective cohort study of 590 patients aged 75 or younger after cardiovascular surgery found that discharge-time clinical factors could predict reduced exercise capacity at 3 months; 139 patients (23.6%) had reduced exercise tolerance on follow-up cardiopulmonary exercise testing. The single-center study suggests routine clinical data can support smarter postoperative risk stratification and earlier targeting of rehabilitation in patients with heart disease.

Why It Matters To Your Practice

• Functional recovery is a patient-centered outcome, but clinicians often lack a practical way at discharge to identify who will struggle with exercise tolerance after cardiovascular surgery.

• A model based on readily available clinical parameters could help flag higher-risk patients before symptoms or poor recovery become fully apparent.

Clinical Implications

• Consider discharge data as more than administrative information: it may support prediction of which postoperative patients will have limited peak oxygen uptake at 3 months.

• Patients identified as higher risk may benefit from earlier referral to cardiac rehabilitation, closer follow-up, and more individualized recovery planning.

• Because the study was retrospective and single-center, any AI-enabled tool should be validated locally before it influences care pathways.

Insights

• The study included patients undergoing cardiovascular surgery between April 2015 and January 2025 at a single cardiac center.

• The outcome was reduced exercise tolerance at 3 months, defined by peak oxygen uptake on cardiopulmonary exercise testing.

• The focus on discharge-available variables makes the approach more realistic for workflow integration than models requiring specialized testing after discharge.

The Bottom Line

• AI built from routine clinical factors may help clinicians predict 3-month exercise limitations after cardiovascular surgery and target rehabilitation sooner.

• For practice, the near-term value is risk stratification, not replacement of clinical judgment.