Patient routing algorithm in differential diagnosis of cutaneous neoplasms with the combined use of Derma Onko Check and Melanoma Check artificial intelligence software tools

Abstract

Objective. To develop and validate a patient routing algorithm for differential diagnosis of skin neoplasms using a combination of the Derma Onko Check and Melanoma Check AI programs to optimize diagnostic time and reduce the burden on the healthcare system. Results and Discussion. Data analysis using a Python program and the pandas, numpy, scikit-learn, and matplotlib libraries resulted in an optimal routing algorithm threshold of 62%, achieving 100% sensitivity. The combination of the Derma Onko Check and Melanoma Check AI programs increases the detection rate of malignant neoplasms through complementary analysis: Derma Onko Check evaluates overall malignancy, while Melanoma Check evaluates specific features of melanoma. The algorithm provides eight routing options, in which patients with a high probability of malignant neoplasms are referred to an oncologist (diagnosis time: ~23 days), while those with a low probability of malignant neoplasms are referred to a dermatologist or immediately placed under dynamic observation (clinical decision-making time: ~1–15 days). This minimizes unnecessary referrals, reducing the workload of specialists and optimizing healthcare resources. Conclusion. The proposed algorithm improves the efficiency of skin neoplasm diagnosis, striking a balance between high sensitivity and resource conservation, which may serve as the basis for integrating AI into clinical practice.

Ähnliche Arbeiten