Objective To establish an artificial intelligence voice follow-up system for differentiated thyroid cancer and evaluate the completeness and accuracy of information collection by this system.

Methods Based on the postoperative medication and follow-up status of patients with differentiated thyroid cancer, an artificial intelligence voice follow-up system for differentiated thyroid cancer was established. From February 25 to 29, 2024, a total of 2, 471 differentiated thyroid cancer patients without follow-up in the thyroid cancer database management system of the Department of Thyroid and Breast Vascular Surgery of the Hospital were followed up by telephone, and 220 patients (approximately 10% of the total) were randomly selected for completeness and accuracy analyses of their call records.

Results This study successfully established an artificial intelligence voice follow-up system for differentiated thyroid cancer, and the overall call connection rate was 88.9%. For patients with successful calls, the average follow-up time of the artificial intelligence voice follow-up system was (1.7±0.5) minutes, the overall information collection completeness rate was 83.8%, with a completeness rate of 92.9% for follow-up at 3 months after surgery and 82.7% for long-term follow-up, and the differences were statistically significant (χ²=15.200, P < 0.001). The accuracy rate of information collection by this follow-up system was 94.1%. The accuracy rates of information collection for patients who spoke dialect and Mandarin during follow-up were 92.2% and 95.4% respectively, and the difference was not statistically significant (χ²=0.957, P=0.389).

Conclusion The artificial intelligence voice follow-up system for differentiated thyroid cancer has high efficiency, completeness, and accuracy in information collection, which can reduce human costs for follow-up and has good application prospects in large-scale follow-up of populations with differentiated thyroid cancer.