Wearable Device-Based Early Warning of Postoperative Complications in Thoracic Surgery

Postoperative complications are an important concern after thoracic surgery. Conventional perioperative monitoring mainly relies on intermittent vital sign measurements, nursing assessments, physician rounds, and routine post-discharge follow-up. This approach may miss short-lasting, nighttime, or activity-related physiological changes that occur before a complication becomes clinically apparent.

Wearable devices can continuously collect physiological and behavioral information, including heart rate, oxygen saturation, skin temperature, activity, sleep, and wearing status. Combining these data with electronic medical record information may help identify early changes associated with postoperative complications.

This study uses a retrospective and prospective cohort design. In the retrospective stage, previously collected wearable device data and electronic medical record data from thoracic surgery patients will be analyzed. The researchers will evaluate the agreement between wearable device measurements and routine clinical vital signs, describe physiological signal patterns before postoperative complications, and develop a preliminary early warning model.

In the prospective validation stage, newly enrolled patients scheduled for thoracic surgery will wear the study wearable device from hospital admission to approximately 30 days after discharge. The early warning model developed from the retrospective data will be applied prospectively using predefined parameters. Model-generated warnings will be recorded. The occurrence of postoperative complications will be verified through inpatient medical records and follow-up after discharge.

The primary reference outcome is postoperative complications requiring medical intervention within 30 days after surgery. These complications may include pulmonary infection, pleural effusion requiring treatment, prolonged air leak, wound infection, postoperative arrhythmia, emergency department visit, or readmission. Model performance will be evaluated using sensitivity, specificity, positive predictive value, negative predictive value, accuracy, and warning lead time. Wearable device adherence, data completeness, and the added predictive value of wearable device-derived features beyond routine clinical variables will also be assessed.

Inclusion criteria

• Age 18 years or older.
• Hospitalized in the Department of Thoracic Surgery of Tongji Hospital and scheduled to undergo thoracic surgery.
• Able to wear the study-designated wearable device after hospital admission and expected to continue wearing the device and/or uploading data within 30 days after discharge.

Exclusion criteria

• Patients or family members are unwilling to wear the wearable device or unable to meet the required wearing time.
• Severe or unstable psychiatric disease, such as severe depression or schizophrenia.
• Pregnancy or lactation.
• Allergy to the watch strap material or local skin conditions that prevent wearing the device.
• Unable to complete follow-up within 30 days after discharge.